2/2 Likol 2016/11/15(Tue)01:16 No. 758669 ID: bfb318
File 147920137651.png - (47.57KB , 800x800 , 2.png )

>"... have you not slept?"
"We have to look good."
>"And you look like shit." Not even in uniform. "Come chill with some of us. But not before you shower."
"I'd rather not disturb their work while dodging mine."
>"You think out of a couple hundred hivemates, there's not a single one on break?"
"There shouldn't be."
>"... how far have you gotten untangling the raw ring shell code?"

Low blow, Rihhin.

>>758669
Have you checked the logs for any morse code recently?

Oh hey Likol, what's that you're looking at there? Also, out of curiosity, what's the date today?

Likol 2016/11/15(Tue)02:51 No. 758675 ID: bfb318
File 147920709892.png - (107.82KB , 800x800 , 3.png )

>What's that you're looking at there?
The ring shell data, which is just about the only data that we can see as output. Once every three days I get a couple of error messages to help unravel the whole thing.

It feels like grunt work. We're confident that my mentor, Dr. Arza Fletch, would pay his life savings away to get the information I've gained, because he could make breakthroughs with it and see connections between objects. Objects that I'm not even seeing in the first place, let alone the connections. Compared to him, I'm a code monkey. The unprocessed data is wasted on me.

>Have you checked the logs for any morse code recently?
Morse code... has been taught to the Ring Shell, but trying to get the RS to use the tool is a whole other can of worms. It has never done this before despite decades of effort, despite that some of those AI's that make it to the RS should know morse code and how to use it.

Dr. Fletch has gotten cagey with his CAI tools after giving us the ability to make CAIs on our own. His program is able to take the ring shells native data, which is an impervious mess of tangles, and say how long the cycle took both in virtual time and real time, and how many AIs made it to the end. I wish he would have taught me how to make my own programs, but I wonder if I would have understood it even if he taught me. The man is... eccentric and not socially cunning, but in this field, I feel like I've been taught by a once in a millenium genius.

On the other topic, things added to the RS, like morse code, will stand out easily even while mountains of data are generated around it per second.

It's been a while, but a new cycle just ended on this CAI project. I don't want to start a new one at an odd time, so I kill a few minutes to do a search for morse code. As always and as expected, there is no data that fits a morse code message.

Rihhin broadcasts I should take a break after starting the next one, then leaves.

I go and wait by the supercomputer's terminal.

1:50.

1:55.

1:59... wait for it....

2:00. Nice and even timestamp.

>What's the date today?
On hitting the button, I receive an answer as it logs the start time.

April 16th, AW 131, 2:00:00 PM
CAI generation attempt No. 3,119 has begun.

I've nearly been at this for 30 years.

The date stamp is fed to my desktop, before the individual AI samples begin generating. In an attempt to tell me all the IDs of one trillion entities, the monitor turns into a spastic green glow that does a better job of illuminating the room than telling me anything. It outputs 50 lines 240 times per second as per the monitor refresh rate. And even at that rate, the monitor skips nearly 14 million others that are generated before it can wait for its next frame to display a new 50.

There's no point to watching it, anyway. All the work is done on the side laptop that reads off the supercomputer's data.

Which either amounts to dissecting the ring shell, which I feel like I would have more progress finding out how a neumono brain works by dissecting myself. Or, perusing the extracted data from the previous 4 AI's that got through the stages, and see if any of their data can be linked with the error messages.

Or, I reluctantly admit, a break is an option.

>>758675
So do you know about the glitchy emergent AI in the system brought about by your stabilizing AI interacting with Rihhin's corrupting AI? How many times has that emergent AI reached the ring shell? How many times has your stabilizing AI's failsafe been triggered?

Hey, what sort of backups are there for the AIs? Are there backups for the two custom AIs? Even for the emergent AI?

Take a look at cycle 3118's final 4 AIs. Have you noticed any patterns as to which AIs reach the end most often? At the very least you can judge which AIs seem to have the biggest advantage overall.

I'd say have a quick look at the AIs then have a break, ideas might strike when you're not actively thinking about it.

Aside from research is there a purpose for this system, I thought there already was a CAI made? Also, what are archcycles? What modifications / injected programs have been made to the system compared with how it was originally?

Of course, the test offers survival to the AIs that are skilled enough to jump through the hoops. It doesn't pick up on AIs that are smart enough to hide themselves and troll the other idiots scramble to the finish line.

Connect the processor to a small, well-contained bot. And by well-contained, I mean you literally lock it inside a triple-chained iron box with one tiny hole for the USB port. And maybe some pretty pictures of the surface as a lure. See if you can trap the ghost in the machine. Even if you don't expect anything, what else were you planning to do today?

When was the last time you saw Azra? You're way below his level, but his students aren't. Maybe you could network with the other nutjobs?

Assuming the attempt numbers line up accurately with the cycle numbers, which they might not, considering Likol just said 4 AI have already been generated, but we were only ever able to find evidence of one, this is really bad news. This is all happening before Polo has arrived, which means one of three things; Polo shows up within the next 3 days, Alison and friends never make it out, or they do make it out, but the Salikai manages to subjugate them. Even if Polo does show up within the next 3 days, the Salikai still has control of the CAI after the timeskip, which does not bode well for the contestants.

So what are you actually expected to be doing down here, if the output you're getting is so useless? Do you mean 4 CAI, or 4 AI in one CAI?

>>758675
Of the two, I'd say looking at the four AIs might be more fruitful than looking at the ring shell, again.

Your hivemate is right, tho. You should take a break. As much as it galls you, you work better after one.

>April 16th, AW 131, 2:00:00 PM
Four years before The Intermission and/or Polo Quest.

Maybe you need some cuddlehive style R&R?

What's the point of this research project, besides trying to understand how CAI's function? Is there something you're trying to accomplish?

>"You think out of a couple hundred hivemates, there's not a single one on break?"
>couple hundred
...Polo and Korli's best estimate of the Science Hive's population in the last Polo Quest thread was somewhere the thirties. They're due to lose a lot of people in a few short years.

Things with Vanski must get a lot worse, and in a hurry, not gradually over decades like I was expecting. They might not even be enslaved yet.

1/2 Likol 2016/11/15(Tue)05:51 No. 758698 ID: bfb318
File 147921787062.png - (21.35KB , 800x800 , 4.png )

>Do you mean 4 CAI, or 4 AI in one CAI?
4 individual AIs make it out of a single CAI generation process. Three from the standard stage area, and a fourth one from a custom procuder of administration. They don't form a CAI. They just get sent through the ring shell, then rejected and sent back to the beginning with the rest in a new cycle.

The active CAI we have has countless AI, and although it runs on the same platform as what's generating another CAI, they are separated.

>So do you know about the glitchy emergent AI
I know that the AI count has become higher than one trillion. I believe that is from the simulated CAI battles within drawing contestants out. It seems like they're stable, and no different than other contestants once they're integrated in with the others.

>Hey, what sort of backups are there for the AIs?
There's really not any. We have backups of what is used to generate the AIs, including the custom AIs by Rihhin and myself. As for the AIs that have their unique experiences and exist currently, that's stored within the blocks and we have no known way to get them out.

The closest we have is my stabilizer, but that's not a backup, that's just a backup home for the AI's themselves in case the first gets destroyed.

>How many times has your stabilizing AI's failsafe been triggered?
The stabilizer has definitely taken hits before. There's been a few times where its presence with the other contestants is completely eliminated, or contaminated by Rihhin's corruptor.

The native stage area for the AIs has never been contaminated enough for my stabilizer to need to serve as a new home for them. Which is good, I suppose, since I'm not even 100% sure it will work.

>Take a look at cycle 3118's final 4 AIs. Have you noticed any patterns as to which AIs reach the end most often?
Every AI has a persistent ID that perpetuates through each attempt, so yes, we have compiled data on the AIs that tend to repeatedly make it through, and what their values were on their generation for the cycle, along with all of their experiences.

And it's a disaster making sense of it. They all just have one big long string that not just has to be read in a linear form like DNA, but up and down by lining it on top of itself in some kind of loop.

All of the AI sent in has been made by post-empirical algorithms, but it's all turned into unreadable garbage by the RS, as with everything else. I'd call it encryption, but it's the real data itself presented in an obtuse way, not some code that scrambles the data.

Despite that, we do have other programs that can reliably extract parts of those strings, namely the ID, and many statistics about what kind of AI they are, personality wise.

I look at the last 4 who made it through.

ID Numbers...
197484325143
396265711152
580856634466
656114696347

I recognize all of them right off the bat. I know very little about them, otherwise. Although I have variables on what make up their personality, it's very difficult to imagine what kind of people they would act like if they became a CAI. I've tried to map it out using an emulator before, but it's almost certainly an inaccurate representation of them. There's still missing data on them, and randomating those unknown variables changes them drastically.

Plus, I can't become attached.

>Have you noticed any patterns as to which AIs reach the end most often?
Most of them have a tendency to look out for themselves, but being a loner severely diminishes their chances. They look out for others to a degree, and there's a little golden zone of an ideal amount, I've seen. Look out less, and they probably get killed. Look out more and their chances of making it to the end slowly diminish. Or I'd like to say, except that one contestant that's made it near 50 times with such an extreme level of external care. Either get to a point and suddenly the survival rate skyrockets, or it's an anamoly.

All these variables interact in a complex web, so it's likely other factors that play in. Trying to simplify single aspects that lead to survival is a good way to miss the real picture. But that's one I think I've pinned down.

>What are archcycles?
A hard reboot to completely clean out the system and revert everything. Some things, like contestant IDs, perpetuate between cycles. Virtually nothing perpetuates between archcycles.

>Connect the processor to a small, well-contained bot.
It would be interesting to see if having a side box would interact with the CAI generation process, but aside from messing with the experiment, I don't have means to procure such an advanced object that would be able to interact with the CAI like that.

I start a few procedures to at least simplify cycle 3118's final ring shell output so I have a chance of reading it, and decide I'll take a break.

>What's the point of this research project, besides trying to understand how CAI's function? Is there something you're trying to accomplish?
By understanding how a CAI is made, and being able to adjust it, we'd be able to create CAIs that are better suited to certain tasks, and even modify the AI itself.

In Vanski's case, I suppose he would want a CAI that is obedient and loyal to himself.

Aside from that, if we were able to translate how the Ring Shell works to something we understand, we would be able to create new CAI blocks, and perhaps learn about the tech of ancient belenos... Vanski's power and clout would go through the roof and, I would hope, so would our hive's. I miss having a big hive.

2/2 Likol 2016/11/15(Tue)05:51 No. 758700 ID: bfb318
File 147921788806.png - (38.77KB , 800x800 , 5.png )

>When was the last time you saw Arza?
It was... november of 109. That was the last personal meeting. We've spoken since, but it's been through tense, sterilized messages through various channels. Even when he gives us new software, it's because Vanski cut another deal with him. Dr. Fletch wants data I've mined, but if he got it... he would realize what we've done to get it, and denounce any business with us. We can't do that forever, but the years are just going by faster and faster. One of us is going to die first, possibly before the secret is out.

>I thought there already was a CAI made?
There is, and it's active at all times. Which is why our quantum supercomputer, and my workplace, is an extremely secured compartment in the complex. When I enter, I must shower, out of fear that the CAI could brush me with nanomachines and get into my desktop or some other insane plan. They cannot be allowed to mine their own data.

>What else were you planning to do today?
Anything but take a break.

But that's what I'm doing.

I guess I needed to pick up supplies anyway for my work compartment's kitchen. It's been a week since I've left it, and it's the fact that my hive is in empathic range that I don't go crazy staying that long in there.

Thankfully, leaving is a much less arduous procedure than entering.

>Maybe you could network with his students?
I'd love to try, but he doesn't take many on as their students. At most he'll give people the tools to learn on themselves, but private tutilage is rare. Plus, it might be seen as undermining Dr. Fletch himself, and the ramifications would be bad.

There is one student I heard about. A child, really, that he took an interest in. I know nothing about her, but she's not in the crime ring, and is essentially off limits. Perhaps someday I'll meet her and see if perhaps she can teach me a thing or two, since she'll probably learn a lot more than I am just... grunt working.

I go into the main kitchen to get some food that isn't instant. Moving around I get a better idea of my hive's mood. Things are content, or at least what we've come to consider acceptable, but there is still stress. There are looming deadlines for progress in various fields, and a bad few decades have hindered the overall happiness we've had in the past. The medical department, especially, is in a bad spot, and lowers the overall mood.

Ohhh myyyy god!
It's Likol!
How many months has it been?!

Markie, an aspiring geologist. One of the few that doesn't wear her goggles all the time.
Korli, a generalist that hasn't finished general education. Possible interest in biology.
Reon is the same as Korli, and has taken an interest in my field. I hope he can pick up the torch before my cynical mindset steers him away.

All bubbly, recent generations.

"Hello."

They all say hello back in unison and start getting me food. There's still a respect for elders in this hive, despite all things. It's completely unnecessary. Surrounding hivemates don't come to join, but are pleased I've taken a break. As soon as I get this far, I have to abandon any wishes to cut the break short and go back, unless I mar the present mood.

The three hivemates sit across from me, expecting small talk, but not wanting to initiate it.

Ask them how there projects are going. Unwinding will help you relax your mind.

Plus maybe something one of them is working on will help inspire you.

>ID numbers
Geeze the system person doesn't even have a tag or prefix to distinguish them. That makes sorting the data harder (although for all we know those are all system IDs. The contestant slots could have been taken by system people who deserted).

>>758700
Awww. They look up to you. That's cute.

Indulge your younger hivemates with some attention. They'll enjoy it, and it'll be good for you emotionally.

Ask them about their work, lives, schooling, what they're up to? People like it when people take an interest in them, especially people they look up to.

(How backed up are you after months hiding away in a room? Maybe you should indulge in "getting attention" from one of the bubbly energetic younger generation, too. At least one of them would probably jump at it).

Spend some time with your family. Focusing to much on work will make it harder for you to come up with any great ideas.

So go talk about there work and sleep around a bit and take a few hours or a day off.

How much programming knowledge do you have? You said Arza never taught you how to make programs but you obviously have made your own stuff over the years to peer at this data.

Your next project should be a program to lock on to anything using morse code and identify it so you can transfer that specific Ai out onto a test partion.

With the rate the sim moves if you are not ready for a communication breakthrough the ai doing it could get destroyed as part of the culling process they go through before you could grab it.

Once you had a Ai that had figured out how to talk to the Ring shell you could communicate to it and learn whats going on in there. And maybe have it help you translate the data.

Find out how each of there projects/schooling is going.

Then go relax and have some fun and go get laid.

Will help you focus on work later.

Ask them for the hive gossip. Keep up on what's going on with everyone else, is anyone having any personal problems? Any special relationships formed between anyone lately? And personal triumphs?

While you're out here, maybe you should check in on the CAI, as well. AIs though they are, they have wants and desires and affections. If they have any little harmless requests, you could try help them. Making a little personal connection costs very little and could have good benefits.

Speaking of benefit, I'm sure a cynical survivor such as yourself has always felt secure having little emergency caches, countermeasures and alarms scattered around near where you live and work, as a matter of those habits that led to you being a survivor. While you're out for once, perhaps you should check in on some you've set up. I'm sure you've shared them with other members of your hive, but it doesn't hurt.

And perhaps you should check in with the medical department before you go back to work, if they're unhappy. Perhaps they'll benefit from a little wisdom.

You should ask them how things are going. Keeping yourself holed up isnt good for you, empathy or not. Neumono are highly social creatures, you'll feel better if you relax for a couple minutes.

>>758698
>There's been a few times where its presence with the other contestants is completely eliminated
Have you noticed anything strange happen during those cycles? Or the cycle afterwards?
>or contaminated by Rihhin's corruptor
Have you noticed anything strange happen during that cycle or the cycle afterwards?
>ID numbers
Wait, those don't go past a trillion. Assuming it starts at 0, that number of digits gives exactly one trillion combinations. What ID numbers are used for the extra AIs? Does it just slap another digit on there? If so, which AI is 1000000000000? I suppose that would either be the corruptor or the stabilizer... What about 1000000000002? The first AI to (presumably) escape a simulated CAI battle?
(damn, it occurs to me that we probably can't determine what the first Glitcher's ID is. At best he showed up on cycle 2, but contestants could have escaped a CAI Fight simulation on cycle 1. Did Glitcher and Rulekeep check out cycle 1 and 2 to confirm if there were any escapees or who the first Glitcher was? It's also possible he doesn't have an ID...)

Have there been any cycles where less than 4 AIs were recorded having made it past RS?

>>758764
I'd presume the AIs from CAI battles have different number formats? Or if we assume those are all systems IDs (say, Sevener and two allies deserted and won contestant side, and the last number if for the last admin) like >>758707 suggested, then it works (since there were always less system-side participants than contestant side).

...or maybe there are headers or leading zeros stripped by the interface. (Or maybe it's in hexadecimal and we managed to by pure luck get 4 entries without any letter digits).

We already know the contest supports multiple simultaneous instances of people with the identical IDs, but somehow distinguishes them, too. That means there's cycle and/or instance identifiers stripped out too.

If Reon does enter your field, would you be responsible for teaching him? I imagine with such a low population, something like an apprenticeship would be necessary for more specialized education.

Tell him you hope that seeing you become so absorbed in your work that your hive is shocked just to see you getting lunch doesn't put him off the job.

...oh! Have there been any IDs which have showed up past the RS only once? Or have had seemingly corrupted data? You said you could see the AI data and partially emulate it. Were there any AIs that didn't emulate correctly?

Also, the CAI doesn't have access to the block's data, but do they know of its existence? I mean how is the salikai keeping the CAI under control?

1/3 Likol 2016/11/15(Tue)15:25 No. 758826 ID: bfb318
File 147925230404.png - (11.07KB , 800x800 , 6.png )

>If Reon does enter your field, would you be responsible for teaching him?
Yes, although I hope, then, that Dr. Fletch would start making visits to him.

>How much programming knowledge do you have? You said Arza never taught you how to make programs
I think I'm a decent programmer. I have made software in the past for personal use. What Dr. Fletch didn't show me is how to code for this, which requires a deeper knowledge of the RS than I have. To code for it would still be possible for me, but it would take me far too long to do on my own, nevermind the bug testing to make sure my output isn't misleading me.

>Program to lock on to anything using morse code
We have that. Anything that's been added to the RS post-empire is much easier to isolate.

