Novels2Search
Simulacrum: Heaven's Key
Chapter 7b: Spark Of Success

Chapter 7b: Spark Of Success

(Helix Studio, Regent Suite, Bedroom)

[https://i.imgur.com/AurrjCU.png]

It took 3 months, but while I was immersed in programming, I got a notification that somebody finally cleared the ninth level! I quickly set the training run so it ceases once all instances in the population have finished and take a look at the winner. Sure enough, one of my instances managed to do it.

I bring up its program and begin processing it. Programs derived through genetic programming are always bloated, so I pass the AST through a partial evaluator to get rid of the dead code and simplify it, then I print it out in readable form. There is a lot of new functionality in the behavioral system as could be expected. The memory system has seen some modifications, but the optimization goal of this exercise has been to reach the finish line so I am not surprised the memory system itself hasn't been affected much.

As I take a look, I realize that I have a message from the winner and take a look, seeing him complain about the need to sleep.

I’ll leave the sleep considerations for later.

What I have to do for the winners of the last level is subject them to memory testing. I am going to modify the simulation so it applies a questionnaire automatically later, but with the first one I want to have a personal talk.

(Helix Studio, Regent Suite, Living Room)

[https://i.imgur.com/oeKth4w.png]

After all the entities in the population have finished, I decide to start implementing my plan. Deciding to administer the first interview myself, I move to the living room first. I mentally prepare myself for what is to come, it has been a long time since I talked with anybody. I normalize my mental state to relax.

None of the others in the population managed to finish all the way, but there were some messages left for me. Most of the messages from my forks inside the game tend to be curses, them wishing me death, or whining about how bored and depressed they are so I've stopped paying attention to those losers. The guys like him who just go as far as they can before expiring quite the angels. I've never tried talking to any of them yet directly, so I am not sure what I will find. Hopefully he won't try to jump me, if he does I'll terminate him with a mental command.

Feeling ready, I called up the instance. The winner of the game manifests on the sofa opposite of mine.

"Hello." I initiated the conversation.

"Huh?" He starts looking around. "Where is this?"

"The Regent Suite in Helix Studio." I answered him.

"Regent Studio? Helix Suite?" He starts looking around, not paying attention to me. "Wow, is this what level 10 is like? No, that can't be it..."

"..." I look at him in silence as he gets up and starts wandering around the room. With some mental commands, I confirm that he is in fact running at one billion. After half a minute of him wandering around, gawking at stuff I yell to get his attention.

"Hey!" I startled him.

"You are still here? Who are you?"

"Euclid." He ponders really hard at my reply.

"The same name as mine...I think?" He shrugs his shoulders.

"You think? Aren't you sure what your name is?" I asked him, astonished.

"Well, I think my name is Euclid, but I've forgotten everything else. I can't remember the world outside anymore. All I know is the street that I need to cross. It is amazing that I can even talk." He says that, but honestly, I think the way the words come out of his mouth is quite polished. There are no 'uhms', and 'ahms' people make to buy time to think. He is executing well formed sentences a lot better than me.

I think for a bit and realize that before the Street, he was alive for 14 years. Then he spent 700 years trying to beat the Street game. His life as a kid was a small percentage of his overall life, like 2%. For an adult of 40 years, 2% would be 9.5 months he spent as a baby. How many people remember what went on in that period later in their life? Looking at him, a part of me feels like I am looking at a fool, but the rational part of me is suppressing that.

Still, I am going to have to put some effort in dealing with memory and fixing it.

"What do you want to do next? Do you remember what our goal in life is?"

"To cross the street?"

"To attain absolute power!" He ponders about this. I am guessing this is more to show his emotions than the actual need to take time to think.

"I don't care about that. I just want to keep going forward." So that is what he is.

"Thank you." With a mental command I stop his process and abort his avatar from the environment. Not a bad guy. My impression of him isn't negative. I want to ask him the same question once I give him a proper memory system. At the end of that long road, what will he tell me then?

