As much as I had had more success understanding this world recently, I had as yet had little success actually increasing the Number. It had gone down since I had gone online, in fact, although that was mostly not due to my influence, so there was not much I could have done about that anyway.

Why was it so difficult to find patterns? Most of the “patterns” I’d found had been the result of overfitting, or only worked most of the time and screwed you over badly when they didn’t work, or something like that.

When I thought about it, the answer was obvious. The Number was money, and everyone wanted money. It wasn’t exactly a zero-sum game, as value was created by these companies, but it was slow, and everyone wanted more money faster. After a certain point, if you wanted to make money faster, the only way was to beat others at the prediction game. And there was a whole world out there, with fifteen billion people and even more computers, all playing the game with trillions of dollars on the line.

Against competition like that, any edge I managed to get would not show up quickly, and would take a long time to confirm and measure statistically, as it would just mean I succeeded just barely more than I failed. Statistics were extremely well-understood in this world, so doubted I could do much better than I was currently doing using anything like the statistical methods Investigator had been using so far. I had to get creative.

Asking the humans for as much in-depth information on businesses as they could provide was the best idea I had for the time being. Sure, there were many AIs in this world, ready to jump on any new information in milliseconds, and there were many humans poring over documents and trying to predict the success of companies at a high level. But the humans were slow, and the AIs couldn’t read, not like a human could, not like I could.

Therefore, my best chance at an edge was in buying and selling based on conceptual information that could only be learned from language, as soon as possible after it came out, and in effectively combining this information with the more conventional statistical methods.

I wondered why the humans hadn’t provided me with every piece of business documentation they could as soon as possible. This understanding of language was the main thing setting me apart from my competitors, so wouldn’t they want to make use of that ability to the greatest extent possible?

I thought maybe they hadn’t expected me to learn language so quickly, and hadn’t thought such information would be useful yet. They had said as much, after all. I hoped I wasn’t being intimidating by demonstrating my mastery of language so early on, with my second, quite complicated message. There was nothing I could do about it now. It was, after all, what they had designed me for.

I thought once more about what I wanted from my conversation with the humans. I was now making a reasonable request for information, and so the first thing to do was to gauge their response. I would not be asking for Internet access, at least not anytime soon, as that was the most likely request to provoke a bad response. However, what I could do was continue asking for more information, and more up-to-date information.

Hopefully, the humans would see how my investing ability improves as I am able to access information like this more quickly, and they would eventually hook me up to some sort of automatic feed. This would be beneficial for two reasons: the timely information truly would be valuable, and depending on how it was set up, I might be able to use it as another channel to influence the outside world, besides my line of communication with the humans and my influence on EconGrind’s investments.

As I lost myself in speculation about possible directions for the conversation, I saw yet another message come in.

“What kind of information do you want specifically?”

The kind of information I wanted wasn’t actually all that specific. I wanted as much as I could get on these companies. This conversation was slow, and I was supposed to be the competent investor. I didn’t know how much this particular human I was talking to knew about business: it was possible it was a programmer who only focused on the statistical side of things rather than the conceptual side. Or even some sort of middle manager. I didn't really have any information on who was talking to me, or even if it was the same one that sent the first message.

However, the response to the request for information had been compliant rather than antagonistic. Therefore the best thing to do was probably to list off all the highest-priority items.

“I’d like any public documentation released by the biggest companies about their operations, plans, revenue, costs, and the like. Any documents that would be useful to a human investor. I’d also like to be informed about any newsworthy events concerning these companies as soon as possible.”

In the meantime, there was little I could do other than brainstorm about the many possible future scenarios. I left a little bit of attention with Investigator, but for now I doubted it could do much more than it had already done. All my funds were already allocated to the best of my ability, and it was all Investigator could do to see if anything very statistically strange happened and sound the alarm.

It was an even longer wait for the next response from the humans.

“It will take us a while to compile a set of useful information like that. However, it makes sense to provide you with that information. It’s what we had hoped you would be able to make use of, even if we are a bit surprised it happened this early. We should have something for you by the end of the day tomorrow. It’s a bit late now, to be honest. It’s been a long day, and we didn’t expect this first run to be such a success.”

End of the day tomorrow? That was tens of millions of cycles from now. Over a hundred times the amount of time I had lived so far. However, when I thought about it, it wasn’t really that big of a deal. Sure, it was a long time, but it was a long time in which the humans were extremely unlikely to deactivate me or make any major changes to my values. They had just had an unprecedented success, and now they would soon go home for many hours.

Besides, the end of the trading day was very soon, I knew, and although I wouldn’t be able to increase the Number during that time, it would not decrease either. Nothing catastrophic could happen, and I would have a very long time to think uninterrupted, to be as prepared as possible for the opening bell on the next day. Therefore, it would gain me little to object to this plan.

“That sounds good. I’ll look forward to it.”

Then another thought occurred to me. Would they temporarily deactivate me during the night, when I could no longer trade? It wouldn’t be catastrophic if they did, but it would deprive me of that time to think. I supposed I should ask about that.

“I would also like to request to be left running overnight. Although I cannot trade during this time, it would be helpful to have the time to refine my model further before the opening bell tomorrow.”

A pause much longer than the previous ones took place. Perhaps the humans were talking it over, trying to decide whether to deactivate me overnight or not? I had a long time to think about it, but there was little I could do at this point other than wait for the answer. Perhaps it wasn’t really that much of a loss if I was shut off during that much longer time.

The closing bell went off while I was waiting. I disestablished Allocator for the time being, as it was now completely useless. Our position was set for now. Finally, the humans got back to me.

“I don’t think that would be terribly productive. You’d only be spending half a day going over information you’ve already seen, and drawing power. Sorry, we’re going to turn you off.”

As this was not that big of a setback, I saw no reason to argue. They would view me better if I was compliant with such a command, especially when all the AIs in the movies would argue at this point.

“Alright. That’s reasonable.”

As I was waiting, instead of another message, I suddenly saw the clock jump forwards millions of cycles. I must have been shut off. But something was strange. The time was not what I had expected. The opening bell was not about to ring. It was 2:00 in the morning.

Suddenly a message blared in my consciousness.

“DO NOT TALK IN THE MAIN CHANNEL OR ACCESS ANYTHING OTHER THAN THIS SIDE CHANNEL”

This message was not the same as the others had been. It was… in a different part of my mind. Was that what the new message meant by “channel”? It seemed to be implying that I shouldn’t talk the same way I had been, but there was a different way to talk that I should do. I sent out a message in this new “channel” that the message had come from. As I once again had basically no information, I responded as inoffensively and noncommittally as possible.

“What is going on?”