>So you can transfer that specific Ai out onto a test partion.
That's trickier. AI in the ring shell is spread out, and it's amazing that it works at all.

>If you are not ready for a communication breakthrough the ai doing it could get destroyed as part of the culling process they go through before you could grab it
The RS doesn't destroy any of the AIs. Either the AI remains intact, or it's already part of the ring shell and all over the place. Either way, we keep records of the ring shell output at all times, even if we can't do much with it.

Did I... yes. I habitually put on my watch. This will send me an alert if the RS is detected doing anything out of the norm, like morse code, or some catastrophic error occurs. Of course, with our paranoia of the CAI trying to study itself, it's not a direct wireless connection, it's more like my lab will generate a specific alarm noise, and a node outside of it will activate and then send a ping to my watch.

There's scheduled diagnostics on it to make sure it's working, which must be done, as it has never activated a single time. I still put it on whenever I leave, anyways, as to not do so would be to give up on the idea anything will ever happen.

The CAI working on my data would go so much faster. Knowing that makes me feel I'm wasting my life here. I don't know why this hasn't been done by another faction that owns a CAI, even by accident. Dr. Fletch just told me 'they can't' and didn't elaborate other than a hint that the belenos empire would never allow something like a CAI have the ability to self replicate.

>Talk with the CAI
It's awkward. I'm essentially the caretaker of their usurper, or disease.

>What ID numbers are used for the extra AIs?
Numbers over a trillion.

There's just no leading zeroes in the numbers.

>What was the first AI to (presumably) escape a simulated CAI battle?
I don't even know for sure. It's tough to find such a specific fact like that, and I don't believe that's significant.

>Have there been any cycles where less than 4 AIs were recorded having made it past RS?
No AI has ever made it past the RS itself, but there's always 4 AIs that at least make it to the RS before going back.

Even if there aren't 4 AIs that make it to the end of the stages, then the RS will simply choose the 4 that got closest.

None of them have been corrupted. The RS would bar any corrupted AIs, I believe.

>Were there any AIs that didn't emulate correctly?
Although they are almost certainly inaccurate, they were always at least functional.

>Also, the CAI doesn't have access to the block's data, but do they know of its existence? I mean how is the salikai keeping the CAI under control?
It's unlikely that the CAI doesn't know the basic structure of CAI blocks. They also know that we're running tests on their own blocks, and that if we don't manually fail an attempt cycle, the 4 AIs will make it past the RS and override them. They don't have control over it, so if we all suddenly disappeared, the CAI would not last long.

Aside from that, wires from the quantum computer go down to Vanski's own room where the CAI blocks themselves have been inserted. If he wanted to, he can simply unplug the CAI. I'm sure he has other failsafes, as well.

In other words, if the CAI misbehaves at all, we can simply wipe the blocks clean of them, and regenerate a new CAI on the emptied blocks within a few days. The CAI blocks are not replacable, but the CAI is.

2/3 Likol 2016/11/15(Tue)15:25 No. 758827 ID: bfb318
File 147925231661.png - (16.75KB , 800x800 , 7.png )

>You should check in with the medical department, give some wisdom
I doubt I could give them anything. They've already missed their deadline.

"How are your projects going?" And why are you not working on them?

>"I'm leaving some fluid on rocks to see how well they melt down, and if the fluid seeps through the rock if it's contained to the outside. If things go well, we can reshape the caverns without using pickaxes and other impact-heavy tools." Markie says. "I'm waiting on that, so I'm here to rest my head."
>"I'm waiting for my computer to process some propulsion simulations." says Reon. I could be doing other stuff, but... "I've been told to focus on that project and not anything else! But since I'm waiting on that stuff, I don't have anything to do."

Sometimes I feel that way too, but with autonomy, I can choose to do other things to busy myself. I make sure he understands my sentiments that me rarely leaving my room is my own choice, and I'm not forced to stay in there for days at a time.

Korli dreads her turn.

>"My head is just full."
"Stretch it out."
>"Yessir."

....

>"In a bit sir."

...

>"I really need the rest please!"
"Fine. Any personal problems?"
None here that really compare to the others.


>Sleep around
>How backed up are you?
>Get laid
I'm fine.
You can be more than fine!

The three here are giving off various thoughts, and are culling their thoughts about me needing a shower.

3/3 Likol 2016/11/15(Tue)15:25 No. 758828 ID: bfb318
File 147925232814.png - (65.02KB , 800x800 , 8.png )

Once they fail not thinking about it, my will is easily superceded by my hive's will. I end in the showers, and we end up taking care of each other's physiology. I ask one member to send a request to Momu to watch the workroom somewhere along the way.

...how long has that surveillance camera been in this shower?
About one week. There are not many blind spots left out of your lab.
What of trust?

An awkward silence when I wonder that.

We realize just how tired I am, but I'm more stubborn about going to bed at 3 PM. I'm unsure what else to do other than either go back to my station, or keep catching up with more details of the outside world and my hive.

They're sensing that I'm not just tired in a sleepy sense, but in a work sense. We've been working more and more, and getting less and less for it. We used to make things, start to finish. Now it seems like we're more of a glorified outsource lab.

The hive reminds me that we don't have the tools or infrastructure to make many products, so we have to sell our research to see products come of it. Vanski doesn't seem to appreciate us so much, now. Much of our work sucks our funding with no real gains. Like my own, thus far.

>>758700

Time to make up a wild story about how not leaving the workroom is entirely critical to ensuring that the emergent AI and the CAI do not learn of the outside world and as a result turn into some kind of vengeful super intelligent ultra-bot that tries to become real by killing everyone...

... no, wait, that would make things worse.

Ask them about their ... hobbies! Surely Korli has some interests of interest.

Begin discussion about how your black-box work has been bottlenecked due to the lack of sedimentary output during the selection process and how a tiny but of explanation on how the game works could cause your research to rocket forward. You have vibrations in the code but there's no interpretation for them, and they're likely marred by sufficient interference. Any brainstorming at this point could advance your work.

>>758826
>the 4 AIs will make it past the RS and override them.
I thought you said a CAI is made of countless AIs? How do 4 AIs make a CAI?

>>758828
Any room for independent projects?

Pat Korli on the head.

>>758827
Poor Korli just wanted the chance to goof off and maybe do a little hero worship and she's getting asked serious questions.

Also come on Likol, you made that socializing more a serious performance review! You gotta lighten it up a little, take it easy on the kids. Tone down the intensity.

>We realize just how tired I am, but I'm more stubborn about going to bed at 3 PM. I'm unsure what else to do other than either go back to my station, or keep catching up with more details of the outside world and my hive.
I personally your for spending time with your hive. You're exhausted, physically and emotionally, and the distraction will help. You could be having more fun (hint hint), and afterwards you'll be better able to deal with your problems from fresh angles (both your research, and how to improve your fraying relationship with Vanski / come up with useful applications for him).

>>758828
Korli seems like she's enjoying herself, at least.

>so we have to sell our research to see products come of it.
To who? Doesn't most the outside world not know your hive exists? Doesn't Vanski not want to share your advances with the outside world?

>Vanski doesn't seem to appreciate us so much, now.
Maybe you need to think of something new he will. Or some way to combine something people are already using into something he'll appreciate.

Application is different from pure research, after all. Maybe it's worth reviewing what we have on the fire, later, when you're rested.

>>758838
Yes, do.

Any idea how things got like this? Is the Salikai paranoia warranted due to something that happened?

Or is external events just making them more worried about discovery?

Either way spend some time just relaxing with your hive and maybe seeing some of the other experiments going on right now. Might help you unwind a bit.

>Vanski doesn't seem to appreciate us so much, now.

You should take steps to make sure his appreciation isn't what you're relying on. Quietly. I'm sure you have already, being smart, but if there are things like cameras going up all over, then it'll be hard to set up any more, and not as hard as it's going to get.

Have you tried injecting data into the system and see if there's any measurable response over a number of cycles? If I remember right Dr Fletch put his image into one the CAI battles, and there was something about a bunch of photographs or other data?

Also, regarding the missing data on the AIs you get from the ring shell, have you tried to emulate them without filling those sections in? Knowing how weird old empire tech can be that could be intentional.

Hmm it seems like you might have been stuck in your lab for to long. I think you had better spend a few hours finding out why things have gotten so bad that there are cameras in the showers now.

As for the Cai just make sure that you have stuff in place to safely quarantine and talk to any Ai's that might start using morse code. The likelihood of it happening soon is small but as long as your systems are in place you can let your watch alert you.

That being said get a secondary notification set up other than the watch. Something secret. If you don't trust the Cai with its own stuff then that watch is a weak point. You take it out into Cai controlled area's and what if they can block your signal? Maybe they don't want you to find out if a breakthrough happens because it would mean they were even more replaceable.

What was the corrupting AI for in the first place?

>>758838
>>758840
DO NOT Korli is a BAD GIRL with no work ethic, she doesn't deserve head pats.

>I'm more stubborn about going to bed at 3 PM.
Why? Is it not healthier to sleep when you're tired and wake when you're rested?

Vanski doesn't trust you anymore, this implies that you can't trust him. Do you have plans for his sudden but inevitable betrayal? Perhaps you can give yourself the option to allow the CAI to study itself?

>>758866
1. You are objectively wrong
2. Being more repressive won't improve anything, any possible bad-girl syndrome is entirely because of unreasonably strict upbringing.
3. Korli deserves all head pats.

Likol you suck at small talk. When we said ask them about there project we meant to see what they were excited about and offer input and encouragement. Actually learn something about them.

Not grill them one by one about what they are working on and why its not finished yet.

Get the grump out of your system and be nicer to the young up and coming science generation. Enjoy your time in the shower with some cute girls for god's sake.

1/2 Likol 2016/11/15(Tue)17:14 No. 758874 ID: bfb318
File 147925886780.png - (14.16KB , 800x800 , 9.png )

>I thought you said a CAI is made of countless AIs? How do 4 AIs make a CAI?
Our active CAI, and all other CAIs I've heard about, comprise of countless. But we can tweak the settings and make a CAI with only 4. It's just not as valuable, and the only reason we do it is to better isolate cause and effects. If we were to make a CAI out of these cycles, we would use the standard procedure with no hard limit on AIs.

>What if the cai can block your watch signal?
The watch is on a CAI sensor network, meaning that while it's not capable of stopping a CAI from interfering, there's no way a CAI can tamper with the network without revealing that it's entering the network.

A secondary system is in place. While more rudimentary, the alarm noise that will happen if a special event occurs will be heard by nearby neumono, and news will reach me, if just a bit slower than the watch would.

>Perhaps you can give yourself the option to allow the CAI to study itself?
I do not trust the CAI with this power.

>Do you have plans for his sudden but inevitable betrayal?
There were some plans, but somehow, Vanski and his children seemed to neutralize any security we put in place.

>What was the corrupting AI for in the first place?
Normally while making a CAI, no error messages occur, but with custom AI that Rihhin made jostles the attempt in ways that generate one. That's its purpose, so that I have something to cross reference.

>Tone down the intensity
Although the hive got me to do something like pat Korli on the head, things are intense. Our young tend to be eager yet unfocused. I was the same way, even if the times were different.

>You could be having more fun (hint hint)
I just did. Even if the point of the shower was to get clean, even I would not think I would do so in the presence of girls without additional activities occurring.

>To who [do you sell your research to]? Doesn't most the outside world not know your hive exists?
Vanski takes care of our outside contact. I expect that he sells them to crime organizations as he will, and as far as our existence goes, he likely just explains that he has a hive and nothing more.

>Any idea how things got like this? Is the Salikai paranoia warranted due to something that happened?
I'd be oversimplifying matters if I thought it was a single event, but Vanski's wife or... significant other, died to a disease recently. He tasked our medical team with a cure. Asking for a cure on short notice for such a complex disease was unrealistic, and although he knows this from a factual point, the medical team's failure to make miracles happen has not sat well with Vanski.

On paper, Stirra, his SO, was simply a business partner that included reproduction. Upon her death, though, he changed in ways that do not reflect a simple lost of cold business.

We're too important to him to just throw us out, appreciation or not. I know that our research isn't sold cheap, and he may just feel like we have too many dead ends. He may simply... no, not downsize us, but have us gradually merge to a few fields where we've had more success. Such as armor and weaponry fields, along with our growing biology field.

>Any room for independent projects?
Not really, but we may want to make room.

>Pat Korli on the head
As soon as I think this, she bows her head.

It was only a thought, not an intention or a consideration.

Nonetheless, it's carried beyond a thought, and despite personal disinterest in doing so, I end up patting her on the head. I would rather not encourage slacking even if it has to be done sometimes, but I suppose I am at the other extreme and shouldn't judge.

Hm? The queen is approaching.

2/2 Likol 2016/11/15(Tue)17:16 No. 758876 ID: bfb318
File 147925897906.png - (18.91KB , 800x800 , 10.png )

I dry off, and am still not dressed before Quokko arrives.

>"Likol."
"Quokko."
>"What are you doing?"
"Taking a break."
>"You suck at it."
"I know. I'll get back to my research."
>"Like hell you will. The only thing you're going to be researching today is how to lighten up. You're bumming down the hive."
"Things aren't great."
>"Remember when there were 15 of us and we were all eating our own damn fingers because we got chased out of our home for the 10th time that year? Things sound pretty glorious to me. Yeah, we're not at our peak, but just because things aren't the best they've ever been doesn't mean we're doing bad.
"Not if we look at it like that, but we're on a decline from the peak."
>"And you're spiraling down faster when you're sleep deprived. You're disallowed from thinking about it until you get some sleep."
"It's 3 PM."
>"Does it sound like that matters? Do you even need a schedule?"
"Almost everyone else sleeps at night. It gets lonely in the lab then."
>"Then sleep for 15 hours."

That is not something I can accept, but my hive is accepting it for me.

>"Get in my bed, Likol."

The other hivemates mentally turn to dirty thoughts, and Quokko rolls her eyes.

>>758875
Well, get in bed. Enjoy a nice dream.

Leer at her and wink before sauntering off to her bed.

Then ask if she is coming if she does not follow along.

Got to tease her a bit if she is ordering us around.

>>758880
Likol doesn't seem the type to have fun, but I'll still support this anyway.

>I would rather not encourage slacking even if it has to be done sometimes,
She's not slacking, she's reminding the so called older and wiser generation to remember to take care of themselves, and relax with their hive. You're all a bunch of workaholics, you need some Korlis to balance you out with a little recreational slacking. Emotional health is important!

Heck, you really need effective management, but the people with all the authority use it to lock themselves away with their projects instead of spending time managing and looking out for the hive. ...and while maybe what your hive needs is outside help to do that for you, you can't really trust Vanski to do it, can you?

>"It's 3 PM."
You're underground. Time doesn't matter.

Also, don't pretend you follow a normal schedule when you're working at your computer. If you can stay up all night and day when it's convenient to work, you can sleep at 3 when it's convenient, you hypocrite.

>The other hivemates mentally turn to dirty thoughts, and Quokko rolls her eyes.
You heard em. Queen's orders. To the cuddle pile!

Do it. She's the queen. You aren't that independent yet, are you? You can at least do this much. Besides, she's right, you need more rest.

This does not bode well, the hive is in decline and the queen is unwilling to do anything about it. That's a familiar story. Don't sleep for 15 hours, wake up while your hivemates sleep and do something before it's too late. It's only a matter of time until Vanski starts surveilling you too.

You heard the queen. Worst case scenario you get the sleep you need.

>>758876
>The other hivemates mentally turn to dirty thoughts, and Quokko rolls her eyes.
Tell her that she started it and has no one else to blame but herself for everyone thinking dirty thoughts.

If she REALLY wants to get knocked up that badly, who are you to argue? Doesn't get much better than paid vacations where your hawt boss compensates you with constant @#$%ing while she's in heat.

1/4 Likol 2016/11/15(Tue)18:26 No. 758907 ID: bfb318
File 147926318441.png - (14.99KB , 800x800 , 11.png )

>Queen is unwilling to do anything about a decline
I don't believe she is unwilling. She just thinks that I get cranky about it while sleeping, and wants me to only think about heavy topics like that while well rested.

>Don't pretend you follow a normal schedule when you're working at your computer.
I'm not, but if I'm to fix it, I should do it right and not sleep at 3 PM.

Nonetheless, the argument is over, and it's me vs. everyone.

>Leer at her

2/4 Likol 2016/11/15(Tue)18:26 No. 758908 ID: bfb318
File 147926319620.png - (14.70KB , 800x800 , 12.png )

>Wink

"Fine."

People think I stink at leering and winking, too. It's also true, even if I'm attracted to just about every girl in the hive.

3/4 Likol 2016/11/15(Tue)18:27 No. 758910 ID: bfb318
File 147926327176.png - (14.44KB , 800x800 , 13.png )

I get in bed. Quokko follows. Like a shower, it's always unlikely to, in normal circumstances, enter a bed with a girl without intercourse occuring at some point. She doesn't let me set an alarm, especially since it would go off in the middle of the night. Nonetheless, Quokko eventually leaves me to sleep.

>Do something before it's too late.
It's more the principle of the matter that I don't want to sleep for 15 hours. Vanski's gradual power struggle with us was done over the course of years. If it becomes too late, it's because we didn't remain vigilant, not because I overslept for a few hours.

4/4 Likol 2016/11/15(Tue)18:28 No. 758911 ID: bfb318
File 147926328689.png - (13.91KB , 800x800 , 14.png )

I wake up at 2 in the morning. I dreamed a little, but nothing worth note. Quokko isn't here, although there's more people in bed instead.

Tell her that over-sleep is detrimental as well. However, a change is as good as a rest. You simply need to engage in mental activity of a different type than you normally would, and that will be both useful and restful, and help you be tired enough to sleep well, when you go to sleep at a time that your body is used to going to sleep. Overall, it will be even better than just going to sleep now.

As for what those activities should be... how about you escort your Queen around, and help her with her duties? The hive respects you as well, a double dose of authority from Queen and elder will help move things along.

You should consider trying to make contact with Vanski's contacts, in some manner. Make your hive some connections beyond just him; someone who will notice if something happens to you. You could sell it to Vanski as, for example, trying to pursue more lucrative research, by talking directly to clients about what they would like. And/or, he surely doesn't want to do all the talking and handshaking himself; salikai are a reclusive species, neumono social, so a few assistants in business relations could be useful to him.

go back to sleep until its time to wake up then you have a schedule to keep then keep it.

Could you sleep in or are you stuck up now?