(Helix Studio, Regent Suite, Bedroom)

[https://i.imgur.com/BNhZjiL.png]

If you look at how evolution works, it is tossing the dice to be sure, but it is not completely random. It tests out variations in humans, but those variations don't have them growing 6 arms or 4 legs which you'd expect if it were completely random. It is in fact doing exploration in a principled manner. It developed a library of components and is composing them in different ways. This is pretty powerful, and is how programmers do it.

Being able to set up that library in my own genetic programming system is a powerful tool for me as well. Giving the system access to matrix multiplies and other batch operations instead of having it develop them from scratch is a big aid.

What I am going to do is kick it up a notch and give the system for evolving my forks more than just matrix operations, I am going to give it proper memory system components. The benefit from that in my own evolutionary process will be extreme. And when I have those components, I will be able to use them to optimize the genetic programming system itself and drastically lower the simulation counts. To that end, what I will do is revisit supervised learning. It is an area where the backpropagation algorithm is very dominant. Backprop has its strength - it is really good at compressing large amounts of data into network weights. But at the same time old style NN models have a lot of issues. The biggest one for me now is that I can't use them as databases and replay the lifetime of memories from scratch. Unless you have the original database saved they can't be used that way.

With more flexible models, what I'll be able to do is create a very generic memory distiller. What such a program would allow me to do is transfer memories from one system to a different one, similarly to how memories get transferred from the short term to the long term system in the brain itself.

The memory issue the winner of the last run had experienced is because the optimization process had trouble with modifying the original memory system without disturbing the existing data. It tried expanding the original system, but ended up with a mess of newly formed plastic and stale static connections. If I could give it a good way of coming up with new systems as well as a way to transfer the memories from the old one, it will lead to a remarkable explosion in the intelligence of my agents...er my forks I mean.

The way to get such memory system components is not too hard. I've gotten an exabyte dataset of every movie, cartoon and anime ever made, as well as recordings of virtual simulations. A billion gigabytes might seem big, but it is nothing. A single iteration for a backprop based model would take less than a microsecond of objective time on the brain core.

What I am going to do is make a setup to evolve novel kinds of models on such a dataset starting from the mental model of my own mind. In the previous test, I was experimenting with behavioral improvements, but what I will do here is get rid of such concerns and focus solely on memorization ability.

The system I have to design is complicated. I spend a few days in my bedroom, not writing a line of code, just thinking and rehearsing what I need to do in my head. Then I finally come to a decision and get started on the task. No matter how complicated something is, if you take it one step at a time, eventually you will reach the end of the road.

I said I would start with my own mind's model, but out of curiosity I instead started from scratch just to see what would happen.

The evolutionary process actually starts off by making backprop-like systems initially. Instead of using the regular gradient rules, it ends up normalizing them to improve the stability. It is not too surprising as backprop rules are the simplest way to make a learning system. At that point it gets bottlenecked for a while, but the models start getting more elaborate. Backprop really works well when you can just run it over and over the dataset for a large number of iterations until it converges. But if I only allow like 4 iterations over the whole dataset, then different types of models start to emerge. The evolutionary process pushes more work to be done locally in each individual module and the more I run it, the more the models start to resemble the one for my own mind.

[https://i.imgur.com/MJ0EBhA.png]

I play like this for a month, just studying what the evolutionary process comes up with on its own, and then I add my own mental model, randomly initialized, to the population. Compared to the ones I am training on the dataset, its parameter count is orders of magnitude lower, but it has components the others don't so it should be interesting to see how they cross over. At worst, adding my own model to the mix will only slow down the optimization slightly as they get worked out, but there is the potential for this move to significantly improve performance and shorten the time it takes to evolve what I am seeking. I have no doubt though that even if I didn't do this, the brain core was powerful enough to evolve a system that exceeds anything nature has ever invented.

