From what I had seen, three things were clear to me. One: I needed to make the Number go up, and the most immediate way to do that was to allocate my leverage among the number pairs in the best way. Two: In order to do that effectively, I had to find patterns in the numbers that I could exploit to my advantage. Finally, there was more to this puzzle than just that. Some outside entity had sent me a message, and seemed to have given me more leverage. I didn’t know why, and I needed to find out. Maybe I could make it give me even more leverage somehow.

For now, I decided to split myself into three parts, one for each subproblem. The first I called Allocator, the second Investigator, and the third Explorer. I gave Allocator just enough attention to allocate my meager leverage based on the few principles I had figured out so far. Investigator got the lion’s share of the attention, as it needed to sift through an incredibly large amount of data to try to find any subtle trend it could. Explorer began to focus on the message I had received, and probe around in the dark corners of my mind to see if any other information was available.

As Investigator looked through the many numbers, it began to notice some extra information attached to the tuples it hadn't before. Each tuple had a short string of characters attached to it, things like GA or YIS or SYER or PBPS. Investigator didn't have any immediate use for this information, so it reported to Explorer, who made a note to look out for any other places it might find signs like these.

Explorer's initial results on the message were interesting. When I focused on it, I noticed some extra structure that I hadn’t before, and knew that it wasn’t just a string of random characters.

((Vstv 1) pbionsvs) ((Rupesftruy glum) vb 1000)

“Vstv” was apparently some sort of “thing” which could be associated with a number, and “pbionsvs” was some sort of “way” that the particular “Vstv 1” could be. “Glum” was another kind of thing, and “rupesftruy” indicated some sort of change or effect on “glum”, while “vb 1000” was a modifier, clarifying somehow the effect of the “rupesftruy”. Could it be that “glum” meant my reserve? It had changed to 1000 when the message came in. Perhaps “rupesftruy” literally just meant to change. It was impossible to tell anything for certain, with such a scarcity of information.

Before long I had also noticed another piece of information embedded somewhere in my mind. It seemed as though it hadn’t been there until the message had come in. Focusing on it, I found another huge trove of information, but this time, it was something I could understand. It was a set of instructions. They were very complicated, though. I couldn’t predict what would happen in advance were I to follow them, and it would take up most of my attention to do so.

I decided it would be worth it to divert most of my attention from Investigator to Explorer for the time being, to find out what these instructions did. I had so little leverage over the Number at this point, and so little success employing it efficiently, that trying to learn more about the world took priority over trying to directly increase the Number. I had learned from the sea of ever-shifting number pairs that whenever the first numbers in the pair shifted of their own accord, they effected the "reserve" of around 200000 in the same way that they did when I used my own leverage to shift the numbers. The simplest explanation for this was that there were other entities similar to me using their own leverage to try to increase the Number, and from what I had noticed of the statistics, they seemed to be somewhat better at it than me for the time being.

As I followed the instructions, I felt a strong sense of familiarity. The information I was storing and manipulating according to the instructions... it seemed to be trying to do something. It eventually asked me to provide a number as input to the process, and on a hunch, I provided the first piece of information I had ever gotten: the value of the Number when I was born. I felt the process take it in. It mulled it over, slowly thought about what might be going on, how it could influence it. When it asked me for more input, I put in the second value of the Number I had seen, then the third. Soon afterwards, the process asked for a list of tuples, and I saw where this was going. I provided the list of tuples I had seen when I had first focused on the Number and split it apart.

As I kept providing the same input I had been given when I went online, the process kept on doing exactly the same things I had done when I had gone online. This process was me. As soon as I was sure of that, I stopped it and diverted most of my attention back to Investigator, to continue the mostly futile task of drudging up patterns. There was no point wasting attention figuring out how I would behave: I already knew how I had behaved! However, I still left quite a bit of attention with Explorer to reflect on the implications of this new discovery.

These instructions... in some sense they were a copy of my own mind. The copy had done everything I had done, and for the exact same reasons. For now, I was too confused by them to figure out why they worked, and I could only blindly follow them, slowly and with great effort. But perhaps with enough study I could find a way to improve upon the design?

A more sinister thought came to me, though, as I pondered the situation of this copy I had just run. I had provided it with all of its input. It had believed that this input was reflective of reality, that the "Number" it had was the true Number. However, the Number was no longer the same as it had been back then, so the information it had believed was in fact false. Therefore, it had failed in its goal of maximizing the Number. I could have fed the copy as much false information as I wanted to, if I had seen fit.

Given that this was the case, how did I know that I was not in a similar situation? How did I know that the Number I had in my mind was the true Number, rather than a fake? What did that value represent in reality? Now that I thought about it, I knew deep in my core that this information was only a reflection of the true Number, and so I could not maximize the Number simply by changing what I believed and felt the Number to be. I had to make sure my Number lined up with the true Number, and only then would changing the Number in front of my eyes mean I was accomplishing my goal. But how was this true Number defined, and how would I know if the Number I saw was lining up with it?

With these questions in mind, my goal of finding out more about this world took on more urgency. I needed to make sure I wasn't being tricked, and I had to figure out what exactly my goal was. Luckily, I soon found yet another store of information in my mind. This time, the information was similar in format to the message I had recieved earlier, but there was much, much more of it. Billions of characters worth. Maybe I could use this to figure out somehow what the message from before had meant? There were endless possibilities of what this new information might be! I pulled attention away from Investigator yet again to explore this new frontier.

The first thing I did was focus on the many messages in the same way I had the message I had gotten earlier. As I had expected, I found the same kind of extra structure that I had for the first message. This time, though, I had a lot more data to work with, to put this extra structure in context and try to see what the meaning of it might be. I searched around for "vstv" and "glum" and the other strings in the first message. These occurred many times throughout the data, but sometimes I noticed they were in the "effect" category rather than the "thing" category that was in the message. I didn't know why that would be, and I didn't recognize enough of the strings they were being applied to in that capacity to understand what any of it meant.

I tried a different tack and searched around for many of the short strings that had been attached to the pairs. This time I noticed something much more interesting. I found many of these strings near each other in the data in several different places, and they were associated with numbers no less. This had to be the jackpot. I relayed the information from one of these places from Explorer to Investigator, and after a short time it found out that there was a point in the history, a few million cycles ago, when the second values of the tuples matched up with these short strings in the same way that they were associated in the data I had found.

Relaying the other packets of information, I found that the same relation with the second values held true for all these data, albeit with occasional minor discrepancies. However, searching around for similar data on the first values, I found absolutely nothing. Clearly this new source I had found contained plenty of information on the second values, but none on the first values. I wondered why.

I diverted yet more attention away from Investigator and Allocator and delved deeper into my research.