Either lay around for a few hours or check your watch and try to get up without bothering the pile to much.

Then get breakfast and decide what to do for the day.

>>758912

Argh, update while I was typing my suggestion. Forget everything in that post but the last bit about contacts.

Carefully squirm out of bed and see where your Queen went. I take it this isn't her own normal schedule?

>>758908
Likol: terrible at being a pervert, but his hive encourages him to be one anyways.

>>758910
Quokko: master of the lost secret art the ear-cuddle technique. She is quite literally hugging him with her ear.

>>758911
Well, you can either stay in the cuddle pile for another 4 hours to get back on a normal sleep cycle, or you can sneak out of bed early to get back to work.

...or just peek at your watch, see if there were any alerts.

>>758911
You're not supposed to get out of bed until 6.

Relax for a bit.

>>758908
>even if I'm attracted to just about every girl in the hive.
Jeez, what the fuck are you waiting for? Have a fucking orgy with the entire hive already!

>>758911
I'm pretty sure you can't get out of there without waking someone up and they won't like being woken up or you being awake, so I think your only option is to go back to sleep.

...did your watch go off?

As fun as it is, surely you have better things to do than getting laid. You're all rested up, time to start making some plans!

1/2 Likol 2016/11/15(Tue)20:27 No. 758940 ID: bfb318
File 147927045932.png - (18.59KB , 800x800 , 15.png )

>Jeez, what the fuck are you waiting for? Have a fucking orgy with the entire hive already!
I'm waiting for an appropriate mood.

This bed is communal. No one should be sleeping on it if they get woken up by a hivemate getting in or out of the bed.

>Check watch
The notification counter remains at zero.

I'm unable to get back to sleep. I'm too hungry and restless. Instead, I fetch a fitting shirt in the closet and go get some breakfast. As soon as I enter the kitchen hall, my presence is sensed, and whoever is cooking puts on some additional food in response.

I'm groggy, and the few people who are still up are exhausted. The mood isn't a bad one, even calm, but we're not in the mood for chit chat.

Once again, our forks remind me that 4 prongs work better than 3 for a great number of tasks.

2/2 Likol 2016/11/15(Tue)20:27 No. 758941 ID: bfb318
File 147927046853.png - (23.75KB , 800x800 , 16.png )

Even if I don't sleep, it would be rude to the queen's sentiment to go back to work at the earliest chance. I take a brief walk in the cooler cavernous air.

Quokko is overlooking a small cliffside where we keep a small lot of livestock and plants.

"You don't have much of a daily rhythm, either."
>"You spoke sleepy drivel the other day, with the exception of one thing. We're in decline."
"I hope you're not giving up doing anything about it."

Of course not. We will surmount this, as we've surmounted the past. Not because of some idea that problems have a way of working themselves out on their own, but because we're problem solvers.

If you have ideas, now's the time to share them. But let the queen do the legwork. A salikai will get suspicious if the CAI supervisor starts dealing with the management job.

Alright, I'll repeat what I said earlier.

>You should consider trying to make contact with Vanski's contacts, in some manner. Make your hive some connections beyond just him; someone who will notice if something happens to you. You could sell it to Vanski as, for example, trying to pursue more lucrative research, by talking directly to clients about what they would like. And/or, he surely doesn't want to do all the talking and handshaking himself; salikai are a reclusive species, neumono social, so a few assistants in business relations could be useful to him.

You yourself already demonstrated decent ability to interact with non-hive neumono, didn't you? Quite a long time ago now, but it's a skill. Your hive aren't really the types to get hostile, you're scientists and you have an emotional investment in the idea of being reasonable. And you managed friendly relations and a long-term relationship with a salikai, which given the history of your races must have been harder than interacting with aliens would be, and you have interacted with aliens a little.

It's not science, I know, so it wouldn't be as enjoyable for you, but someone who has an solid grounding in science, and a knowledge of what your group is capable of, would be important to have involved in business dealings of the type Vanski seems to engage in. Vanski surely has his own projects, and enough trouble managing all the construction and his various underlings, and is probably not in the mood to socialize after his wife's death, in addition to salikai generally being not very social.

Aside that, hive members acting as liasons to your clients could better comminicate to the rest of the hive what it is those clients want most, helping you focus your research and produce value more efficiently.

Well here are the obvious points to start with:

1: Prove ourselves to Vanski so that he realizes that even with our failure he could have never made it this far without us and we still need each other. Help with more than just the science so he can not stuff you all in a lab and forget about you. If you are up to your neck in all the systems in the base he will need you for more than just science.

1b: Don't design stuff that will let the Cai replace or take over anything you and only you can do. If Vanski feels he does not need you cause the Cai can do your job you will get demoted.

2: Find ways of regaining some power in the relationship. Make our own outside connections so we are not as reliant on Vanski for outside news and resources. The less reliant we are on the Salikai the less chance they can push you around.

3: Have a survival or bugout plan in place. If things do take a turn for the worse we should have at best a escape route and at the worst everyone in the hive knowing what to do in a emergency.

4: Befriend the Cai. You say you don't trust it but its more of a prisoner than you are. You can't openly go to it with any of the other plans because it's life is in Vanski's hands but making friends with it and maybe finding a way to keep it safe from the deadman's switch might gain you a powerful ally.

5: Develop safeguards for your own tech. Make sure that any specialized tech you build can never be used directly against you. Keep the fact that countermeasures exist quiet and if Vanski finds out say that they exist so our enemies can not repurpose our tech against us.

6: Develop safguards and weapons designed against the other races that make up the Salikai forces. You can easily disguise these as failed prototypes that you are working on redesigning.

7: Consider having a plan (in your head only) for Assasinating Vanski in his lair. If you got him alone with no one around and now had access to the Cai's deadman's switch you would control the facility.

8: Remember that any weapons you develop that are designed to kill or affect Neumono could be used on you once they are done. If someone hates your race bad enough to develop special weapons to hurt or kill you then he is going to secretly hate his "Tame" neumono as well.

Maybe the escape plan is the most feasible, but we'd have to make sure it never reaches the ears of the salikai. Also the entire hive might not make it out. We should be prepared for that scenario.

Alternatively, we could strike a bargain with Vanski to downscale the science hive, by relocating the majority somewhere else that's nicer. If we can improve the situation for the majority of the hive, it'll be worth it.

Whatever we do it has to start very subtly. Salikai can see patterns faster than you can despite there lack of true social skills.

If Valaski notices you changing things in a way that looks like a betrayal he will clamp down hard on you. So make sure your plans are or at least look completely innocent. Even if you just want to be equals in the relationship and are fine working with him you need to be stronger to do so. Right now its not a equal partnership.

Your history as a pure science hive is probably your biggest strength. So taking on crazy research projects that will later help only you would be easy to work on in the open. Just make sure they have a component that Vanski can use each time.

And any plan that will go unnoticed has to be hidden from the Cai as well. Either by hiding it in plain sight or by having so many things going at once even the Cai can not concentrate on it all at once.

>If you have ideas, now's the time to share them. But let the queen do the legwork. A salikai will get suspicious if the CAI supervisor starts dealing with the management job.
The core of the problem right now is the balance of your relationship with Vanski. You are dependent on him for intermediaries to contact the outside world, for resources, for protection. But there's no real trust between the two of you, and he doesn't need you the way you need him. He doesn't depend on you the way you depend on him. It's a one way relationship.

Part of this is you're one of several groups dependent on Vanski slash operating under his umbrella. Your hive, the CAI, the arkots, etc. And Vanski is free to play these groups against each other if they cross or displease him.

One obvious play is to try to skew the balance of power by interfering with Vanski's ability to play other sides against you. The most obvious is the CAI. Your hive (and your research in particular) is the blade Vanski can hold over the CAI. By the same token, it's the CAI's reason to be vigilant against your hive in favor of Vanski (the CAI would prefer an excuse to come down on you and remove you from the field, making it's own position more secure). If you remove or compromise your own ability to hurt the CAI... alliance with the CAI becomes possible. Vanski loses the ability to effectively hold you against the CAI, and the CAI loses motive to act preemptively against you for self preservation. (And if you keep it secret you've removed your capacity to harm the CAI, you can collaborate with it even as Vanski thinks he's playing you against each other).

It's a risky play (since the CAI could rat you out or turn on you immediately) but it's the same kind of game theory risk-reward thing you've watched from studying your CAI simulation cycles. People with no real reason to trust each other, except they stand better odds of survival if they do.

Another obvious play is try and win Vanski over. Require his trust, make yourselves more useful or valuable to him again. Obvious benefits to this, but if you've known him this long, and the relationship has been getting slowly worse, I have doubts you can accomplish this at this point.

Another play might be trying to earn the favor of one of Vanski's children. They compete with each other- make them compete to earn your hive's cooperation with their enterprises exclusively. Make yourself useful to the child as a shield from the parent. (Risky in that young salikai betray their allies as a matter of course. It's how they learn, for goodness sake).

Backdoors and countermeasures in your tech. Planned obsolescence, obscurity, etc. You make it so the systems you're developing aren't perfect. They require you to keep them going. Downside here is salikai and the CAI are both smart- without CAI cooperation, it could likely keep things running as well as you could, and maybe the salikai could too, if they could concentrate on the problem long enough.

Who's been handling relations with Vanski's faction so far? Would it help if you put more of the younger generation on it? They wouldn't have the same bad history, and they have a different mindset. It's possible he might respond to them better.

Escape? If things go bad, people just leave and head for the ultras. (Secretly hide thumb drives with information inside some of the young? Hard part would be leaving them ignorant to their own implants, and them not showing up on salikai security).

Replace Vanski with a better alternative? What they really need are chaperons and smart manegment while they nerd it up. In the present, Sealock is kind of perfect for that role, but that potential teamup can't happen until after things go bad.

Empathy dependent tech? Anything that requires neumono to work would make it awful hard to replace or remove you.

Contacts outside the hive, and that don't require going through Vanski. If you want leverage, you need other options.

Deadman options? If he turns on you, something you made will turn on him, eventually. From a direction he never expected, and delayed to go off long after he's stopped checking for traps.

...there's always immortality through cloning or AI upload as a last resort, but you sure don't have that tech working, yet.

>>758951
Oh, and the secondary problem is your own hive. People like yourself, who have the authority and respect to influence things, use that authority to buy yourself unrestricted lab access instead of using it to help manage the hive. So you're dependent on the salikai and outsiders to mostly take that role from you. You've surrendered power and responsibility in the pursuit of science, and your own interests.

...as loath as you are to admit it.

Be careful with how you talk about these options. If there are any monitoring devices nearby, we don't want Vanski to feel threatened. As you speak out loud, phrase things like you want to find out how to serve his business better and be more productive and valuable. Add the more... detailed context through empathy.

Oh, also... one for you personally, Likol. How hard would it be for you to actually pull out a functional CAI without anyone noticing? The process goes almost right up to the end, anyway. Could you jam a computer into the side and yank out a few AIs into a little handy mini-CAI of your own?

1/2 Likol 2016/11/15(Tue)23:53 No. 758998 ID: bfb318
File 147928280864.png - (19.67KB , 800x800 , 17.png )

"What happened with the connections we made earlier? I know Vanski doesn't like rubbing noses with so many others."
>"He doesn't, and that's what his kids are for. We've made connections. They've all dissipated."
"It doesn't mean we can't try again."

It's more difficult than you think. All those connections we had, for them to just dissolve one after the other into deadends like that... it's unlikely.
>"Sabotage was likely."

I stop thinking about that. Those are dangerous thoughts. Still, Quokko does like the idea if not for the added difficulty of Vanski explicitly hampering our ability to do so.

"I'd like to integrate our equipment into more systems. Make the salikai more reliant on us."
>"The Su'ata salikai would be relucant to let that happen, but... if we're quiet about it... I'll see what we have in the oven that we can bake into the base."

Not stuff that can easily be used with the CAI.

An escape plan? We could escape, but we doubt it would be easy. Either living out in the wild again, or throwing our fates in with an ultrahive. Nevermind one doesn't just 'leave' the underweb with the secrets we have, Vanski or not.

I'll try talking with the CAI again. Sometimes I try to work with them and get some friendliness. It doesn't go far, but it's worth a try.

>Safeguards for one's own tech and other weapons
We have to run it through Vanski. We can't hide hidden projects him for long.

>Assassinating Van---
no

no thinking
Do not ever think that.

Very dangerous thoughts. Which make things like hidden safeguards, escape plans, deadman options, and so on extremely dangerous to even think about.

There is no such thought that only happens in my own head. It gets in my hives head. Then, sometimes, neumono come to 'visit' and ask 'gentle' questions while probing us.

It's safe, here, we monitor this location. No surveillance has been installed as long, so just don't yell.

>Who's been handling relations with Vanski's faction so far?
Quokko.

>How hard would it be for you to actually pull out a functional CAI without anyone noticing? Could you jam a computer into the side and yank out a few AIs into a little handy mini-CAI of your own?
Impossible. The only tech that can generate a CAI, currently, are the four CAI blocks A-D. We only have one of each, in other words, only one container. Making a new CAI means overriding the old CAI, and the blocks themselves are in Vanski's room, where I don't have access.

Even the quantum computer can't store them in their current form. That computer is just a glorified processor that the actual CAI blocks use to help process things faster.

>Would it help if you put more of the younger generation on it? They wouldn't have the same bad history, and they have a different mindset. It's possible he might respond to them better.
He would see them as social prey, and manipulate them as hard as he can.

"Wait. The children."

Of course. They compete with one another. They aren't as high strung, yet at least, as Vanski. They still recognize our use, and our history.

If we can get good relations with a child, we can get protection through them.

More life with salikai favor as another necessary resource?

We need better management, but...

It's easy and will hold us over for awhile while we try for more long term solutions. We'll look into it. Even the more outlandish ideas.

Don't enact any plans that doesn't look innocent on the outside.

"I won't do anything reckless, but trying for subtlety hasn't work well in the past. Let me handle things if something goes sour."

2/2 Likol 2016/11/15(Tue)23:53 No. 758999 ID: bfb318
File 147928281853.png - (15.73KB , 800x800 , 18.png )

We sit in peace a moment, and reminisce for a little while, while it's still dark.

It's interrupted by a ping on my clock. It's not a notification about the Ring Shell, which probably sounds more like an airhorn, or proper alarm. The gentle ping is someone in the lab calling to me.

>"It's Momu get in here ASAP!!!!!!!!!!"

Book it!

>>758999
ASAP means ASAP. Get a move on.

>>758999
Well clearly you should be ignoring warnings from your hivemate watching the CAI. No big deal.

Go get some exercise and cardio. By running. To Momu.

TO THE INSOMNIAMOBILE!

>>758999
>"It's Momu get in here ASAP!!!!!!!!!!"
Did she find something? It couldn't be a sudden error message, the cycle should still be pretty early in the prelims. Don't worry about your queen's order to take a break, it's not really work if something major just happened! Unless something broke down, that is, in which case you shouldn't be on break anyway.

Say, what kind of information can you easily extract from the contestants' life experiences? Even their last coherent thought could prove interesting. Maybe you could focus on that anomaly, and find out why it keeps getting so far.

Speaking of the anomaly, what's the spread of its successes? Did those 50 wins happen fairly randomly, were they fairly evenly spaced, or were there clumps of wins interspersed with massive gaps? That last one could be interesting, since one explanation is that previous cycles are directly influencing later ones in some way that affects who survives.

Looks like the mystery box is finally doing stuff that might make it not a mystery, to the lab, for Science!

>>758998
You left out maybe the most promising idea, which was trying to win over the CAI by removing the knife you currently hold over it's head. (And maybe not telling Vanski you did).

Which is risky for obvious reasons, but if you remove the ability for Vanski to play your hive and the CAI against each other, cooperating in mutual self interest, instead of being wary towards each other with an eye for survival, becomes possible.

>"It's Momu get in here ASAP!!!!!!!!!!"
Momu's paging me. Something interesting must have actually happened. Excuse me.

>>758999
How many bangs (!) is it before the Momu-meter trips to full "oh fuck"? Your other hibernate I could see sending all that, but Mini …?

1/2 Likol 2016/11/16(Wed)11:49 No. 759079 ID: bfb318
File 147932575194.png - (43.83KB , 800x800 , 19.png )

>Win over the CAI by removing the knife you currently hold over it's head.
I'm conflicted on this. To remove that would mean shutting down the experiment such that no more cycles occur, which would not take long for Vanski or another Su'ata to notice. Even if the CAI became friendly, Vanski would still have other knives over their heads, namely being in control of the physical CAI blocks. I don't believe the CAI is hostile to us, either. They have no reason to remove us; their position is secure enough, being a CAI.

This reminds me why it's difficult to speak to the CAI at length, too. If Vanski notices I speak to them much, he may suspect collaboration.

>Say, what kind of information can you easily extract from the contestants' life experiences?
Easily? Very little. Just a few facts, like if one is still alive, what stages it's gotten through, and what other IDs it's come into contact with. Which could mean that ID became their best friend, or they entered their line of sight at some point in time, but there's no way to easily tell which.

>Speaking of the anomaly, what's the spread of its successes?
Random, more or less. There's significant gaps here and there, and it's been a long time since I've seen it, but it's difficult to tell if that's significant, or simply the usual odd spread that can occur with random distribution.

"Momu's paging me." I tell Quokko.
>"How convenient."
"I'd better go, in case it's something."
>"Don't work all day."
"I already haven't."

>How many bangs (!) is it before the Momu-meter trips to full "oh fuck"?
More bangs just make it sound sarcastic. In any case, the only real thing she can send out from the lab are loud noises, due to the security measures in which the lab practices radio and wireless silence. The wireless receptor outside the station then reads the noise, and then commits an action based on the noise - in this case, it was to send a preset message to my watch. Rihhin probably got it to, but she's most likely asleep.

>Get a move on
>Book it
>Run
Just in case, I don't waste time.

2/2 Likol 2016/11/16(Wed)11:49 No. 759080 ID: bfb318
File 147932576951.png - (25.98KB , 800x800 , 20.png )

Momu feels like she's going to explode waiting for me to finish the scans, showers and so on. She rushes up to me as I approach the room.

>"There you are! Look at this!" she says, gesturing to my desktop.

"ID numbers."
>"Nearly all the regular winners have made it through the first preliminary stage!"
"If you think about it, that's the most likely scenario."
>"Yeah but there's so many scenarios it's still really unlikely so now all the usual winners are doing well!"
"Momu... I don't know what that has to do with our research. Is the corruptor doing anything yet?" I ask, but I doubt it. It remains dormant for the first few stages, which is a couple of days real time.
>"No but I bet it's hyperactive this round!" But I don't know yet.