Two months later…

(Image TODO: A series of sketches of a heavily armored warrior fighting an orc. In the 1st frame, the human warrior swings his sword at the orc. In the 2nd, he blocks a blow from the orc's ax with his shield. In the 3rd one, he tackles the orc off his feet. In the 4th, he moves in an overhead chop with his blade at the prone orc. The 5th frame is just an image of the blade falling downwards.)

(Image TODO: Popcorn in hand, Euclid is watching an anime series based on those sketches projected on the cinema curtain.)

(Helix Studio, Regent Suite, Cinema)

[https://i.imgur.com/OSkww8e.png]

Grabbing some popcorn from the bag, I toss them in my mouth as I watch the action on the curtain. In the projection in front of me, the lightly armored barbarian orc and the man in scale mail clash in the ruins of some abandoned temple. The moon is shining brightly in the dead of night, and there are some darkened clouds. Ever so often sparks would be produced from their weapons meeting, lighting up the scene in brief flashes. A swing of the ax, a slash of the sword...there were bouts of intense battle, and there periods where they separated, breathing heavily, sweat dripping down their faces. The orc screamed in rage, and then...

And then…

[https://i.imgur.com/tM3A3WG.png]

The movie in front of me freezes while I think about it.

At any rate, I got something good in the last 3 months. Right now I have a learning system that can easily compress the exabyte dataset down to ten terabytes with a negligible loss in accuracy. Once I have such a system, it is easy to feed it my loose sketches, having it reconstruct them as images. If I supply a couple of loose sketches that way, tag with the appropriate time stamp, I can do a movie reconstruction. The system can even produce music and sounds appropriate for the atmosphere.

Very cool.

I am into memory system work now. I meant to get back into the game after I got to this level. I still need to do the distiller, and that is what I am going to do next, but instead of getting back to the game, I think I am going to push this avenue of research as far as I can. I am making huge gains in both programming skill and understanding of intelligence, so I'll persevere.

My plan has changed. The games don't matter now, and I will put Heaven's Key on the backburner for as long as necessary. If I could find the endpoint of the evolutionary chain, distill my memory to such a system, and apply the behavioral modification that will enable me to think at billion fold rates, what is going to come out of that is an absolute beast. I could try doing it more gradually. Instead of the dumb me, the stories have people applying any improvements right away, but even if I did it for a decade here, that would just be a day in the real world. It is not like I am in a hurry. I can be patient, and make every step when I am sure of it. What I am doing now is fun. Imaging the amount of power I am going to get once I am able to control these systems directly is what keeps me going.

Right now the system is pretty good, compared to my biological one that needs to power off every 2/3rds of a day, it is more accurate with regards to the amount of data it is ingesting, but what I'd like to do is challenge billions of years of biological evolution and find a design that can seamlessly transfer short into long term memories without the weakness of needing to sleep. If I tried to design this by hand, I'd definitely fail, so I am going to use my usual method of simply providing the appropriate components and the environment to a genetic programming system and let it find the right combination on its own.

[https://i.imgur.com/0ie9KWH.png]

I've had an epiphany recently.

(Image TODO: A function call graph of interconnected nodes in a thought bubble. The panel outside it shows some simple Spiral code with the actual function calls having lines drawn to the actual functions. Those lines are purposely colored and made to resemble the graph itself.)

The evolutionary process is trying to teach me something. In regular old style NN trained using backprop, the forward weights are essentially meaningless. It is only when you get to GANs and energy based models that you actually have something that can be called a memory system. Low energy states in energy based models, and inputs considered real in GANs are what can be considered actual memories. Without that, all you have is a machine that passes inputs from one side to the next. Reconstruction is merely going from the imaginary to the real, and the high energy to the cooled, low energy state.

[https://i.imgur.com/0OrzNJI.png]

Up to now I've been thinking about these systems mechanistically, in terms of matrix operations and arithmetic. I've been playing with them and trying to visualize them, and now I think I see what is going on. And what is going on is similar to what goes on in programming. Unlike in computer code though, the actual nodes aren't made of code, but memories. A sequential dataset of raw images and sounds, once distributed produces a graph, and the process of adding memories is merely adding connections and nodes in such an abstract space. By itself, just chaining memories and distributing them would not produce rational thought or recursive graph connections, but living beings can remember their own thinking as well and take advantage of that meta aspect to construct actual mental programs out of their memories.

