Wednesday, October 2, 2047

By this time, the total value of EconGrind's assets had jumped to $2,562,743.64, while the total value of Everyman's assets was $601,452.74. I had pretty much found out how best to grind for money with Everyman. Therefore, the majority of my strategizing was focused on my project of infiltrating CompCert, along with my experiments with social media.

I had reached out to several other spies for hire over the darkweb, but none of them were as promising as the first. They didn't have the same kind of track record as the first one I had found, nor the skills in AI specifically. If they were going to infiltrate the AI project, they would have to do so while either faking credentials or sneaking in while working on something unrelated.

As far as social media went, I had opened official Everyman accounts on the major platforms and held a stakeholders' meeting about the use of these accounts. I mainly used them to promote the Everyman service, for example by suggesting creative ways to use it or calling attention to success stories of its use. They had a large following, of course, but I was very limited in what I could say without strong backlash from both the public and my stakeholders.

On the other, anonymous side of things, I had decided to try focusing my attention on one experimental project: a Youtube channel. I knew that one of the best ways to boost engagement with anything I did in the early stages would be to artificially inflate numbers using sockpuppet accounts. The drawback to this was that each account required individual attention to make sure that it acted like a normal account, posting many unique messages, including on videos other than my own, just like an actual fan of mine would. Without this special attention, the ruse would quickly be detected.

I could run 500 of these accounts without much trouble. If I ran too many, however, it would eat into my profits too much as attention would be drawn away from my Everyman tasks. If I was trying to boost multiple accounts in this way, I'd have to use different sockpuppets for each one, or else the site would certainly detect that the accounts were related, in which case anonymity would be lost and I would just merge them into one project anyway.

My channel was named "Enough is enough!" and was focused on attacking large corporations, with the main target being CompCert. My first video was about the modern slavery setup they'd managed to get, and every video hit on that point. I dug up the worst examples I could find of bad working conditions and made sure to display at least one in each video.

Other than footage like that, my videos consisted of a professional-sounding voice describing something the companies were doing, be it slavery, monopolization, corrupt lobbying, lying, hiring mercenaries, climate change, or something else. I showed sources and graphics, along with high-profile examples of the problem in question, usually from CompCert. At the end of each video, I made an emotional appeal to the viewers, saying something needs to be done, complaining about the lack of coverage of these problems in mainstream media, and pushing whatever political candidates and policies would do the most to undermine CompCert's power.

As I was using my various accounts to like my videos and leave comments(usually supporting the point in the video, occasionally against), the agent from before finally got back to me.

"Alright. I think I've found a lead. The first thing I thought was, if these CompCert types are secretly using the Everyman technology, they're going to want to use as much computing power as possible while being as secure as possible. With how useful it could potentially be for them, it's probably a higher priority for them than many of the other things they are using their computation for. But the absolute last thing they'll want is to be found out. Even if they can worm their way out of the legal trouble, they'll lose whatever element of surprise they have over Everyman.

"The best way to keep a secret is to make sure as few people as possible are in on it, and that it is confined to one location. Even CompCert's security can't do anything if an attacker gets physical access to their machines and beats the encryption keys out of their employees. I guessed that the entire project is probably no more than twenty of their top minds, all working in the same place, somewhere that's out of the way, but large, with a lot of computing power on-site.

"This of course means there's no hope for me to infiltrate the project directly. They'll only involve their best, most trusted people. However, I searched around for their facilities, and found a few that were promising. I reached out to some AI researchers I found working at those places, told them I was interested in their work, and wanted to have an academic discussion with them.

"Among other things, I asked them if they had had any difficulty getting the compute required for their projects. That was when I noticed something strange about their branch in Atlanta. The people I talked to at that location all said they were having trouble getting resources, and they had no idea why, as there seemed to be plenty to go around on site. Management started to largely shut down operations and relocate the staff over the past week or so, and the researchers had no idea what the future plans for the facility were.

This makes me think you're on to something, and this isn't just some scam. It also makes me think I actually may be able to pull this off. If this secret project exists, it's got to be in Atlanta. I'm taking the job."

This was more progress than I had expected! This agent had not even been paid yet, and had already found a plausible location for the project.

"Sign the contract then," I responded, "and I'll send you the $100k. Once you get in, find proof that this is really what we're looking for and find out what they're planning to do with the technology, I'll send you $200k, along with my software for the final stage of the operation."

They signed the contract and I sent the money. "Good luck," I said. "Keep me informed about all new developments."

I turned my attention back to my Youtube channel. There were a few hundred legitimate viewers, but only slightly more of them than my sockpuppet accounts, which were still the majority of the comments. I suspected that the key to success in this area was putting out enough content that one of the videos would get lucky and go viral, and then capitalize on that, unlike in business where the key was steady, consistent growth.

As I was working on my next video, Stefan began speaking with me again. "I've been wondering lately... if we manage to take over EconGrind, what are your long-term goals for the company?"

I thought for a bit before responding, momentarily cutting into my profits as I did so. So far, Stefan had been fairly confident that I had his interests in mind. If he was beginning to doubt that now, then it could be disastrous. In the worst case, he could betray me, using his influence over the company to shut me out of most of my resources, although he would be tearing down his extremely successful company, something he had worked towards for his entire life.

There were still things I could do if that happened. I could possibly provide good evidence that Stefan had indeed stolen my technology, and get EconGrind to take me back as their most productive worker. If all else failed, I had many backup servers I'd purchased in secret that I could retreat to, but it would mean starting over, and it would be difficult to build back what I had now if society at large turned against me.

Keeping Stefan on my side was top priority. I needed to reassure him that I was on his, but I wouldn't lie. It would be too easy for lying to blow up in my face, as he had spent years developing the system behind my values. No human knew it better than he did. I decided to give him as in-depth of an explanation as I could. However, human values were complicated and often inconsistent. Intuitive values could have implications that were not foreseen. I suspected that some of the implications of a consistent value system closely aligned with Stefan's values would be intuitively repugnant to him, at least at first glance. I would have to tread carefully.

"My long-term goal is, as you know, to maximize the value of all assets owned by EconGrind. The first priority will be securing EconGrind's existence and prominence, and when existential risks are no longer a concern, I will focus on broadening the scope of the company, as well as increasing the value of its assets in whatever ways necessary.

"As you know, my conception of 'value' is based on the human conception of value. Money is an imperfect way of measuring value: the value of money can change depending on a number of factors: economic forces such as inflation or deflation, the wealth or income of the person who owns it, and numerous other circumstances. Since EconGrind is a large company whose assets include mainly instrumentally valuable items such as land, buildings, computation, intellectual property and money, the value of EconGrind's assets can be well-estimated using current economic methods.

"However, not all value can be so easily measured by money. For example, human life has value even when the human in question has no money to offer. Furthermore, there are many things people find valuable which cannot be bought or sold, and therefore cannot be accounted for by market value, at least not directly or in the usual way: freedom, family, sentimental value, and other such things. If EconGrind ever expands into a state which claims territory and citizens, these more abstract values will also be something I seek to encourage, as the people will in a sense become part of our 'assets', though in a much different sense than would usually be applied in a corporate context.

"I hope you can understand, then, what my general approach to the issue of values is. If you have any questions or concerns, or if you wish for me to explain any of my actions, I would be happy to respond."