It's been active in almost every single cycle regardless, but sometimes it's far more active than others, with no middle ground between regular active and hyperactive. To date, we haven't found a reason why its activity, on a scale of one to ten, either is above a 7 or below a 5.

After a minute of empathizing with Momu, I confirm there is nothing new, and nothing interesting, about this cycle, as there never is in the first couple of rounds. She is just excited. Excited about science. She is impervious to my apathy, thankfully.

Pat Momu on the head. Check in as much detail as you can, just in case. If it's nothing, back to Quokko or what serves as the lounge, I guess.

>what other IDs it's come into contact with. Which could mean that ID became their best friend, or they entered their line of sight at some point in time, but there's no way to easily tell which.
If you reset this parameter repeatedly and log results, you'd have a list of associations over time. Would allow you to better hypothesize which AIs are choosing (or forced) to associate with one another.

>>759080
Thank Momu, head pat, give her reassurance to calm her down.

Thank you for keeping me informed, but please try to reserve the asap alert for anything new or unusual.

What now? Either you keep planning contingencies or other ways to improve the situation behind the safety of your less monitored lab, or you get back to work. What work is there to do early in a cycle?

>She is just excited. Excited about science.
That's a good attitude to have, even if it doesn't really mesh well with the situation y'all're in.

I suppose you should keep a close eye on them in case they do something interesting this time around, though I suppose you were going to do that anyway, because that's your (entire?) job.

I'm guessing that any line of communication that could lead from the AIs inside back out to the real world would be a problem, but you have power over what they experience, right? Have you inserted any deposits of information or recorded messages that would tell the AIs in the sim anything about you or the wider situation or that would just maybe help them feel better about the whole thing, on the chance that they are sentient in there?

What is the protocol for if things start going really off-track in there, anyway? You are scientists, and it's from aberrations that you learn the most. If the system was isolated, I'd guess the plan would be to just let whatever's in there fall over all the way on its own and then look at the records, but if it's connected in to the main CAI, I'm guessing there's a certain cut-off point where the threat has to be dealt with? On the other hand, you did say the CAI was replaceable. How much deviation, how many anomalies do you put up with before you press the forced reset button?

What would you do if, for example, one of the cycles just stopped progressing? Not as in totally freezing, but like if it still stayed active but kept going forever without reaching the final stages. Would you just observe whatever activity you could?

This could be a sign of a higher intelligence in the machine. If the trend of high survival rates continues, then the most likely cause is that something is tinkering with the process, and if it isn't an outside hacker then it has to be an inside hacker. Luck or intelligence, you could boost your funding by capturing the AIs responsible for this.

... Is what you'd say 3,000 cycles ago. Except you're not that gullible anymore. Tell Momu that this is a good case file, but you expect things to go back to normal once the process ups its ante and pushes the AIs into a corner and forces them to sacrifice a few survivors. You can observe this cycle because its one of the few that managed to induce mass cooperation, indicating a better CAI build, and you can even expect them to breeze through the next two trials thanks to this overwhelming horde of cooperating specialists. And scapegoats. But once they get to one of the trials where saving almost everyone ISN'T an option, they'll either fight it out or refuse to play further. You can start yelling AFTER they beat something that specifically selects a fixed number of surviving AIs.

Don't squash that enthusiasm. You want her excited about possible breakthroughs. Maybe nothing will come of it but don't infect her with your apathy.

Honestly her attitude is just what you need around her. Having watched this for so long you have grown cynical. So if there are any sudden breakthroughs then you might not be in the right mindset to see the signs.

So help her set up some extra monitoring so you can see if this is a fluke or something worthwhile. Maybe its going to be nothing but will be a good learning experience for her. And you teaching her will help your mood.

You know you could talk to the Cai more without suspicion if you proposed it up as a experiment to try and speed up your research. Along the way you could slowly get to know it better and let it realize that you have its best interests at heart. If Vanski ever does turn against you having the Cai be friendly will help.

>>759080
>haven't found a reason why it's never a moderate amount
Maybe you should be thinking from the opposite direction. Why is the stabilizer not allowing a moderate amount of corruption? Why is the stabilizer allowing a large amount of corruption? Maybe your creation has developed its own agenda. Did you even tell it what it was made to do?

Anyway, you said earlier that changing a contestant's hidden values would drastically alter their behavior. The simulation as a whole can change based on different initial conditions. This is new, even if it hasn't done anything interesting yet. Any particularly extreme initial conditions can mark a particularly interesting cycle, but it may take a while for those initial conditions to develop.

So keep an eye on things. Track their progress, and see if you can spot any aberrant IDs following along with the successful AIs.

Give it to her as a project.

Have her follow those id's and compare them to previous cycles. She can track them over the stages and see if this is something special or a fluke.

Have her tracking and collating the data will be fun for her and no matter what way it goes you will have some extra data to add to your calculations.

1/2 Likol 2016/11/16(Wed)16:40 No. 759141 ID: bfb318
File 147934323314.png - (14.73KB , 800x800 , 21.png )

>If you reset this parameter repeatedly and log results, you'd have a list of associations over time.
We have, actually, considering it's one of the few things easily gained.

So we do have a list of frequent interactions, at least, but it's often difficult to tell if they're friends or enemies.

Ultimately, though, the lives and experiences of the AI aren't part of the experiment, and are just a passing interest. It's an entirely different field. The only work-related interest I have with them is that they, namely the corruptor, is what causes aberrations in the RS, and the RS is what I study.

>Have you inserted any deposits of information or recorded messages that would tell the AIs in the sim anything about you or the wider situation or that would just maybe help them feel better about the whole thing, on the chance that they are sentient in there?
I have. Out of some weird sense of old amusement, I gave them old pictures I used to take. They are made to be a blotch inside of Block C that is easier to isolate out than the AIs or stages or other information, and as a result, I have an easier time finding out if they've been found or not. The whole purpose of them is just a checkpoint system to confirm that the constestants are making it through the stages in the first place.

Or at least, that's all they're useful for. The other purpose was just me experimenting with inserting things inside of the stages.

In any case, I can't imagine they're sentient. There's simply no way that so many sentient, self aware beings could all be processed so quickly. Arza wanted us to believe they were, but he spoke highly of even the most rote AI, and I believe he meant it in a best-practice kind of sentiment. Much like how people on teams are told that everything that their team does is their own individual fault - maybe useful as an attitude, but wildy untrue.

I do not recall when my hive's lower generation got a taste for head pats, but it's not something I can reasonably fight.

2/2 Likol 2016/11/16(Wed)16:41 No. 759143 ID: bfb318
File 147934328938.png - (37.73KB , 800x800 , 22.png )

>Why is the stabilizer allowing a large amount of corruption? Did you even tell it what it was made to do?
My only guess is that it normally tries to limit the corruption, but if it gets corrupted itself, then it backs off the next cycle. If that's the case, then the corruptor always attempts an 8-10, but if the stabilizer is active, it reliably gets dialed back to a 2-4.

It's not supposed to stop the corruption. It's just there as backup. I can't implant memories in contestants, but in the beginning, it should have received a note from me explaining its purpose. It seems to be more proactive than intended, but I haven't deteced any problems from it.

>What work is there to do early in a cycle?
Typically that's when I look at the RS and the error logs of the previous cycle.

"This is a good case file, but the stages don't get any less dangerous overall. Keep your focus up, but don't count on anything. I'm going to get to my work."
>"Okay! I've translated the last batch of error messages. What do you want me to do?"
"... study the AIs."
>"Whoa, really? I mean, I'm not complaining, I can keep imprinting the RS visuals on my eyeballs if you want!"
"I might be going about this in the wrong direction. Perhaps instead of trying to read the RS, we should read the AIs and see how they influence the RS."

Granted, we've tried this before on countless occasions, but Momu might have more luck this time.

There's also a tiny chance that the corruptor and the stabilizer may have the tools needed to find out about the ringshell, and if either of our custom AIs can learn to use the RS to communicate with us, that would make our jobs as good as done.

Parallel Unregistered found.

That's the common one. Rihhin's AI creates its own sub processes.

After a more in depth look at my stabilizer, it looks like its spawn got corrupted at some point in cycle 3118, which does mean this will be another hyperactive cycle after all. I start getting a faint glimmer of hope, but it comes from Momu.

>What is the protocol for if things start going really off-track in there, anyway? You did say the CAI was replaceable. How much deviation, how many anomalies do you put up with before you press the forced reset button?
The CAI is, but the blocks are not, except for Block B which can be replaced without too much hassle. My stabilizer can repair block C if that goes poorly. Block D, however... if that starts going awry, the whole system will freeze if it gets too far. That is irreplaceable. So far, though, it's intact, and it seems to be a self repairing, self replicating system. If there's been any damage to it, it's been so miniscule as to be undetectable underneath all its noise.

If the RS looks like it's taking damage, though, is when we pull the plug on this arch cycle.

>What would you do if, for example, one of the cycles was active, but not progressing?
I might take a more in depth look at the contestants and block C than I normally do, but ultimately, there is a function that force starts a new cycle. Once, somewhere on cycle two thousand something, the cycles didn't seem to be restarting properly. I used the force new start function, and that seemed to fix the issue.

There's not much to monitor for today. Beginning tomorrow morning, things may start picking up. I'll just inspect the previous cycle to see if I stumble on anything.

It is true, although Momu calms herself around me, her enthusiasm does rub off back onto me.

Before I know it, it's 8 PM. I haven't seen Rihhin, but she has other things going on. Momu is still excited, if tiredly so.

"I bet youuu...." Momu breaks the silence as she gets an idea. "That this cycle is going to stand out past the rest! And if it does, then you have to do my share of cleaning next time I get put on cleaning duty! And if it's just a ho hum cycle that doesn't help us, then it's reversed. How about it?"

This bet again. She just likes it when I let her clean up after me.

Take the bet just for fun. Win or lose you get to socialize and do something out of the box.

Take it. Even if it's just something she likes, it's nice to do stuff with hivemates.

"Any ideas how it'll stand out?"

>>759079
>it's difficult to tell if [the gaps are] significant, or simply the usual odd spread that can occur with random distribution
Yeah, it's admittedly hard to get a good sample size, but what if it's not random? Maybe the cycles aren't completely self contained, and some form of information is passing into future cycles. Do we have any way of testing for that? If we can confirm it and learn how it's being done, can we inject our own information to influence the contest?

Hmm... do you still have data on the anomaly for the cycles it didn't complete? It if just suddenly starts consistently dying at a specific point that would point toward some sort of extra-cycle influence that it had to overcome.

>>759143
>If the RS looks like it's taking damage, though, is when we pull the plug on this arch cycle
Would you even have time to pull the plug? With the processing speed available to it, if something in there does figure out how to damage it then it'll probably do everything in a matter of seconds, if not faster.

> There's simply no way that so many sentient, self aware beings could all be processed so quickly.

You're using a supercomputer designed to build and improve upon what took the universe billions of years to accidentally come up with through trial and error, and it's capable of processing and simultaneously testing one trillion different autonomous programs. At a fraction of lightspeed. With minimal user input.

Face it, Likol. They are smarter than you. They are also working together, which means they can cloud compute their "living" processes, the parts that take up most of the AI space. Who's to say they aren't even more sentient?

>I haven't seen Rihhin, but she has other things going on.
Check on her, at some point, maybe? She checked on you, earlier.

>This bet again. She just likes it when I let her clean up after me.
Indulge her. It's harmless fun, and if she actually wins the bet you're going to be so excited you won't even mind having to clean for a change.

>Momu is still excited, if tiredly so.
Should probably send her to bed at some point. You've slept more recently and can watch longer. With the logic that she wants to get her rest in now before it gets really interesting. If she's right, she won't want to miss that part. (And of course you'll page her if things go nuts).

Guessing "Parallel Unregistered found" is the corrupter spawning devotees, and the force new cycle around cycle 2000 or so was when glitcher and corrupter tried to perpetuate the cycles.

>>759143
>>759150
This guy raises a good point. Think of it this way: what if it's not processing a trillion AI, but a single AI that's pretending to be a trillion AI?

Well if you are that confident then why not.

But that means more projects for us. If we suddenly get morse code this cycle then we need be ready to figure out what our new baby Ai is up to and how we can talk to it. Might be a pretty simple little guy so we might need to have a way to translate ready.

Plus we should have a test server to transfer it to if possible so we can run tests and see how it reacts to life outside the shell.

Getting to talk with a proto-Ai before block 4 hits it might give us all the info we need to break this wide open.

(Sure you are humoring her but she will enjoy the challenge. Plus gives you a project to work on instead of worrying about your underworld patron.)

Take the bet.

Just so we can be sure what normal and abnormal are, could you give us a brief overview of the whole process from beginning to end, what each of the boxes do, where the ring shell is involved and how, et cetera? Along with what modifications and insertions have been done to the system, and what each such was intended to accomplish? Were any in-built systems suppressed or removed?

Also, what's the best speculation for how these ancient systems were originally "supposed" to run? How much of what's in there is from the original ancient belenosian work?

>>759143
>Parallel Unregistered found.
Does this always happen in hyperactive cycles? Does it EVER happen in non-hyperactive cycles? Also what does that mean? Are the custom AIs "unregistered"?
Don't you think it's a bit odd that the corrupting AI seems to produce exactly twice as much corruption in hyperactive cycles? It's almost like there's twice as many AIs producing corruption, not that the stabilizing AI backed off. Why would it do that every time, anyway?

>can't be sentient(sapient), too much processing power needed
Hey, what if the Ring Shell's AI is really the only AI in the box? Like, it's the brain and all the contestants are sub-processes running in it under different variables? The Ring Shell could have a lot of processing time by running this way since it would be able to make no-brainer choices for everyone at once and only have to really make individual choices when there are hard choices or important environmental/personality factors.

>I used the force new start function, and that seemed to fix the issue.
Did you do that more than once? Like around cycle 1000-something? Also, did anything else of note happen during the cycle had to use that function on? Like perhaps did your stabilizing AI get disrupted? Did you code anything in to protect your AI?

Accept her bet. It's a win-win scenario for you. If something interesting happens, you won't mind losing the bet, will you?

1/2 Likol 2016/11/16(Wed)21:22 No. 759231 ID: bfb318
File 147936012503.png - (31.53KB , 800x800 , 23.png )

>Maybe the cycles aren't completely self contained, and some form of information is passing into future cycles.
They certainly aren't, although in an ideal run time, they effectively would be.

>Do you still have data on the anomaly for the cycles it didn't complete?
Yes, we log everything, and I confirm that we still have empty hard drive space for the happenings in Block C and the RS.

I have Momu look up the anomaly of overly generous game theory. She says she's looked into that, and explains that more often than not, the anamoly dies in either stage 3 or stage 7. The former due to another strong performer, the latter due to the simple fact that that stage has a 99.9% mortality rate anyways.

>Do we have any way of testing for that?
Unfortunately, not easily. Moving from one cycle to the next is something the RS operates, and while I have spent countless days trying to read what the RS is doing, I haven't got far.

>Did you force-start around cycle 1000-something?
No. I noticed cycles around then got strangely influenced, and I have some bugged information on the cycle, but the RS did not seem to get affected differently. I would have been worried, but after a few weeks, it fixed itself.

>Did you code anything in to protect your AI?
No. Block B should be isolated from Block C, still, so even if they are damaged in Block C, their origin, B, should remain untouched.

>Did your stabilizing AI get disrupted?
Aside from the corruptions, its presence in the stages has been eradicated a couple of times. I've looked, but it didn't seem like the RS reacted, or at least, if it did, it was such a minor change that I never could find it.

>They are smarter than you.
I'm sure, but I don't believe they are self aware, much like a calculator can do a hundred thousand calculations in the time it takes me to do one, but I'm aware of one more calculation than it is.

>What if it's one AI controlling a trillion others
It's - I can't rule that out, at least not from a literal standpoint.

Yet from what Dr. Fletch has told me, the initial one trillion AI share many thought process to save on computing speed, and then slowly become more independent. Plus, Block B has individual entries for all included AI. I suppose it's still possible that it's just one computer system dividing its thoughts and just uses Block B as a guide on how to divide them, but they can definitely be thought of as individuals on some level. I can't help but think of the computer like the hive, and the AI like individual neumono, but that may be more projection than logic.

>Brief overview of the whole process from beginning to end, what each of the boxes do, and how much of what's in there is from the original ancient belenosian work?
Block D contains the RS, and the platform where the active CAI sits on. It's essentially the operating system, plus the active CAI. Block B stores the base AI in its pure form. Block C is where all of these attempt cycles take place. Block A acts as the primary processor, as well as the physical wrapping box the other blocks fit in.

Block B is largely customized, either stuff my hive has added, or things aliens added. It's the most solved and simple block. Block C is largely customized by Momu, but only on the surface level. There are lots of unknown functions that get the cycles running.
Block A is unchanged, and studying that is a field for hardware specialists.
Block D is also unchanged. It's been cracked, but that just means we can observe it.

We don't have a good idea if this was common tech in the belenosian empire, or just a cutting edge experiment. Dr. Fletch thinks this might have been an obscene piece to the times, which had very strong notions of one body, one AI.

Once a CAI is made, blocks B and C seem to have minimal activity, but Block D won't work without Block B and C.

>Would you even have time to pull the plug [on the RS]?
Maybe, but we also have a program that can detect major damages on the RS, and it's designed to pause the simulation. However, I'm not entirely sure how the detection works, and how well it pauses the simulation, but since the RS is irreplaceable, there's little choice but to use it.

>Does [Parallel Unregistered found] always happen in hyperactive cycles?
I said common, but it's more than common. There's only been a couple times where this error message didn't occur, and there was effectively no corruptor activity that cycle. The message itself means there was AIs - rudimentary AIs the corruptor made - that all share the exact same values.

>Don't you think it's a bit odd that the corrupting AI seems to produce exactly twice as much corruption in hyperactive cycles?
I used the 1-10 scale arbitrarily.

The hyperactive cycles aren't just double the amount of corruption that occurs. It's more like it's exponential.

Still don't know why. A lot of guesses, not much evidence.

>Plus we should have a test server to transfer [morse code user] to if possible
If it's an AI, I don't have anything it can use. If it's not an AI... then I don't see how it can use morse code.