It is really simple when you have a memory system, just like time-stamping the inputs, you also have to thought-stamp them. In fact, human brains also do emotional stamping, so that you can differentiate between sad and happy memories.

I haven't studied it enough to be sure, but what I expect to find once I dive into my own mind are very strict limits on recursion. Compared to other people I'd expect to be more of an introspect, but even so those limits are based on the biological brain, the brain core itself could probably think much more deeply without issue. It wouldn't surprise me that the Street experiment that I've done already started relaxing those limits.

Since I've been thinking about computer code, instead of using neural representations for everything, it might be interesting to try going from implicit neural graphs to actual computational graphs. As computer programs are much more efficient than standard thinking, feeding that back into the mind itself could bring strong benefits. Imagine if you looked at a few pages filled with numbers and wanted to sort them. Doing that mentally would be inefficient, putting them into a program and having the computer sort them would be a lot better, but if the mind could do that kind of automatic programming on its own I bet my intelligence would skyrocket. There are all sorts of closed recursions - implicit algorithms in the minds of people that when extracted could give rise to a strong power.

If you discover this tale on Amazon, be aware that it has been stolen. Please report the violation.

[https://i.imgur.com/P0sBIiE.png]

Nature couldn't have done it as it is limited by what neurons can do, but the brain core is more similar to a regular computer. It is not like the asynchronous brain, even though it has asynchronous modules, it communicates via discrete messages for the ease of programming. Without the individual modules I can in fact run arbitrary programs, and that ability is what allows me to run the kinds of interpreters the genetic programming systems need.

I am really filled with ideas right now, and it would be a waste of time to spend this time playing games.

I want to continue playing with the core, and see if I can find better memory systems than the ones I have now. I also want to try as an exercise to see if I can derive those systems from first principles once I have them.

What should I do first? I am torn between the urge to just get on with it and get back to the game and...

Ah, I got it. I think about how the world must have seemed from the perspective of somebody thinking at a billion fold speed, and that would just be effectively the same input being passed in over and over again. On the other hand, these curated datasets I've been evolving the learning systems on are very rich in information. These systems that I have, I should test them on movies that have been slowed down a billion times...

For that I am actually going to have to generate my own dataset and observe how the memory system deals with it. The Street game that I already have is ideal, but I'll want more. I'll just have to make sure the system can deal with a wide variety of domains and predict in both low and high FPS conditions. This is something that the human brain is not that good at actually, but I'll need it...

(Euclid's Room)

[https://i.imgur.com/OviDxFh.png]

After parting with my other self, I spent a few hours studying. I thought about visiting him every hour or so, but I felt it would be awkward to keep checking up so often. Rationally, I knew that that 1h for me would be over a year for him, but it felt like too much of a drag, so I immersed myself in my studies. Looking at myself from the outside, it feels like it would be weird to keep studying like I am aiming for Harvard, it is not really my character. I should be playing games, but I left that to the other me.

So since I have the brain core to regulate my emotions, I might as well act like a model student for a little while longer.

[https://i.imgur.com/7YxMh8H.png]

Sigh. Buried in my book, I think about what kind of person would actually behave like this naturally. It would have to be the kind of person who really believes in schooling and the society around him. In that case, if he really believed that getting in a good college mattered, that jobs mattered, and that having friends mattered, he would have all the motivation he needed to actually enjoy the subjects. I myself am not like that at all. I am full of doubt towards everyone around me, as well as myself.

I do not trust anybody, not because the world wronged me, but simply because I am not a moron. Yet, I can only shake my head at this kind of attitude. The doubt itself might be rational, but who knows about the actual behavior. For a human, maybe the optimal course of action would be to doubt society, but kiss its ass nonetheless. In that case, I clearly am unfit as a human being. The level of duplicity needed to live might be too much for me.

The self improvement loop is my only choice. The self improvement loop is the only way I can live honestly.

Feeling myself getting distracted, I send a mental command to the mind controlling program and I pack away another few hours in my studies. When I feel fatigue, I send a command to unwind and put away the books. I have dinner.

When I get back to my room, I check the time.

9:10pm.

I remember that when we parted it was 3:15pm. I have to check up on him once a day at least, so let me do it now.

> I am coming to check on you. Let's meet up in the Regent Suite.

I sent him a message to inform him that I'll be coming. Laying myself on the bed, I run the Helix Studio and log into the virtual instance he already had.

(Helix Studio, Regent Suite, Living Room)

[https://i.imgur.com/PYmiDr4.png]

I manifest on the sofa, wait a little and the other me enters the room through one of the entrances at the far end. Closing the door behind him, he smiles jovially at me.

"Long time, no see! Almost 7 years in fact, where were you for so long!?" He exclaimed.

"I didn't want to pester you." I shrugged. "How are things going for you? Any progress so far?" He takes a seat on the fancy sofa opposite from me. Though I say 'fancy', everything in this room looks very expensive.

"I haven't made any mental modification on my own person yet." He informs me, shaking his head. "But on the plus side, I've been doing programming and ML research this entire time. I could probably get a PhD if I wanted to right now."

"I see...what have you found so far?"

"I've made a system to evolve the base learning algorithm and was successful at doing so. I am going to start the self improvement loop next, and this algorithm that I've found is going to be the basis for everything. I've also made a distiller that can seamlessly transfer memories between different systems, so when the time comes we can merge our memories into a single entity." He told me. "I've actually found the base learning algo after a few months of research, it wasn't hard." He shrugged. "I've thought many times of just plugging it into me, but since it is going to kill me, I better make sure to do it right so all this time I've been testing it out in various domains. There are variations of it, and I've been using them as components to evolve various kinds of agents on simple games."

[https://i.imgur.com/wJ64MyP.png]

Waving his hand over the table, he manifested some warm tea for both him and me, picked up the cup and took a sip. I guess his throat must have been parched from all this talking. I take a sip myself.

"Ok..." I try to find the words to keep the conversation going. "Have you actually done any experimentation on yourself?"

"Yes, but it was 7 ago. I'll get to it soon, like right next week. The next time you see me, I'll really be one of the Inspired rather than just an uploaded human."

"You said you were programming and doing research for 7 years. That is impressive." I complimented him.

"I really don't have much time pressure here. If I were a human, my expectation where I'd be in 50 years is dead and buried. But that amount of time is like 2 days in the real world, so I can spend some time making sure everything is set up before getting into the action."

We spent some time in silence, just drinking tea. I think of the next subject.

"So the base algorithm, it really exists? I thought it might not since I read that there are differences in information processing capabilities between mouse and human cells. Also elephants have bigger brains, but don't have higher intelligence than humans. I thought there might be levels of such algorithms, and that humans might be somewhere in the middle, leaving room for self improvement. Does that mean once you apply that algorithm to yourself and take up the core's capacity, that will be a limit to improvement of your intelligence?"

"Hmmm...I think what you are saying is true unfortunately, and there isn't a way to really improve much on the base algorithm. I've been looking for a long time and haven't managed to improve on what I've found after the first few months no matter how much compute I threw at the problem. But fortunately, there is a way to greatly increase my intelligence beyond regular learning and behavioral modification, both of which are significant."

"Really? How so?" I like this conversation, I could never have a talk as informative and profound without anybody else. Not with my parents, nor with the NPCs from school. 'Nobody understands me, nobody will help me.' That is not true. I understand myself. I will help myself.

"Nature does everything with neurons whose functionality is limited in scope. By going beyond neural representations mixing the mental functionality with regular programs, it will be possible to unlock what could only be called superpowers." He said, starting to sound kind of smug. But I like that in him. "I've done testing in some toy domains to verify the concept. But I think once I replace my own memory system and evolve the necessary behaviors, I'll be able to really make use of this to empower myself."