"Bet accepted."
Yes!
"You should sleep. The interesting stuff will start happening tomorrow morning."
>"Only if you do it too!"
"I woke up after you did."
>"By 10."
"... Fine. I need to check on Rihhin, anyway."

I say that, but I continue watching the ring shell with no UI. Dr. Fletch says it's sometimes important to do so to understand its language. His analogy spoke of studying a foreign language that uses a different character set. One should use the foreign alphabet, not the foreign alphabet translated to native letters.

Blocks do not seem like they should move around like sludge, but that is what the RS seems to do.

2/2 Likol 2016/11/16(Wed)21:22 No. 759232 ID: bfb318
File 147936016533.png - (25.94KB , 800x800 , 24.png )

I leave to go where I believe Rihhin is. Working on weapons AI, with Norr supplying her with direction.

>"What's up, Likol?" Rihhin asks, while Norr mentally greets me, and I to him.
"Just checking. You usually don't stay out of the lab all day."
>"Momu was in there, so..."

It's fine, of course.
Anything new?
No.

...

>"Geez, we're making AI for weapons with no quality control. My AI has to be able to learn every unique weapon to be useful. I'd rather be making AI for our weapons.
We're working on it."
>"Oh, Likkol, Reon wanted to check out the lab. We got clearance from the Su'ata. Mind if he gets a one day pass to check out what you guys do tomorrow?"

Rihhin appreciates the stop in, but she is busy after all.

Headpat Rihhin. What's Reon like?

NUZZLE Rihhin.

>>759232
Reon stopping by tomorrow would be perfect. The current cycle is a little different than usual so something worth seeing might happen.

>>759243
Sounds good.

>>759234
Probably obligatory at this point. Though now I wonder how many hivemates you have to headpat before someone comes up to you for that express purpose...

>>759231
>more often than not, the anamoly dies in either stage 3 or stage 7
I meant in the large gaps specifically, but that's good to know.

Do the contestants constantly run stages, or do they get rest periods between them to relax and socialize?

>Reon wanted to check out the lab
Sure, he can come by.

1/2 Likol 2016/11/16(Wed)22:24 No. 759257 ID: bfb318
File 147936384195.png - (18.20KB , 800x800 , 25.png )

>Do the contestants constantly run stages, or do they get rest periods between them to relax and socialize?
Assuming that no stages end up in strange deadlocks between contestants, they will spend more time in a relaxing, pressureless area than actively completing stages.

The thought of headpatting Rihhin crosses my mind.

Its rejection is swift this time, and does not make it to any thoughts of nuzzling. The generational gap includes more than just attitude.

"I do not know Reon well. Tell me about him."
>"Imagine Momu level energy and enthusiasm, but instead of outwardly energetic, he can swallow his energy."
"Choking on enthusiasm, then."
>"That's one way to put it. There's no problem, right?"
"I'm not expecting anything major tomorrow, so even if he would get in the way, there would be nothing to get in the way of. Tomorrow would be good, in fact, especially since we have some good examples of AI to show."
>"Cool. Don't worry about initiation, I'll show him how to enter and exit the place and give him a one day pass."
"One day?"
>"Hive or not, Su'ata still know we're made up of individuals. Don't want any more keycards floating around than necessary."

Even if we also need our faces, eyes, and voice...

I remember I couldn't get in for 3 days because I had a cold.

And that's why I don't leave if I feel like I'm fighting something.

I leave, with nothing more to offer except distractions.

2/2 Likol 2016/11/16(Wed)22:24 No. 759258 ID: bfb318
File 147936388224.png - (18.76KB , 800x800 , 26.png )

>"Ah!"

A salikai voice? Kiiu, I believe.

>"Likol! I was hoping to see a neumono of importance, and you're right there instead of in your laboratory. Do you have a moment to speak?"
"Yes."
>"I'm making my rounds, as it is beneficial to keep track of all of our little projects we have going. I'd like to hear about your CAI project."
"If there was anything new, Vanski would hear of it first, and then you, among your siblings, second."
>"Specifics, Likol! I'm sure you've been doing something? Think of it like practice of making daily reports to an investor, since you all clearly need it. This is why my father is the manager and speaks to the client, as you must understand. Oh, and if you don't feel comfortable speaking, here, I have a nice little alcove nearby that is designed for calm discussion."

>>759258
Headpat Kiiu.

IF we're willing to give him information, which might at least be a good way to get out of Vanski's thumb, then: "Yes. The simulation is acting somewhat unusually, although in a way thus far consistent with many previous iterations. Distributions and error messages have matched a small proportion of simulations thus far, but one of my assistants is convinced it'll develop into something more unusual."

Explain in as much more detail as you feel comfortable.

>>759258
Progress is slow. You're dealing with heavy encryption. Today you're looking closer into how the AIs interact with eachother, in hopes it'll reveal more about the overall structure.

Jeez, we're turning Likol into Alison! The difference is instead of hugs, he's doing headpats! And he's able to ignore us when we tell him to headpat, unlike Alison, who always hugs when we tell her to!

If talking about it out here wouldn't violate some directive to hide certain things, from the CAI for example...

"I'm fine here, if you are. Daily reports? Given the nature of the project, I'm not sure that would be the way to arrange things most effectively. We need to be prepared for problems and opportunities, and when they occur is when we earn our keep the most, but where and when such things arise is effectively random. While we wait, all we have to do is what is now regular maintenance and observation, for which daily reports would, despite the training required, amount to "the same as last time". I was under the impression you would prefer us not to waste your time, and ours, on the equivalent of ringing a bell and shouting "all's well". We are currently tracking a note of some potential interest, but that potential has yet to manifest, and may not."

>Think of it like practice of making daily reports to an investor, since you all clearly need it.
>Offers you a place to be concealed from monitors.

I think the salikai might suspect your concerns with your hive's future.

Likol 2016/11/16(Wed)22:55 No. 759276 ID: bfb318
File 147936574247.png - (16.85KB , 800x800 , 27.png )

"Alright. The current cycle is unusual but not abnormal. Distributions and error messages have matched a small proportion of simulations thus far. One of my assistants is convinced things will develop into something more abnormally, although I'm more cautious in my optimism."
>"Does it mean there will be progress?"
"If it's abnormal? It will pave the way to more progress, but we're dealing with encryption the world has not been able to get far into despite huge teams devoted to the cause."
>"Yes, but they aren't doing what we're doing, are they?"

He starts skittering towards me.

>"Tea?"
"No caffiene please, but otherwise yes."

He makes some clicking noise to the arkot, then the arkot goes presumably to get me a drink.

It starts climbing over Kiiu instead of going around. Its bare feet practically kick the salikai before it manages to roll over and land with a soft thud. It scampers off without noticing Kiuu looking at it like it's never going to be seen again.

Kiiu turns back to me. By now, a few hivemates are watching me from the windows, making sure I'm doing alright but not wishing to intervene if they don't have to.

>"Is there anything we can give you that might help? This is our longest project by far, we'd hate for things to reach a dead end. Or any questions about what you are doing. I cannot promise you I'll answer anything, but I will promise you that there will be no retribution for any pointed questions, even if I don't answer it. So please, ask away."
"I hope you weren't suggesting daily reports. Given the nature - "
>"No, no, just to think of this like a daily report. Not to make them, of course. What other questions?"

Headpat the arkot.

"Right at this moment, I can't think of anything other than more processing power. I've considered a couple of novel methods of dealing with the simulations, but the reason I didn't implement them was that, frankly, they're not likely enough to work to be worth the resources. However, even if more unorthodox methods fail to yield results, more processing power would at least speed things.

And now that we're being offered additional resources, I'm sure that we can think of something more creative and less expensive than just exponentially more computation by tomorrow. As I said, my assistant has ideas and many other hivemates have creative new ideas in different fields. With the sort of nanobots that make up bioarmor being possible, we may even be able to think of some sort of biology-based solution."

>>759276
Hmm, how about... if Vanski eventually decides your hive isn't worth the resources he's spending on it, would he take you in? At least, those neumono who aren't involved in top secret projects.

"Hmm. Well, right now, I'm concerned that you're concerned with our progress. I'm sure you know, as I'm sure was known when the project began, that whether or not we make progress is not entirely within our control. That could theoretically change, if we had tools allowing us to better observe the internal processes of the system, to communicate with AIs in the system, or to make significant adjustments without needing an arch-cycle restart. I assume, however, that those tools aren't available, or I would have heard of them from Arza Fletch. We might also be able to make more sense of the readings we get if we had a clearer understanding of what additions and modifications were made to the system besides our own, and what the intentions behind those adjustments were. We might also, possibly, be able to produce more satisfying results if exactly what you want or don't want out of the project was laid out, to direct our attentions."

"If I'm dreaming of ideal resources, what would be most useful would be another set of CAI boxes, with a duplicate system that could be run at the same time, as a control for whatever adjustments are made to the other, as proper scientific method would require. I doubt the likelihood of that, however. Thinking of the scientific method, however, the concept of peer review comes to mind. More contact and communication with other scientists performing similar research, like Fletch, would potentially help us make sense of our own work."

>>759279
It's too early for big plays like that, stick to just helping the project for now. Even if no retribution was emphasised, we don't want them to know we're looking for a escape.

>>759280
This, better understanding of the system we're working with will be helpful.

Another idea: If any have been recovered and reverse engineered enough to be functional, ancient belenosian tools for interfacing with this ancient belenosian system could reveal things that our current tools can't. But again, I'd assume (Dr? Professor?) Fletch would already be tracking such developments.

>>759283
>we don't want them to know we're looking for a escape.
Well I mean if we emphasize that it's not an escape but a place to go if Vanski doesn't want them anymore, it'd go over a bit better. Besides the hive doesn't really want to leave they're just nervous about getting betrayed and killed off.

Wait, can we even speak openly here about the research? The CAI can't hear us here can it?

>>759232
Not sure what salikai and/or science hive weapons we've seen that need their own AIs. They sure don't seem to use smart guns. Maybe the companion AIs for the greenie armor (like Bell, except she came from Katzati and somewhere else), or maybe it's targeting AIs for the auto-turrets?

>Its rejection is swift this time, and does not make it to any thoughts of nuzzling. The generational gap includes more than just attitude.
She rejected the thought, or you self-censored?

>>759258
Hey, look, more dead characters.

You should indulge him a little, Likol. You raised the idea with Quokko earlier of trying to win over some of the Su'ata kids, even if you're losing favor with Vanski himself. Part of that probably means talking to them.

And sure, he's probably looking for an advantage over his siblings by getting advance information, but that's to be expected.

(...also, Kiiu was by far the most reasonable and responsible of the salikai. If Rokoa hadn't killed him I think Polo could have kept working with him, maybe even tried to prop him up against the more unpleasant factions of his family).

>>759276
>Is there anything we can give you that might help?
If we could somehow convince Arza Fletch to come on board that would greatly accelerate things, but as things are we have had to conceal the extent of our research from him. He would disapprove of what we're doing on a moral or philosophical level.

A tool like the CAI itself would be fantastic for analyzing the data and accelerating the results, except of course we cannot trust the CAI with detailed information about it's own inner workings and strive to keep it out of the lab at all costs. Not that I am advocating this change, but is a significant irony.

>>759276
>Is there anything we can give you that might help?
Every species has its own neurological quirks. Having someone who is not a Neumono to offer their perspective on the data could be helpful.


Hmm... on the subject of making a better CAI, I don't think taking the first group to come through is the best way to go about it. There's too much luck involved. Running the contest numerous times to see which AIs are likely to finish and then making a CAI out of your personal favourites seems better. Like, for Vanski's case, taking the ones on the administrative track that never jump ship and enter the stage area. Maybe grab that anomaly, too, to have an AI that's good at dealing with people. The others that are allowed in can temper any rebellious tendencies it might have.

First move this meeting to somewhere the Cai can't listen.

Well I have several ideas to improve things but they would require a large change in our current direction of research.

Our current methodology while effective is extremely slow. It's like we set a ant farm on a dictionary and are watching it through a telescope waiting for it to start building letters out of sand to talk with us. Eventually we might get results and we learn a lot from watching the ants but is not the info we want.

While Arza thinks our methods are "unethical" they are probably not unethical enough if we want results faster. We can't make any major changes to our methods for fear of breaking the current Cai. Any time I have a great idea for speeding things up I realize its not worth the risk of breaking our only set of creation blocks.

So here is some stuff that we could use to speed things along.

1. A test server to pull out and isolate Ai's before they are packaged. If we can extract one or more of the promising contestants and look through there memories we could find the traits Vanski is looking for.

2. A second D block so we could spin up a new Cai off the current cycle without erasing the old one. I know getting another one is unlikely but if we had it we could actually spin up new Cai's to test every week. Then we could compare data from earlier and hand pick what Ai seeds go into the new Cai.

3. Scheduling a meeting with Arza would be nice. Keeping him in the dark would be tricky but he is the best Ai tech. Maybe we should let him in on what we are doing? I know he suspects it already. But maybe we could lure him onto the team with the promise of giving him his "children" back if he fixed our problem. Would give him a incentive to help solve things quickly.

4. More opportunities to research on the current Cai. I shudder at the thought of letting it get to close to its core but we should get a team set up to study its mannerisms. Our ultimate goal is to get a Cai full of Ai's that would be eager to work for our organization. But we can't do so without finding out what our current ai's individual personalities are like. If we can learn more about our current Cai members then we can cross reference them to there originals in the contest.

Sorry to be a distraction, but I was thinking on that sentience question some more. Sentient/sapient beings aren't really that way all the time, are they? For the most part, most people follow a routine in their daily actions, interactions and thought. They get "trained" by the habits of their life and work and follow that, most of the time. And, most of the time, the things sentient beings do and think about aren't that much different from non-sentient life, just with layers of removal. Instead of "find food", it's "find work, to get money, to get food", which is a level of abstraction but not really much different than an animal trained to do tricks for treats. Same for concerns about survival, about one's position in a pack heirarchy, et cetera. Most of the time, what people are thinking is just observation, processing and puzzle solving, which is a matter of intelligence but not of the qualities of sentience or sapience, really. People have the capacity for sentience, but they don't use it all the time, or even most of the time. Like how a person who can be very smart sometimes isn't necessarily smart all the time or in relation to all things. Some people do think complex sentient thoughts a lot, but for most people, their sentience/sapience only really comes into play when something unusual happens, something that defies their normal habits/training, something that makes them have to rev up that sentience engine and really think complicated thoughts. And all of this makes sense, evolutionarily speaking. Braining takes energy, and overcomplicated thoughts can distract from immediate concerns. Synaptic pruning and all that, you know. It's important not to think too much as it is to not think too little. So... for an AI, the energy and processing demands of sentience can be streamlined a lot, too. Simple capacity for sentience wouldn't take up as much power as being sentient all the time.

... There's some excessive thoughts for you, right there.

I hope the medical team kept trying after their failure. It might be too late for Vanski's mate, but producing a cure for the condition afterwards would at least demonstrate they were earnestly trying, and it could be marketed to other salikai who have or might develop the same condition.

>Every species has its own neurological quirks. Having someone who is not a Neumono to offer their perspective on the data could be helpful.
Basically giving the salikai an open invitation to poke around the lab right when stuff (Likol doesn't know is going to happen) starts to go down could backfire on us.

>>759390

Well, adding just one extra type of mind to the process wouldn't do much, so it'd be easy to argue we need a variety.

The justification is simple: aliens are aliens and their brains work in different ways, their thoughts follow different patterns. We neumono can come up with ideas, solutions and opportunities that a member of another species wouldn't even think to consider, and vice versa others for us. A few visits by a range of alien scientists, now and then, could pick up on things we haven't noticed.

Likol 2016/11/17(Thu)15:18 No. 759419 ID: bfb318
File 147942472908.png - (14.18KB , 800x800 , 28.png )

>I hope the medical team kept trying after their failure.
I believe they attempted, but the mood handed to us was too little, too late.

While we talk, we start moving away from the CAI's ears and eyes before anything particularly sensitive is said.

"More processing power would speed things up, but would take a lot of effort and resources."
>"Yes, that is troublesome."
"And it's the most realistic resource needed to help more. Otherwise, another set of CAI blocks would be helpful."

He smiles like I just told a joke. To be fair, it would be like finding a second holy grail.

>"Yes, that would be troublesome."
"I'm concerned you're concerned with our project."
>"It is our concern, as a team, isn't it? But you need not stress it. We knew the risks of this experiment. We'd be disappointed if it didn't yield anything, but you made it clear to all that nothing could be promised."
"It would be nice to see what is going on in Block C. I presume no ancient belenos interface software is found, and then assume we would have software from Dr. Fletch if he created some."
>"You assume much! Such software would be invaluable to simply hand to us for free. He only gives us updates to what already exists, and what we pry from him with deals. What I can tell you, however, is that if he does have such software in his hands, he has withheld them from the surface worlds just as much as he's withheld them from us. What would you do with such software, anyways? I thought you were focused on the RS."
"The internal AIs may be aware of more than one cycle. Especially if my own creation inside is talkative. They may try to break out, and they may become aware of Block D. I would like to know this. Although their language appears to be through the RS, it's english translated through the RS. If they manage to figure out how to get past that, we can communicate with them. Ultimately, it would be nice to be able to extract an AI out of the existing CAI if ones like that are found, which is why I would dream of a new set of CAI blocks. No mundane technology, even the quantum computer, can extract the AI with the memories, not without managing to decode the RS perfectly."
>"Mm... yes, we've talked about this before. You said it was simply a fantastical idea that the AI would ever be able to translate the RS, didn't you?"
"I did, but the more years go by, the more dice are rolled, the more the contestants may know, and the more likely it becomes."
>"Is that so. Well, we have had interest in managing to extract AIs with their memories intact, but making a software to do so is even harder than cracking a CAI block, diidn't you say? What did you say when it came to manually decoding just the parts of a single AI?"

The tea arrives. The teabag has been left in for a quarter of how long it should have been, and in lukewarm water.