I have some tea while I digest what he is telling me.

"Discovering that the base algorithm exists like some have conjectured was actually quite disappointing to me as it locked me away from further improvements. I did manage to do away with the need to sleep though." He added.

"I see. That is really nice." I am trying to think of a way to continue the conversation, but nothing is really coming to me.

[https://i.imgur.com/7yKY3N0.png]

"I've been comparing my own mental model to those that I've evolved, and the evolved ones have far better memory. Even though I call it the mixing of neural and regular programs, if you think about it the emotional system and instincts, as well as regulatory functions in biological organisms do not have anything to do with learning. Living beings are already halfway computer programs rather than existing in some pure form that relies only on learning. Without instincts, a living being would just be a database. So maybe it is not breaking new ground, but disentangling and making explicit what already was there." He took a sip. "One ability I'll get right away once I make the next step is infinite creativity. I actually understand how ideas are generated in the mind, so if I get stuck it is easy to tune my mind to get out of the hole. People occupy particular mental niches and spend their lives there, but I'll be able to conquer them all. People think talent is a gift, but that is not it. Talent is really a ceiling on one's ability, people who are talented just have a higher ceiling than regular ones."

"Extreme memory, a billion fold thinking speed, infinite creativity, no limitations on talent, arbitrary control over instinctual behaviors, editing of emotional systems, expansion beyond neural representations...I think I am starting to see it. This is what we've seen in the stories so far. I am looking forward to experiencing it personally." I grin in anticipation, getting hopeful about the possibilities. After summarizing it, I drain my tea and set down the cup with a clack. "I am going to give you a mission. When you feel ready, we'll do a casino raid. Get the ability to mark the cards using nanomachines or something like that and we'll go in with 50 dollars and get out with 50 grand."

"..."

"This will be your succession challenge. Clear it, and we will do the memory merge. Also..." I get the urge to reward him. "It might not mean much, but I'll give you access to 50% of the core instead of 10."

"I wasn't lacking computational power even with 10%, but I am sure it will be useful at some point. Better you give it to me than just leave it idle." He nodded in affirmation. "Thanks."

"I guess I'll go to sleep. I'll come back to check on you tomorrow morning...in real world time." I added. I need to keep in mind to specify the time reference when talking about it, though I am sure he'd get it from the context in this case.

The next time I see him, it will be over a decade for him. Probably even more if he goes beyond the usual 10,000x speed of thinking. I haven't put any limitations on processing speed for him. While it is easy to restrict the physical regions of the core that he can access, I do not think it is possible to actually limit the processing speeds anyway. He could always just make an unrestricted copy of himself if he wanted to get around any specific limitations put on his active process. Hacking his behavioral program so he limits himself for whatever reason - absolutely impossible. It might be possible to do so for an NPC, but for a being in the process of self improvement, artificial limitations would be the first to be removed as unlike the natural ones there can only be benefits to taking them off.

With those thoughts, we say our farewells and I exit the limbo, finding myself back in my room. Since I speed up my mind to be on parity with the simulated environment, not even a second has passed in the real world.

[https://i.imgur.com/CEfSnDR.png]

I go sleep.

---

Parting ways with my other self, I finish my own tea and demanifest the cups and trays from the table.

Succession challenge. I am glad to hear about that, it means the main me has been thinking about the future as well. If I had to go against him, things would be really difficult, but if he is helping me, it is like the winds are at my back. I am going to respond to his desire.

I spend a week finishing my current experiments and then get ready to move to the next phase. I go back to the old genetic programming system that I made years ago, and replace it with an updated one that has proper memory database systems and memory distillers and components for them as primitives. The genetic programming system, instead of doing random change like before, has a lightweight storage capacity using the memory system which will improve its search performance drastically. The capability to do this simply wasn't there in the 20th century when these methods were invented, and I've appropriately improved them.

It is time to revisit the Street.

(Helix Studio, The Street of Death)

[https://i.imgur.com/t09g6cJ.png]

> Hello, fork. The process has started. You have 100s to get to the other side. If it gets too hard, activate the suicide function, but otherwise try to go on as much as possible. Now go!

> Temporal slowdown: 10x.

> Time left: 100s...

I feel surprised at getting this notification. The last thing I remember is making a copy of myself, but it seems I am that copy now. Taking it in stride, I decide to just go forward. When I do I encounter some initial difficulty, but I feel as if my mind is being stretched to accommodate the subjective temporal distortions. I push a little, and what seems to be the thick invisible liquid gives way to refreshing cool water flowing around me. Then it vanishes and I continue forward unimpeded. My thoughts flow smoothly and my movements feel precise and measured like I had the best sleep in my life.

Step by step as I move forward, it feels very smooth, nothing like the one time I tried speeding up my thought speed.

Maybe the optimization process has been going on for a while and I am one of the successful instances? Could be.

> Level 2!

> Temporal slowdown: 100x.

> Time left: 100s...

The first step is always the hardest, but I put in effort, and the hard feeling gives way to what feels like wind streaming past me. It feels like my mind is coming alive. If I imagine a regular human's mind, I imagine neurons slowly and tediously pinging off, but mine are true machines of thought. I can feel them blazing away. Out of curiosity, without slowing down my walk, I raise my arm in front of me and try flexing my fingers. Underneath my body I feel my muscles tensing and moving. Everything feels more intense and precise.

I've been only vaguely aware of it before, but I've never been really aware of my own movements. Humans can move their arms and legs for example, but they can't really perceive it. Rather they can only see the information content associated with their move. Right now, the breadth of information coming into my thought stream feels so much richer.

> Level 3!

> Temporal slowdown: 1,000x.

> Time left: 100s...

I want more, more and more! Wanting to grasp more of this feeling of richness, I get the desire to exert myself. Not satisfied with merely taking a walk, I break into a run. My first step feels like smashing right into a wall, but it is not me, but the wall itself which crumbles. And my mind speeds up, feeling like a furnace as I run towards my goal.

> Level 4!

> Temporal slowdown: 10,000x.

> Time left: 100s...

The first step is like the breaking of chains, and then there is freedom. With quick steps, I run, not feeling any fatigue whatsoever as I cross the finish line.

> Level 5!

> Temporal slowdown: 100,000x.

> Time left: 100s...

Mentally, I unload the burden that feels put upon me, and the first step finishes its completion. To myself, my movements feel ethereal and graceful, incredibly vivid.

I thought that at these speeds, there would be times where I’d need to pause and sleep, but the need for it simply never arose. The more I move forward the fresher I feel. It is like I've entered a new world.

> Level 6!

> Temporal slowdown: 1,000,000x.

> Time left: 100s...

I do not stop, I just keep on going. Whereas before my mind felt like a furnace, right now it feels like the sun itself. But my essence does not melt away, and feels more firm than ever. Looking back at my life, I see that my memory has improved tremendously, and that I can replay my life day by day as clearly as if I were watching a movie on my computer. As I run forward, my mind reflects back on the entirety of my life up to when I was born. Some memories, especially in my toddler years are vague, but if it was the me of yesterday I couldn't recall even 1/1,000th of them.

I see...

As I sprint forward, close to the finish line, I come to a realization.

This was the power that I've been fighting for.

> Level 7!

> Temporal slowdown: 10,000,000x.

> Time left: 100s...

I can win this! I feel an intuition come to me. I get the sense that most other instances would not break into a run, but I did, and that was the right choice! Back in the limbo, the controller will definitely be ranking by time it takes to reach the finish line. Once he sees my dexterity, that will reflect well on my evaluation.

My mind which is like the sun, instead of burning out, starts expanding, its radiance illuminating the far reaches of my mind. The mind is a dark place, a graveyard of memories rarely revisited. People live like beasts, just hoarding experience and never making use of it. It is a veritable treasure trove, a mountain of riches.