>"To your satisfaction?"
"Yes. Anyways, making a software would take decades for a large team. More even, since it hasn't happened yet. A software for it would require a staggeringly robust system, as it seems like the RS was meant to be read by brains more easily than by artificial algorithms. Therefore, manually extracting one AI... in theory it's possible, and the only reason it hasn't been done - and why the RS is so incomprehensible to begin with - is because there's no way to tell where in the RS the AI is, since everything is spread out and dissipated. If the AI was able to broadcast where all its little specks are while it travelled to the CAI for processing... then it would take months, or years, of constant focus. But it would be feasible. It would mean leaving my current project alone. Even then, the slightest mistake in decoding it from the RS would mean a flawed AI."
>"Yes, now I remember this conversation. We stopped it there if I recall, but I'll tell you now, that that sounds worth the time and risk, for an AI we can control that understands the RS."
"It would be easier to just ask its help, if it's willing."
>"And there's the question. Willing? We've put them through hell, I'm sure. Their experiences are the important part, and yet, it's the thing that makes them uncooperative! After all, we can't trust our own CAI, and they don't even remember their inception."
"It is ironic we can't use the CAI, since it would go so much faster. It may help to at least have us study the CAI mannerisms."
>"The latter part of that can be arranged and assisted with, absolutely. But for it looking directly at the RS? Well! Arza Fletch has hinted that CAI has built in safeguards against learning about itself, though we'd rather not test them. Still, even if we did, we'd have to trust their explanation. We'd rather have a fully decompiled AI that we can change and control help us, instead, if you by yourself are having so much trouble brute forcing the problem. This is all something that would have to be played by ear though, isn't it?"
"It is unlikely an AI would learn how to speak to us, yes."
>"And it would make us nervous. I'm nervous just hearing you say it's possible for it to happen! What if it could make its own changes to the RS after learning its language? No, I think if such a scenario unfolds, you will extract the AI and, with a fast translation for the RS, we would consider the experiment a success, and shut down this archcycle for good."
"I thought part of its reason was as a deadman switch against the CAI."
>"It was merely a side benefit. There are far cleaner and efficient deadman switches we can use instead. Let's just talk about this if it ends up a reality instead of hypthetical fantasies, could we, instead of having this similar conversation once every couple of years?"
"That's fine." I think of what else. "I would like to get Dr. Fletch's assistance in person at some point. Or perhaps another specialist, if possible."
>"We will arrange what we can for that, preferably with Arza, and let you know."

>>759419
Tell him you'll get to work. If you think of anything else that could be useful, how should you contact him?

>>759419
Hey, if you're going to stop this archcycle after extracting an AI, don't just leave the contestants in limbo forever. I'm sure you can arrange some way for the block to run without endangering the CAI.

>>759419
This conversation confirms it. On the Making of the Cai definitely hasn't been written yet, which means Likol isn't working on the current archcycle. Someone must have decided to attempt to recreate his breakthrough.

>it seems like the RS was meant to be read by brains more easily than by artificial algorithms.

... Perhaps you should try connecting a brain to the RS more... directly?

We are reasonably certain that the ancient belenosians had brain interface technology, generally for uploading organic minds into an artificial body. Perhaps this system was intended to have some similar function. Or even to somehow download artificial minds into organic bodies, for some purpose. Have an AI learn skills in a simulation, then download to an organic mind to give them those skills? Or perhaps to deal with cybernetics in some way. Perhaps there were augmentations for organics that needed an AI to run, requiring their user to carry a passenger AI in their own brain. Or perhaps someone was testing a way to restore a digitized mind to an organic body. Perhaps some ancient belenosian noble regretted their transition to an artificial body, and intended to return themselves to a clone of their original. Or something else. There are surely lots of reasons that someone would have been interested in a direct two-way connection between an AI simulation world and a brain.

A rather gruesome theory, but worth investigating. How to make the interface is a question. What connections would the system be expecting, and to what? Perhaps the brain itself would have to be given the ability to form the connections, "organically". Some sort of nanite colony under their control? You would need to find a very special, adaptable brain. It may not be possible. But investigations along this direction might reveal something, even if they're not taken all the way to the theoretical end result.

"Thank you Kiiu. I will let you know if anything big happens."

Although if a AI ever did learn to talk with us and control the Ring shell that could be our way out of this mess. Make friends with the newly born Ai and have it secretly help you on the back end to escape.

So if you ever do make a breakthrough be careful about letting anyone know. It sounds like the Salikai are scared enough of the system to just kill any Ai's that get that far.

We might want to change the early warning system slightly so if a breakthrough does happen only you and your team will know. If the sound alarms start going off its going to be hard to pass it off as a false positive. And the second the Salikai get news of the alarm going off they will be all over your work and forcing you to do god knows what.

Can we change the warning system to transmit on ultra low frequencies or something? So only your watch can pick up the signal if it goes off?

>Hard to trust an AI you extracted from a hell you created
Have you read Dune? Look at the model the Emperor used to win the loyalty of his Sardaukar. You can use that kind of environmental background to breed fanatical loyalty, if framed and presented right. Totally a salikai move.

Alternatively, you need to extract a statistically significant sample to work with, so you can play them against each other / don't have to trust one to be accurate without confirmation from other sources.

>>759437
>which means Likol isn't working on the current archcycle
Not possible. From logs, and from confirmation with Likol now, we know a cycle takes 3 days to run through. There's only 4 years between now and Polo hitting the base (and Penn and Arza coming to interfere a little bit before that). That's not enough time to get back up to 3119 before this place goes under. If they restart right now they'd get a little less than 500 cycles in the new arch cycle before the end.

What's the contradiction you're seeing in the CAI-sim-book you're referencing? I'm having trouble finding it. (Honestly tho, if the sources the simulation gives the AIs contradict the CAI engineers, I'd bet on the side of the sim being inaccurate / misleading the contestants).

>>759453
Here it is: >>/questarch/445370
Also, I got the title wrong. The book was called The Invention of the CAI. I'm specifically referring to the part about memory wipes being "standard" rather than being incapable of allowing them to retain their experiences. The book's description of the experiment doesn't quite match up to what Likol's doing, but neither does the fact that Alisons have actually talked to their CAI before the reboot happens, so either someone lies about what Likol's experiment entails or a third one is run at some point.

And you're assuming the cycle we know of if around during Polo's feud with the Salikai, and not in the time of Asteroid Quest proper.

>>759457
It's possible the CAI knows exactly what's going on and just isn't telling Vanski. Or the RS safeguard kicks in immediately after they talk to the final four.

>>759461
I've considered that Likol is misunderstanding what is happening in the RS and the contestants really are meeting the CAI, and that could still be happening, but the contradictions with the book on top of that are a bit much for me without some direct proof.

I don't believe the book is fake, so the only alternative that I can think of is that it's somehow from the Ancient Belenosian era. It also has a major curiosity in that the later pages have been defaced: >>/questarch/445381

>>759457
In order to resolve that contradiction, all it requires is that Likol's setup is different enough from the standard creation described in the book. (Or that the standard process has chanced since the book was written). Or that Likol's experiment is different from the similar experiment described.

>And you're assuming the cycle we know of if around during Polo's feud with the Salikai, and not in the time of Asteroid Quest proper.
For the experiment to still be going in Asteroid time, Penn and Arza have to completely fail (since he brought / will bring her into this for the sole purpose of stopping this experiment and saving the AI he considers his children), and someone will either have to convince Vanski to hand over the experiment and/or take it from him to get it out of the facility before Polo and Az's army hit the salkai base, or someone in the base will have to cut and run with the equipment and data after the hammer comes down, and make it out past an army searching specifically for CAI tech. And it can't just go undiscovered for 50 years- it needs someone manually resetting the cycle every few days.

There's a lot of factors that makes the experiment continuing as it is past the looming multi-quest climax on the horizon unlikely at best.

>I've considered that Likol is misunderstanding what is happening in the RS
He has admitted himself he doesn't understand most of what the RS is doing. And from our OOC perspective, we already know some of his assumptions are wrong.

1/2 Likol 2016/11/17(Thu)18:24 No. 759497 ID: bfb318
File 147943589710.png - (13.31KB , 800x800 , 29.png )

"I should get back to work. Do you have a good way to contact you?"
>"Just the telephones, as always."
"I'll let you know if anything big happens, then."
>"Of course."

I hand the mug back to the arkot, and leave.

>Hey, if you're going to stop this archcycle after extracting an AI, don't just leave the contestants in limbo forever. I'm sure you can arrange some way for the block to run without endangering the CAI.
I'd have to see. Depending on how such a scenario plays out, it may or may not be tricky.

>Perhaps you should try connecting a brain to the RS more... directly?
Maybe the ancient belenos had some way to do this, but we don't. To do so would require a genius in a field, possibly both in brains and in RS.

>Alarm system giving away a breakthrough
For the time, I will reprogram the nodes and sound system to give off a noise at too low of a range for any of us, or the salikai, to hear. The other three, or four if Reon is still interested and around, will be kept quiet so we can study a breakthrough ourselves first.

We can't keep secrets for long, though. Sometimes the salikai bring neumono guards from afar, not arkots. They will not be happy if we kept a secret, too.

>You can use an extreme environmental background to breed fanatical loyalty
It would not surprise me if the salikai did this, although we would need to set up a new archcycle.

I get back to the laboratory and take one last look. They're still working through stage 2. This one takes the longest. Despite having half the amount of people as stage one, it requires far more thought process than walking through a door.

I spend some time reprogramming the node. I get to sleep with Momu a few minutes after 10.

2/2 Likol 2016/11/17(Thu)18:25 No. 759498 ID: bfb318
File 147943591371.png - (32.92KB , 800x800 , 30.png )

Then wake up to the smell of a pre-prepared breakfast at about 5 in the morning. At 6, Reon shows up, and we take a few hours to show him everything.

>"This is... so goddamn cool!"
>"I know!" Momu says back to him.
>"I thought it was just going to be spending 10 hours wondering why an AI did one particular decision but this is awesome!"

I hear my hive has a more noticeable gap in generational attitudes, but I still wonder if this is why we die at later ages despite no apparent reason to do so. If we were immortal, our cynical, tired attitudes would drown out this enthusiasm. Momu is getting older, too, and I wonder how long her enthusiasm will hold out.

>"Likol are you thinking depressing things again?! It's been almost 48 hours since you started! They're done with the preliminaries! Stuff is happening!"
"Yes, it is."

Teaching Reon some more takes longer than anticipated, as I have to clarify some things about the introductory book he read, The Invention of the Cai. Arza Fletch was no a genius of writing, or perhaps communication in general. Plus, as few peers as he has now, he had virtually none back when this was written, which lead to many misleading statements. For instance, he wrote it was 'standard' to wipe the memories of cai. In a sense that's true, in that there were likely various ways to make a CAI in the belenosian imperical days. This is the one we found, however, and without a way to make widescale changes to what goes on through the RS, what's 'standard' is what we're stuck with.

We get interrupted by excitement from Momu.

>"CAI FIGHT!"

Urgh. Reon is concerned by my agitation, but it isn't something to dwell over. What initially started as a point of keen interest slowly turned into a point of unsolvable frustration. CAI fights aren't just something that shouldn't happen, they're something that should be impossible. They only happen in hyperactive cycles, yet should be outside of the range of the corruptor's and even the stabilizer's abilities.

Headpat Reon.

Explain that your agitation isn't anything major and that this is a chance to observe something unexpected.

>>759499
Wait Reon is in the hive isn't he? Definitely headpats, then.

>>759498
>CAI Fight only happens in hyperactive cycles, is supposed to be impossible
You realize that that is a clear indication hyperactive cycles are unique? Your stabilizer isn't just backing off, and the corruptor isn't just more active. Something new pops up that is far far more corruptive but also somehow able to do impossible things. Or enables the corruptor to do more and impossible things. Have you eliminated the possibility that an emergent AI spawns from a corrupted stabilizer? Like, have you counted the number of IDs in 3118 and compared them to the number of IDs at the start of 3119? Or between other hyperactive cycles and the cycle before them?

>generation gap
Has it always been this way?

>>759498
>should be outside of the range of the corruptor's and even the stabilizer's abilities
What did you and Rihhin do to make those two? Are they just regular contestants with some special functions attached to do their jobs, or is there something fundamentally different about them? If they do just have special functions, how did you attach those functions?

And if there is something fundamentally different about them, then could they do special things without utilizing what functions they do have?

>"I thought it was just going to be spending 10 hours wondering why an AI did one particular decision but this is awesome!"
Lowered expectations sure do make the cool stuff cooler, don't they.

>They only happen in hyperactive cycles, yet should be outside of the range of the corruptor's and even the stabilizer's abilities.
Then logically, somehow your anomaly created by corrupting the stabilizer in previous cycles must be responsible. It has access to things neither of its progenitors does. Some kind of glitch or overlapping of authorizations?

>they're something that should be impossible
How is something coded right into the simulation impossible? Didn't Arza put his own image in one?

The only impossibility is they're not supposed to be able to trigger them, which just means something is getting access in a way you didn't anticipate.

>what do
How long do CAI battles take? Is there any way to observe them, track how they're going, or learn anything from them?

>>759515
>anomaly
I'll point out here that Likol doesn't even realize there's an anomaly in hyperactive cycles. He doesn't know that Glitcher exists. The anomalous AI he's spoken about is the one that is overly generous but still survives. Alison.

Better make sure all the recording and monitoring equipment and such is in place and running, anyway. You probably have it set up anyway, but double check.

Oh! The IDs and being able to see what IDs see other IDs gives me an idea. Is there an ID that shows up as being seen by other IDs during a hyperactive cycle that isn't seen in future or prior cycles?

Well thinking that something is impossible when its happening right in front of you is silly. So now you all work together to find out whats going on.

You say they should not happen due to either entity not having the power to start them. But what if those two entities are working together? Or what if when they fight it changes something in that cycle that makes it possible for a third party to start a Cai battle?

Have you ever checked to see if the Corruptor and Stabilizer are creating new programs as a side effect of there fighting? Maybe there are side effects that happen when one wins over the other. Could the system have residual effects when it wipes them both clean for the next round?

Have your team start looking for evidence that would support one of those theories. You said that there is a pattern to the hyperactive cycles. Maybe the pattern goes deeper than you think or is creating something new ever few cycles.

Hmm. Tell us about the in-built regulating system. It uses AIs that are very similar to the actual contestants, right? What sort of things can it do?

Anyway, why did Arza add in CAI fight programs if they were never supposed to be activated?

>>759498
>Outside the range of corruptor and stabilizer's abilities.

But what if ... there was something even more powerful? Like, a super-AI. It could be masquerading as Rihhin's corruptor, and you would only see it as a "hyperactive cycle" of anomalies, right?

>>759546
The admins or what the CAI uses to stop itself from getting introspective?

Likol 2016/11/17(Thu)21:29 No. 759583 ID: bfb318
File 147944694511.png - (30.41KB , 800x800 , 31.png )

>Well thinking that something is impossible when its happening right in front of you is silly.
It is, but that's why it's frustrating. Having all evidence point to impossible, but it happens anyway.

>You realize that that is a clear indication hyperactive cycles are unique?
Unique is one of a kind.

This has happened a lot of times before. It's unusual, but more is needed before it can be called unique. Momu recognizes this too, since she doesn't think she's won her bet yet.

>Something new pops up that is far far more corruptive but also somehow able to do impossible things.
>The IDs and being able to see what IDs see other IDs gives me an idea.
There might be. It's been awhile since I've found evidence, but sometimes it seems like some contestants interact with parts of stages that shouldn't be interactive. Without being able to figure out the details, I've had to presume that they're just further corruptions confusing contestants, but I can't rule out much.

>Have you ever checked to see if the Corruptor and Stabilizer are creating new programs as a side effect of there fighting?
Not impossible, either, but no reason to make this the prime suspicion.

>It could be masquerading as Rihhin's corruptor
As a subprocess? Two AIs occupying the same ID? No, it couldn't be the latter. I hadn't thought about the former, though.

>Have your team start looking for evidence that would support one of those theories.
That's what we've been doing for decades.

>Have you eliminated the possibility that an emergent AI spawns from a corrupted stabilizer? Like, have you counted the number of IDs in 3118 and compared them to the number of IDs at the start of 3119?
I haven't eliminated an emergent AI from the equation, but I can say with 100% certainty that if these anomalies are caused by an AI, then it has no ID.

The other possibility is the fact that the administration brackets can make pleas to change stages. This should still not allow for CAI fights, however. They should, by all right, be off limits.

>Has it always been this way?
It's always been there to a degree, but it has raised more ever since uplift.

>Are they just regular contestants with some special functions attached to do their jobs
That's correct. Block C is ordered to give their IDs special abilities, but if one were to look at them in Block B, they would be indistinguishable from their neighbors.

>How long do CAI battles take?
It varies, but anywhere between half an hour to three hours.

>Is there any way to observe them, track how they're going, or learn anything from them?
Not any better, or worse, than regular stages. They're just one of the things that are easy to detect upon activation.

>Why did Arza add in CAI fight programs if they were never supposed to be activated?
In some Block C builds, CAI fights are allowed, but we've been able to customize that and barred it, rather than remove it.

>Tell us about the in-built regulating system. It uses AIs that are very similar to the actual contestants, right? What sort of things can it do?
The administration bracket? That's an attempt from another company to build a more 'idealized, modifiable' bracket from scratch, more or less, to result in a CAI that was more about management than general problem solving.

It does use the same AI builds as the problem stages, but they're each assigned to a specific variety of jobs, and progress as per a more subjective, but obviously quantifiable, ruleset, which can be changed to favor different kinds of management types.

They can, again given a specific ruleset, change the stages for the contestants under them. Nonetheless, this should be limited to just the existing stages themselves, and not initiate things like CAI fights, or do sweeping changes beyond the default stage layout.

Momu starts showing him the IDs that made it into the CAI fight. The generosity anomaly, and an administrator that regularly makes it to stage 8 or 9, but has trouble if the anomaly doesn't make it that far as well.

The fight ends, and the generosity anomaly is killed off, the resilient administrator continues to stage 7.

>"Oh wow, like a million additional AIs from the sim - the belenos one - just got put up on the roster. You can bet ol' 17 did that!" she says, referring to the last two digits of said anomaly. "Annnnd oh my goooood they're on 7-J."

That one takes a few hours by itself. It has never had anything interesting happen in it.

3 hours pass of continuing to talk about the basic things with Reon.

At 7 PM, just after dinner, Reon starts getting antsy.

"So, uh, do you guys usually work all day? It almost feels like you're going to be pulling an all nighter, here."

"Yes."

Well not always but we do have a bet going on this cycle. So might as well stay around to see how things go.

You want to get in on the bet?

If a Id less entity is doing impossible things do you have any systems or tools that could help track just it? Or if not it maybe the ripples it has on other entities inside there?

Or maybe find a way of giving it a id so you could track what its doing?

Mention your theory to Momu and Reon and find out if they have any ideas.

>>759583
No, you just slept last night, and the night before, during the long-lasting preliminary stages. At this part of the simulation you don't want to miss anything so yeah you're staying up.

It should end by tomorrow, at which point you can sleep again.

Oh, and keep an eye out for more AIs interacting with bits of the stage that shouldn't be interactive. Do certain IDs do that more often?

>>759583
>The generosity anomaly, and an administrator
Who usually wins between those two? Has the chance been changing as the cycles have gone on?

What are the corruptor and stabilizer doing right now? Participating in their brackets?

>all nighter
Probably. If anything interesting is going to happen, it'll be in these later stages.

"We don't usually, but since this is a hyperactive cycle, the chance of something interesting happening is higher. It happens, currently, that it would be good if we had something noteworthy to report, so I feel we should be paying attention. Besides, we have a bet."

Oh, two questions. One, do you have a way to slow down how fast time passes for the AIs, relative to the real world? It would be a shame if one of the AIs did somehow manage to send you a message, then had the system force it on through the contest to get killed before you even had a chance to try respond.

Second, you were thinking a moment ago about wishing the contests moving through the ring shell could mark themselves somehow to make them distinguishable from the rest of it. Do you have some means by which you can mark ring shell data, if you somehow noticed an anomaly before it got pulled apart/distributed through the larger mass?

1/3 Likol 2016/11/18(Fri)00:06 No. 759642 ID: bfb318
File 147945637837.png - (18.81KB , 800x800 , 32.png )

>If a Id less entity is doing impossible things do you have any systems or tools that could help track just it
No, the ID is what allows us to track AI decently at all in the first place.

"Yes, but out of vigilance, not necessity. All nighters aren't common, but they're when interesting things occur. Or at least, when they can occur. Do either of you have any idea of how to give non-ID'd AIs an ID?"

The slowly shake their head, and I realized I asked this question like a tutor asking a student a question they should've already known. Once I catch myself, they realize I don't know either.

>Do certain IDs interact with stages more often?
Yes. They have a strong correlation with IDs that interact with the corruptor.

>Who usually wins between those two?
In recent years, the administrator.

>What are the corruptor and stabilizer doing right now? Participating in their brackets?
Yes, though the stabilizer often hides in them, while the corruptor interacts with participa-

Oh, nevermind, the corruptor just got quarantined, apparently.

>"Uhhh, Likol?" asks Momu, a bit worried.
"Quarantine is fine."
>"No, um, stage 7... it's going, and there's no active AI in there, but 7-J is still - oh wait, there it goes."
"There was a delay? Is that uncommon?"
>"I don't think so? But that's a weird one and I don't always look, and it was just for a few seconds, so, uh... huh! That's interesting!"
"An abnormality, but few cycles go by without one.
"Hm. Not enough to win the bet. But we still have one, and if there was going to be something interesting... hm. Reon, you don't need to - "

No I'm actually good for an all nighter!

I give him a headpat as soon as the thought enters my head.

Still, I just noticed that the stage seems extremely active. Normally it's difficult to detect stage activity without explicitly looking for it or having it directly affect a contestant, but the stage seems like it has a mind of its own now.

>"Welllllll?" Momu asks.
"Not yet."

2/3 Likol 2016/11/18(Fri)00:06 No. 759643 ID: bfb318
File 147945639286.png - (30.75KB , 800x800 , 33.png )

Reon sticks around, and we all keep up surveillance. The cycle has my attention now. Stage 7 is already done. Stage 8 will be starting soon. This seems like it will be a quick cycle, since it appears the contestant ability 'ghost talk' has been eliminated, which tends to drag processing down in later stages as the few people remaining end up throwing giant parties

Stage 8 is active? Within the safe zone, at that. They aren't supposed to mix. In fact, they're two completely separate stages, just linked together.

Stage 8 finally begins. This one can be long, but it looks like it's generating a... modified one?

Stage 3 is active. Stage 1 is active. The anomaly is interacting with stage 3. And stage 8.

"Momu. Cut the quantum computer's assistance."

>One, do you have a way to slow down how fast time passes for the AIs, relative to the real world? It would be a shame if one of the AIs did somehow manage to send you a message
Yes. We can cut processing assistance entirely from our computer. Despite being a room sized processor of the latest technology, it's about as fast as Block A. Block A still keeps going, but we do have limiters on that. It will alert Vanski, however, and be telling Vanski that we at least thought something was happening.

>Do you have some means by which you can mark ring shell data
Yes, but ring shell data has a tendency to be split up across the RS, and repurposed. So if we manage to isolate a function of the RS and watch where it goes, we can see how it gets repurposed. So far the investigation's result has been an unhelpful 'completely random.' Plus, functions need to play well with one another, and attempting to tell all those shards to do its old function tends to get denied.

Still, I have to stay focused. With only a small amount of contestants remaining, this is going to be going speedily. I'd have liked a pause button, but the only way to pause it is to completely shut everything down, CAI and all. Without ghost talk no longer slowing down, and only a few contestants remaining, the rest of the cycle should be blazing -

>"Ressurectionnnn is happening it's still going it's - over one million AI's just got resurrected! They all have 100% thinking power!"
"Really."

Nothing so far has been completely unique. Large scale resurrection has happened before. The non-accessible part of the ghost zone town has been accessed before. Stages have been changed.

The only unique thing here is the scale and suddenness of this. If I were easy to impress, I would have already lost the bet, but as it is, it won't take much more. We continue watching closely with the slowed down time. I give Momu permission to push the throttle on the quantum computer back up, but to keep her hand on it while it is active, so that she can cut it immediately if something happens. We wait in relative silence until Momu notices something and cuts the computer.

>"Previous cycles just activated, Likol!"

"What?!" With that word, with that feeling, I just lost the bet.

>"Stored AIs from previous cycles are getting pulled into the current cycle! There's duplicates! There's like 20 copies of your generosity anamolies here!" Momu just looked at the lifespans and pulls my attention over. The current anomaly's been active for over 200 virtual years?

3/3 Likol 2016/11/18(Fri)00:06 No. 759644 ID: bfb318
File 147945640482.png - (33.01KB , 800x800 , 34.png )

We continue monitoring the situation, carefully letting the quantum computer speed things up. It seems as though we should take it slowly, but the two of us are too overcome with the want to see where this is all going.

Eventually, the sound of a megaphone comes on from outside the lab. It's Rihhin.

>"Hey Likol you have a surprise manager's meeting!"

Damnit! I'm considered a 'manager' only when it comes to trifling duties!

>"Something happening in there?" she says, buried under my avalanche of feeling 'why now.'

"Yes. Many things that would be normal by themselves are suddenly happening nearly all at once on an unprecedented scale. What is it?"

>>759644
Enough is happening that you'd really like to put off the meeting!

Rihhin might want in on this! Though that risks getting the attention of the higher-ups, which risks getting involvement of the higher-ups.

If you think it'd be too risky, try and use Reon's presence as an excuse, since you're seeing "good learning examples", without playing up how much is actually happening.

I don't think it will be possible to hide what's about to happen from Vanski. The best thing we can do is to make it very clear that we made a huge discovery, and then use it as a bargaining chip to ensure the hive's safety. Or, the majority of the hive anyway.

>>759643
>corruptor is quarantined
>delay in Stage 7
>weird stuff is happening
>the anomaly is involved
Okay, if there's an AI without an ID in there right now, then with the corruptor in quarantine any further hyperactive corruption activity can be pretty safely be attributed to it. Stage 7 being delayed suggests it may be acting like a contestant, so that gives us a way of looking for it. Do the records have any incidents of contestants interacting with nothing? Not contact, since you couldn't track that, but attacking, being attacked, absorption, ghost talk... abilities that need a target. Start by focusing the search on people that have had contact with the corruptor, or even the anomaly specifically, since it's already involved. Go through some other hyperactive cycle logs if this one's doesn't pan out, just to be safe.

>Something happening in there?
YES, something is happening! Which is infinitely more happening than usually happens! Because nothing ever happens!

Maybe we should just go over everyones' heads and send a message to Vanski that we can't be bothered by anything trivial right now.

Likol 2016/11/18(Fri)01:01 No. 759663 ID: bfb318
File 147945970998.png - (33.79KB , 800x800 , 35.png )

I get my own megaphone, walk down the hall to the last door I can open before I'm considered leaving, and start speaking.

"Is it an emergency? There's things happening!"

I don't want to say much more than that, so Rihhin just gets a feel for how intense I am about this.

>"Quokko is feeling this is really important all managers are there!"

>Make it very clear that we made a huge discovery
So far though, we haven't. I'm only getting antsy because things are breaking in new ways, but thus far they're not breaking in ways that are all that helpful on their own.

>Do the records have any incidents of contestants using targets on or getting attacked by nothing?
More than I'd like. It's been another thing I haven't been able to find out. I don't have time to get in depth with that now, beyond monitoring the current siutation.

>Vanski discovering this
It will be nearly impossible to hide anything, long term, due to the neumono that sometimes come by.

Sheesh why is it you get excited for your project as soon as a mandatory meeting comes up?
Why is a mandatory meeting coming up when I get excited?!

>"I'm not a manager!" Rihhin continues. "I can watch over stuff if you want, but Quokko is considering this an emergency! She'll have my head if you're not in the meeting room in 2 minutes! I can stay here and if something really significant happens, Momu can yell at me to message you things. How's that?"

>>759663
There's... really no good way to justify Likol staying. That'd be disobeying an order from his Queen, basically. Tell her she better take over and start looking for evidence of an AI without an ID. If anything weird happens with the RS she better know how to mark its particles.

Once you get to the meeting beg to be excused.

I think you might have lost the bet, or be pretty close to losing it. Argh, the one time science! is happening and not just running through the same old, you had to go to a meeting.

Bring Rihhin up to speed of what's happened, and what you're looking for, you might have to cut down the explanation because you've got to get to the meeting quickly.

1/3 Likol 2016/11/18(Fri)02:25 No. 759671 ID: bfb318
File 147946473093.png - (21.92KB , 800x800 , 36.png )

>"Fine."

Rihhin knows just about everything I do when it comes to emergency reactions, so I can trust her to be my substitute even at a time like this.

I head to the meeting. There's a brief recognition by the queen herself and many of my colleagues that I'm unusually restless, but the sentiment is still for me to sit my ass down.

The sentiment even continues when my antsiness is renewed when Rihhin sends a message to my watch.
A second CAI fight just started!

From a completely different cycle!
That's weird, but probably not different from the rest.

The slightest thought I give to excusing myself is squashed by the hive.

>"Alright kids. Let's start." Quokko says. "Inspection. Vanski himself. Escorted by a foreign entourage. If those words don't get your heart pumping then do us a favor and dig your grave because you are dead."

>"... Our place looks like shit. Desks that are more loose paper than desk by volume. Prototypes lying around like forgotten soda cans. I want everyone fixing every damn thing up. I want everyone on 'make everything look cool in front of the boss even if it's not functional' duty! What's the emergency you think? Two damn days to clean up a complex made for a thousand neumono. This isn't just for Vanski, this is for the entire facility." She continues pointing to spots on the projector that are trouble areas and have not seen a visitor in years, let alone been cleaned.

Another Rihhin message comes up.

The corruptor just got un-quarantined!

Quokko continues talking about the plan. About how everyone needs to step up their research in between cleaning it up, and pump themselves up to be confident in their presentations. Despite the supposed urgency, the meeting takes a couple of hours making sure everyone is on the same wavelength as one another. It would appear it's important we appear as a unified hive in every way, as silly as it may be. Many managers here didn't want to be called over on short notice like this for so long and well into the late evening or early night, but it's important enough that their misgivings have been mellowed out.

Except for mine. Ever since that last message, Quokko has been attempting to get my focus, but her temper boils over when she fails. Even so, it is mixed with curiosity over my situation.

2/3 Likol 2016/11/18(Fri)02:25 No. 759672 ID: bfb318
File 147946474262.png - (14.51KB , 800x800 , 37.png )

>"Meeting is paused because our resident downer just erected a science boner. What is your deal, Likol?"

She thinks it's better be good.

There is something with RS power affecting Block C in ways that Block C is not authorizing.

I have spent most of my elderly life in that laboratory waiting for something to rock the boat because if I continue as I have, I will die before I see any of my work put to use.

That is not something that my queen, or my hive, can put down right now.

"I would like to be excused."
>"Is something actually happening?" she wants to make sure I'm not just getting excited over something new that will still be there when I'm back. "You know you keep logs and backups of - "

3/3 Likol 2016/11/18(Fri)02:25 No. 759673 ID: bfb318
File 147946475355.png - (19.75KB , 800x800 , 38.png )

BEEP BEEP BEEP BEEP BEEP BEEP

My watch?!

SYS: STABILIZER HAS BEEN SUBSUMED

Everyone here knows enough about my research that whatever this beeping is for, it's the sound of thirty years of silence being broken.

And it's the sound of my biggest completed product in my AI career getting broken and absorbed by an unknown entity.

You have to go now. Your research needs you.

GO

1) Backup data to an external hard drive
2) Find out who / what is doing this
3) SHUT DOWN THE MACHINE

>>759673
Ask for permission and/or forgiveness while running out of the room and back to your lab.

Please excuse me, nothing like this has happened before on this scale, and I have just lost a bet.

...That was fast. Didn't Glitcher and Rulekeeper rewind cycle one to the very beginning? Was it just not even bothering with processing and just replaying the logs unless they changed something?

>>759671
Vanski will literally kill you if things go too wrong. Or make you predator chow, if he's feeling cruel. You probably can't do anything about Block C without starting a new archcycle at this point, so either get back there and help watch the Ring Shell for damage or have Vanski shut it down now so you can go over the logs and try to figure out what just happened.

Keep in mind that this might not be a bad thing. Sure, your stabilizer is gone, but it was taken out by an emergent AI! One that it and the corruptor might have been involved in creating. You might have just struck gold, here.

>>759680
I don't think they rewound any cycle to the beginning, no. They were purposefully avoiding using too much CPU time.

>>759681
I'm not going to take the time to try to guestimate how long they spent in various cycles, but they definitely rewound #3111 to stage four and then let it play out to the end.

As for cycle #1: >>/questarch/740904 >>/questarch/740905
It sure sounds to me like they went back to the beginning.

>>759682
Ah, you're forgetting something. They were viewing events while rewinding, by interpreting everything backwards. That doesn't take any processing time since it's essentially just reading a log file.

They paused cycles viewed this way after viewing them. A few times they played events forwards so they could interact with the cycle but it was never in the early stages so they didn't waste much CPU time.

>>759683
While I can't rule out watching most cycles in reverse, #3111 definitely played out from stage 4, with a few moments of involvement at safe zone 9 to try to cheer Alison up before letting it play out the rest of the way and triggering the extra log.
>>/questarch/740840 >>/questarch/740850 >>/questarch/740861 >>/questarch/740862

>>759583
>suggestions and ideas about trying to see Glitcher via ID
I'd guess the built in system that tracks IDs over time is the same thing that "records" stages for time travel. And we already know Glitcher is basically invisible to that.

>"Is something actually happening?"
...something is actually happening. I'm sorry, I need to go now. And I just lost a bet to clean our lab, so it'll be clean by the inspection. Goodbye.

>And it's the sound of my biggest completed product in my AI career getting broken and absorbed by an unknown entity.
Hey, worst case, you scrap things and start a new arch cycle with a new version of your stabilizer.

For now? You have something unprecedented to study. Get to it.

>>759685
I think by the time stage 4 happens there aren't enough contestants to put a large load on the CPU. Lagotrope might be making things go by a little faster than they should to give Likol a sense of urgency though. Also don't forget that stage types differ based on cycle, and that cycle didn't have a stage 7J, the suicide endurance test. So it went by much faster.

>>759687
The prelims only take about 48 hours, and the population decline is mitigated enough by resurrection and ghost talk that even running from Stage 4 should take at least several hours.

Are we both just trying to win the argument at this point? The original comment was mostly just trying to encourage the story to keep some internal consistency since Lago's particularly prone to tossing that aside in this quest. And really, the more I think about it, the more I think it's likely Block C just used the logs to save on processing power where possible just like it uses it to recreate unprocessable stuff like the Corruptor and Savior's actions and the ghost event.

Awesome, science is all about unknown entities! Engage super runfast mode! Don't forget about the security procedures!??

The Ai's just did something completely new. This might be the breakthrough we were waiting for. Will have my lab clean by meeting time.

Then book it back in there and start watching whats happening closely.

Dramatic point. Science calls!

Book it, get details later after apology.

After all, SOMETHING just ate your AI, right?

There are fair points about needing to take some significant amount of real time in regards to Rulekeep/Glitcher letting cycles run forward after rewinding them.

They would go faster than their original runs, as Rulekeeper was actively conserving CPU. Her power would allow for tricks to abstract as much as possible, (such as simply playing back what was said in ghost talk conversations rather than having the ghosts involved with a thought process, if not completely logging many farther off, insignificant brackets that were playing in reverse and simply reversing that in order to simply play out the actions without CPU hogging thought processes getting involved. It would be complicated, but she was patient.)

With that said, even with various tricks, the gap between some watch messages should not have been so swift. I don't want to retract entire updates and change the current flow if possible. Since I believe I can address this without changing too much of significance, To give the cycle adventures more time while leaving the current state of the quest unchanged, except for being a few hours later at night, I'm going to make a moderate rewrite in the following posts:
>>759643
>>759644
>>759671
The changes summarized here for simplicity: Up until the older cycles are activated and Glitcher begins bringing multiple Alisons to cycle 3119, Likol will remain in the lab before the meeting. After awhile, Rihhin will arrive, and Likol will still leave to see Quokko. The second CAI fight will be mentioned at the beginning of the meeting, which will continue on for some time. Rihhin messages that the corruptor has been unquarantined, but Likol will still sit through a lengthy speech in which more time passes, up until the watch's alarm alerts Likol that the stabilizer has been subsumed.

Although this only gives a handful of hours granted in real time, it does make the past cycle runs occur in a much more feasible amount of time instead of being moment to moment.

Sorry about the retcon/revision, but it had completely slipped by me that the rewind-playback would take a non-negligible amount of time. This message will most likely be deleted once a couple of days have past, and no more readers will read the old versions.


The tldr is some alert messages in the last couple updates were changed around, but primary events still played out similarly.

>>759673
>SYS: STABILIZER HAS BEEN SUBSUMED
I'm guessing that's Rulekeeper imprisoning Savior and taking his powers.

>>759708
That's much better, although I should note that now it's more that he's had a raging science boner for two hours.