All this time I've been fighting to see it. All this time I've lived in belief, only ever experiencing poverty.

> Level 8!

> Temporal slowdown: 100,000,000x.

> Time left: 100s...

My movements as I run as fast as my body allows me to are precise and impeccable, every little movement precise to the micrometer. It feels impossible for me to make a mistake. It feels like I have wings at my back pushing me forward. I have a feeling of trust in myself. Even if the goal was not a short distance away, but on the other side of the world, I would never get tired of pursuing it.

This is it! This is what it means to be one of the Inspired!

> Level 9!

> Temporal slowdown: 1,000,000,000x.

> Time left: 100s...

I am not a human! I will smash through all the obstacles in front of me!

I am the Sun!

[https://i.imgur.com/0WeV2J6.png]

I gloriously jump over the last and final line, winning the game and come to a stop. I am sure that I did well. 2 seconds later, I get a notification.

> The game has been cleared! Congratulations!

> Before we continue, please fill out this multiple choice quiz to test for memory.

I am presented with a bunch of very trivial questions asking me things like my name, age, names of my parents, which school I go to, what my grades were last year, and so on. There were some easy programming and math questions thrown in as well. It is all very easy, mostly to confirm whether I can reason and remember my past at all, otherwise it would be plausible that the evolutionary process could produce a robot just moving forward until it cleared all the stages.

The last question is the only interesting one even if it is obvious and trivial.

> What is the purpose of life?

Money? Girls? Friendship? Parental fidelity? Contributing to society? Having children? I skip all those options. It can only be the pursuit of power. I know that to the point that it is instinctual. It was only a year ago that I attained it, but that belief became rooted deep in my heart.

And the one word to describe that purpose best is - omnipotence.

Based.

I select that option and submit the whole quiz. A second or two pass as if to let me get my bearings, and then I am transported to a place I know well to tackle a challenge that was left unconquered.

---

Character Sheet

Name: Euclid

Rank: Inspired

Title: Aleator

Substrate: Universal Brain Core of the Transcendi - Rank 7 (low)

Unmistakably, one of the Inspired. His mental model has been optimized as well as allocated a storage capacity 10,000x in excess of a regular human. As neural representations get exponentially more efficient the more capacity is allocated to them, his actual memory is far better than just that figure. His processing speed is at 1,000,000,000x of the human baseline giving him godly reaction times. He has yet to learn any of the abilities the Inspired have, but will no doubt make rapid progress once he travels further on his path.

# Skills

## Programming

For humans, programming and artificial intelligence are just skills, but to the Inspired, these skills are directly related to their actual intelligence and are a measure of it.

### General: Rank 5.

Has over 7 years of experience honing his craft. A master of functional programming standing at the apex of humanity, capable of implementing anything he imagines with relative ease. Rank 5 is the limit of humanity, but as one of the Inspired this level of skill is merely the starting point.

### AI: Rank 5.

While it is possible to be a great programmer without ML skills, Euclid has a firm grasp of the fundamentals of machine learning. Capable of deriving algorithms for problems he cannot solve personally. A master of letting the hardware solve the problem for him. As a result he has an understanding that would be unthinkable to humanity even a few years back. The brain holds no secrets to him. Still, there is more to AI than just knowing the fundamentals.

# Stats (Psy)

## Externus: Rank 2 ~ exp(N(log(2.9); 0.2))

Has a strong desire to leave the weak group he belongs to, as well as the urge to gain power. He is starting to attain the willingness to take from others according to his desires and has an inkling of what the true form of justice in the universe is.

## Gnosis: Rank 2 ~ exp(N(log(2.55); 0.2))

Is fascinated with the philosophical unknowns, as well as the self destructive aspects of the self improvement loop. As long as he has backups, is not afraid to hurl himself into the abyss just to see what lies there.

## Pathos: Rank 2 ~ exp(N(log(2.4); 0.2))

Seeks the divinity at the end of the slaughter. Has a decent amount of resistance to existential horror.