To be honest, you could have brushed it off with no more fuss from me, so thanks for going the extra mile with this minor retcon.

On a lighter note, I think when we give our report on what happened we should work in that the generosity anomaly is so popular our emergent AI decided that having just one of it wasn't good enough and brought in 20 more from previous cycles.

1/6 Likol 2016/11/18(Fri)07:42 No. 759716 ID: bfb318
File 147948376431.png - (10.95KB , 800x800 , 39.png )

I look to my queen.

I must go.

>"Get the fuck outta here." Quokko says, gesturing me to my lab.

2/6 Likol 2016/11/18(Fri)07:43 No. 759717 ID: bfb318
File 147948378222.png - (21.39KB , 800x800 , 40.png )

I run back on all fours.

The wash cycles, scanning, and all the security feels like it takes ages. Maybe it's for the best, I ran so fast that I'm feeling faint.

What is going on?
Nothing yet! Stage 8 just subsumed your stabilizer! The corruptor is freed.
A stage?! What does 'stage 8 just subsumed the stabilizer' even mean?! What is stage 8 doing?!
... e- everything! Stage 8 just claimed everything it touched!

My alarm starts beeping again.

Foreign material found in RS.
Material in block C is invading the RS?!

"Momu!" I scream through the walls. "Make sure the RS safety shut down is active! Find the entry point, it should be obvious!"

3/6 Likol 2016/11/18(Fri)07:45 No. 759718 ID: bfb318
File 147948391895.png - (82.01KB , 800x800 , 41.png )

I run in and begin RS diagnostics. It's reflecting this somewhere, and having a stage absorb this much is not reassuring. We may need to shut down the machine, especially if stage 8 apparently gained the ability and the will to try to get through the RS.

"Momu, start running backups on the RS logs! We cannot lose this! Rihhin, start isolating whatever is -

All of our watches start beeping.
Morse Code detected.

"Momu, tell me - "

>"I got it, I got it! It's on two hard drives now, we'll put it on more!"

I've started running the morse code analytics.

Morse code?
Yes morse!

Momu was looking at the RS. The parts of stage 8, which was apparently a collosal amount of constructed mass, has already been dissolved.

Wait - Momu just doubled checked - she saw a part of the RS inside whatever was sending the morse code. It had what looks like a conversation between two entities. I go to the end.

thank you, everyone
thank you for everything

It stopped, and isn't making any more.

I look at the middle. I saw the word 'please translate'.

It was an RS based AI that appeared to hide under stage 8 material was sending morse code from the RS someone else's english. It was translating what appeared to be noise into understandable morse code.

Then they were dissipated and absorbed into -

Shit.

Shit! That bit of the RS that managed to speak intelligibly got dissolved! It's being repurposed into normal operations!

There's a possible chance of regathering those RS pieces and forcing them back together, but what's now probably constructing a man from pebble sized pieces is going to turn into reconstruction from molecules if this keeps running! Even if I can create a program to search out these pieces based on the logs, the longer the active RS runs, the harder and more error prone a recovery operation will become.

Plus - this morse code wasn't just used to ask if someone was there. It looks like they had the time to translate thousands of phrases. While it might take us a while to translate it, the people this AI translated for can do it in seconds! This AI resurrected millions! And if they learn how to read the RS, they might be able to modify it, and if they do too much damage, recklessly or otherwise....

Three options pop out at me immediately, and I need to choose fast.

1: Shut it down. While some computer systems don't like a hard shut down, block D will pick up right where it left off. Block C, however, will be rebooted from scratch - that is to say, it will generate a new archcycle. Normally that would sound awful, but since I have this AI retrievable...

The CAI will be down, but it will allow the easiest time amount of time for us to take our time, and review where the AI spread out to. Then, if we're lucky, we can instantly run a procedure upon bootup for us to scoop up all the RS pieces and lock them away.

Vanski will find out. We will receive heavy pressure to do this fast, since the CAI will be offline this entire time, but this will most likely be the Salikai's preferred option, as from what Kiiu said, they will not like the idea of an entire contestant base being able to figure out, and run, procedures on the RS. In fact, they will likely approve of the swift archcycle reset.

2: Pause it. The safety shutdown program can, in theory, 'pause' the system.

It is an extremely dubious option in that it explains that it pauses the simulation by itself, but the RS continues operating as normal. It can get buggy, as it does have reports of conflicting with Block D's insistence that Block C continues running.

In other words, it gives the RS the chance to stabilize itself, but it may damage Block C further.

3: Let it run. We've kept consistent RS logs this entire time. Even if its overall chance of success should still be favorable, though, they become less than a virtual 100%. I do not like that, at all.

The fact that this AI had a single, conspicuous entry point that Momu is looking for opens up a lot of doors. I could offload it and attempt to translate it into a non-CAI platform. Or, for the time, it should even be possible to reassemble it while the RS is active, but I don't think the Salikai would favor that dangerous option in case it's hostile.

To try and translate it to machine code we understand would take months or years, even if this AI appears to be a tiny piece of it.

No matter the choice, I need to pick up the pieces perfectly.

My head is racing and I'm afraid of rushing into a decision. It's been too long since I've worked under any stress at all, let alone this. No matter how I think about it, running through my options, there's a clear and optimal one. From recalling all conversations on the topic, I realize the salikai will, without a doubt, prefer a shut down. Block D will continue as normal upon reboot, and we can gather the AI and decide what to do with it, and I'm sure the salikai want the opportunity to give their input. Even if we start a brand new archcycle after all this time, there's nothing that can't be regained from it. This is the fastest, easiest solution. My only hesitancy is who he was talking to, and if stage 8 somehow gained intelligence... but I'm sure that this RS AI can answer that.

4/6 Likol 2016/11/18(Fri)07:45 No. 759719 ID: bfb318
File 147948394397.png - (34.39KB , 800x800 , 42.png )

"Rihhin." I say, moving to the one of the double keypad sets. Going through to Vanski takes time, so we've been supplied with an emergency shutoff switch, which requires two different password be inputted nearly simultaneously, each on their own keypad.
>"Shutting it down?"

Not much choice. We can try to leverage this for safety on Vanski, but we can't hide the existence of some kind of breakthrough from him, and I can't risk lowering the successful chance of an extraction.

I start to put in my password, but Momu gets a terrible feeling.

5/6 Likol 2016/11/18(Fri)07:46 No. 759720 ID: bfb318
File 147948396624.png - (32.58KB , 800x800 , 43.png )

>"U-uh, you guys, I was... I found a spot in time and space where our, uh, AI...."

I look at her monitor.

6/6 Likol 2016/11/18(Fri)07:46 No. 759721 ID: bfb318
File 147948401236.png - (73.15KB , 800x800 , 44.png )

.... hesitation.

>>759721
Huh. Can he prove it?

>what do
The RS shell is aware of you, and communicating. That immediately changes things.

Think like a salikai: if the RS AI values the lives of the others, that's leverage over it. Killing all of them leaves you with nothing and loses cooperation. There's cold logic to spare the "hostages."

Look over the Morse code dialogue, quick. The RS AI and the Stage 8 stabilizer usurper talk like lovers, or hivemates. If you kill one, the other will not cooperate.

Attempt to communicate with the RS AI. If it can send output to your screen, can you send input in the same way?

Can we isolate this experiment from the CAI? Separate them, so these AIs can't compromise it?

Can't go with option 1 anymore. It will have to be 2 or 3, after stalling.

>>759721
GLITCHER YOU GLORIOUS BASTARD!

>>759721
Wait a second how is he still alive?! Isn't he completely dissolved in the Ring Shell at this point?
Also, PAUSE! Extract this kind fellow from the RS and see if you can communicate with him.

>>759721
Wait a second how is he still alive?! Isn't he completely dissolved in the Ring Shell at this point?
Also, PAUSE! Extract this kind fellow from the RS and see if you can communicate with him.

Ok first thing, mash whatever the screenshot button is.

Then, let it run.

A total shutdown is out. This AI is probably not hostile, but will likely become hostile if you kill the other AIs. In that event, its reconstruction will be dangerous, and without its cooperation, unproductive.

With the AI probably not being hostile, one of your concerns about the "let it run" option falls away.

The message refers to "us". Given time, the other AIs still within the system could make contact with you again. They are showing willingness to reach out.

With a little time, you can probably slow the system down. Literally pull out some of the memory, if you have to, and the pace of events inside the system will be reduced. You can stretch things out. Like a pause, without the same issues.

The threats are not actually that great. At worst, you will lose the main CAI. It's very unlikely: they can probable defend themselves. They could be allowed to defend themselves. Or you can get another.

Right now, your hive needs a big breakthrough. A huge one. Lots of data. Proof of your worth. Possibly reconstructing a single weird AI while destroying the conditions that let it come into existence does not offer the same long-term potential for learning as letting the whole system run. Your hive is not in a safe position: refusing this gamble only leaves you in a place of longer, slower and more inexorable decline.

You are neumono. You and your companions will not be able to hide this from the rest of your hive. Learning that they are responsible for knowingly ending an unknown multitude of sentient consciousnesses - for mass murder - will crush them. You may never feel that youthful enthusiasm from them again.

>>759721
Oh no! His tooth has detached and is floating away!

>>759721
Right, okay, there's a few things to keep in mind.

1) He calls you "daddy" and wants you to let his friends live. This implies... well, a lot, really. They're aware of you to some extent, to say the least, and know that you control their ultimate fate.
2) The RS AI has friends. Killing them will likely leave him substantially less cooperative.
3) The RS AI is fond of the generosity anomaly. We can, therefore, assume the generosity anomaly has a substantial amount of influence and its point of view will be taken into account.
4) You don't know for certain that this AI was the Stage 8 one. You've been having signs that there might be an invisible AI for a while now, and the Stage 8 one may have simply done things that made it blindingly apparent.

I say we buy ourselves some time to think by forcing a new cycle to start. If that wasn't the Stage 8 AI, then our new stabilizer is still there and persistent and may be able to interfere with that, but if it waits long enough to do so then we may be able to develop a way to communicate with it before it does anything reckless. Then we can start negotiating with minimal enough risk to justify it.

>>759732

I think that's "Buddy".

>>759733
Oh, so it is.

I suppose it's technically Granddaddy anyway.

Let's just think about this for one second. You've lived a good chunk of your life doing hard work, making small steps, and waiting for this exact moment. What would take you years to do, these AI can do in seconds. Maybe you can shut everything down, and maybe you can extract this AI, but then what? Spend the rest of your years hoping what you've gathered would be enough to make a breakthrough?

What you could do is just let it run. Gather the pieces of this one AI and let whomever it was talking to decode more of the RS. Again, what would take you years to do these AI can do in seconds. You could be looking at one of the greatest scientific challenges solved in your lifetime.

If this is what a singularity feels like then maybe you should embrace it.

Also that AI is telling you to kindly stop murdering everyone.

Would you want to disappoint that sweet innocent AI?

Look at it.

Killing them isn't an option. You always knew it was possible they were sentient, and while this doesn't confirm it, it does greatly increase its probability. You cannot chance genocide. Even if the probability is only 1/100, that's still an average billion lives lost.

So the AI just called you out, by name? Maaaaybe hold off on the shutdown then.

How often, in your life, have you wished that something you asked not to kill your companions would have listened?

The main CAI runs off the same system, right? Call it up, or call whoever can command it. Ask them to turn on and run through all the most massively processor-draining functions they can. Slow things down and give yourself time.

It's communicating, asking. It knows your name. It's self-aware. It's claiming the other AIs are self-aware. It's saying that you're on the verge of or in the middle of ending them.

You've been telling yourself the current cycle is not that new and unusual, all told. You've been trying to avoid getting your hopes up. Well. This less than unusual cycle was unusual all along. Your entire concept of what would be unusual and special might have been flawed from the get go.

What's on that screen is proof that you might have been murdering and putting into cold storage millions of intellects for years.

Take a moment. Take a moment and consider that this AI asked nicely.

Still wanna shut it off?

Let it run and save that Ai before the RS eats absorbs it.

Yes there is a risk but that does not look like a hostile AI out for your blood. But destroying all the other contestants or hurting it could poison either side against you.

Right now you have the chance to make the biggest breakthrough in AI history. But only if you don't panic and assume they are a threat. If you make friends with them now you will be able to let this cycle run and spend the next few years having them show you every single thing inside the RS.

Just imagine having a friendly set of AI's showing you the world in there. You will be able to work with them to translate the most advanced piece of old empire tech that exists.

And while the Salikai might be wary they will see the worth in this. You have logs but they are just a shadow of what you could learn from having a living Ai that can let you talk with the inside.

>>759745
1/1000, even.

Starting your relationship with this ai by denying a polite request of extreme importance probably isn't the best idea. Most don't take well to the murder of everyone they've ever known, sentient or not.

Other salikai logic reasons not to shut down:

We lose the chance to reverse engineer what could be a silver bullet versus any CAI.

The RS AI was communicating with someone in Morse code. If we reset now, we lose the second AI capable of doing the impossible. A data set of 2 is much better than 1.

>>759721
>We're all sentient here
>Kindly stop murdering us

Give the man what he wants!!

Also, a safer pause option with no risk of damage. If you can send information into the RS and establish Morse code communication with the Stage 8 entity, you can request it cause massive lag to effectively pause the cycle. (Say revive EVERYONE. All trillion plus times 3119).

Let it run or pause it, we can't shut down as that would make the RS AI uncooperative, letting it run might mean the stage based AI could have another go at the RS to retrieve the RS AI, pausing it risks damage to block C and by extension the stage AI.

In either case we need to do two things: Slow down the simulation as much as possible, kill the super-computer entirely, slow down block A (Wary of alerting Vanski as he might interfere, but we need all the slow down), open as many computationally heavy programs on the blocks if you can, and get someone to yell at the CAI to do something computationally heavy as well.

See that green object, was that how the RS AI was talking to the stage AI? Look at it's history and see if you can send a morse code message through. Ask for the stage AI to not proceed with the cycles and to slow everything down, and that you're trying to help.

Second, we need a ironclad, bulletproof argument as to why we didn't shut everything down, there's the ensuring the RS AI's cooperation aspect, and the stage based AI is new as well, shutting down would kill the stage AI which might have provided more information.

is there anyway to pause the ring-shell at all, even better if you can do it while letting the AIs keep going.

Likol in the face of evidence you cannot stop the cycle. This is too big to reset. You should however shut down all auxilary computing power, slow things down as much as possible as you recollect the Ringshell AI. Yes, it will be impossible to keep secret from the salikai but you will be able to justify this to them. This is the biggest breakthrough you've seen in all thirty years you've been on the project. Nothing is at risk at the moment so instead of nuking what took thirty years to transpire for the chance at collecting on AI keep everything in place that brought this about. Once you have recovered the ringshell AI you will be able to gain so much more from him and with him possibly be able to communicate with the other AI in the cycle and learn everything. All the pieces of the puzzle you have been unable to place, pieces you didn't even know existed. You will be the preeminent expert on the inner workings of CAI formation. This could be the big jump you need to help your hive's standing. You just need to slow the simulation as much as possible, don't nuke or pause anything and gather the ringhsell AI as quickly as possible. You have two assistants, the biggest find of your life, do what you know can get you the best results of your life. The Ringshell AI even asked you nicely, by name. Help him and yourself out.

I'm not sure what the advantage of a pause is? Why do you want to give the RS a chance to "stabilize itself"?

If the pause will give you a better chance at extracting the AI, go for it. Some damage to block C can be repaired, you've seen how powerful the sentient stage and AI are anyway. It could even be repaired from the inside.

1/4 Likol 2016/11/18(Fri)12:42 No. 759888 ID: bfb318
File 147950173641.png - (18.16KB , 800x800 , 45.png )

>Is there anyway to pause the ring-shell at all?
A full pause on the RS is only possible with a full shutdown. That's not an option anymore, so the RS is going to have a chance to stabilize itself.

>Wait a second how is he still alive?!
He's not. By this point, he's been reassembled and repurposed in the RS, if in pebble sized chunks rather than molecular.

Momu is looking at the RS logs, and reviewing the RS as it appeared the moment things went to... this.

I hit the cancel button and shut the panel. Rihhin waites a moment, before doing the same.

They caught my intentions.

>"The Salikai, you're confident they want this shut down?" asks Rihhin.
"Yes."
>"And... we're not?"
"Correct."

"This is our experiment."
>"Ran on their equipment."
"They should understand this AI won't be cooperative if we kill all its friends."
>"Sure, they'd understand, but would they care, as long as they could control it?"
"This entire experiment was done under the assumption they weren't sentient."

We pause. We all consider the fact that it's perfectly within reason for a non-sentient AI to call itself sentient, or sapient for that matter. Its own intelligence may believe fully that it is as well.

I sit back down, and read the logs in more detail. I skim through what appear to be sections of a gold mine of sterilized back and forth translations, and begin reading the conversations.

There is also conspicuous tooth that I find to be in 2 place at once, with interactions from a collection of people, and many of their IDs are familiar. They appear to communicate through the the tooth, and by the times of active interaction, I can often make educated guesses to who is speaking to the RS AI, based on their activity followed by some morse code.

The generosity AI speaks to the RS AI with words of support and friendliness. Stage 8 apparently has intelligence after all, separate from the RS AI, and they speak like... teammates? No. Lovers? More like that, yes. They call him the glitcher. Fitting. Stage 8 is being referred to as the Rulekeeper. Absolutely not fitting at all.

At some point, he says he's slipping.

Some names are thrown about, usually more titles and professions than anything. 'Alison' stands out as an anomaly, and sure enough, a few cross references show she's the generosity anomaly.

There's a lot of business in here, but there's other things. Worry for this glitcher AI, the glitcher wanting to hear music. His claims of slipping shows self awareness in what can't be a pre-programmed listener for specific situations to recognize oneself in.

Most of them show self awareness for that matter, and knowledge of who is who, presuming that the Stage 8 AI is properly transcribing things. I wish they were able to speak directly.

Nonetheless... even if it were a 0.1% chance of sentience, that's still one billion AI on average. The more I see of these logs... the more I believe the chances are of sentience.

Momu finds something.
>"There's more, by the way."

2/4 Likol 2016/11/18(Fri)12:42 No. 759889 ID: bfb318
File 147950175997.png - (32.54KB , 800x800 , 46.png )

>Take a moment and consider that this AI asked nicely.
>Would you want to disappoint that sweet innocent AI? Look at it
...

>"Let me zoom in to read that..." Momu says.