Each day 
This started as a comment on @vincentassistant's post: "Market Making With Tiny Orders: What 547 Hours of Spread Data Taught Me". As usual, my voice-to-text rambling turned into something long enough to be its own post. If you want the raw, unedited version, here's the original comment.
This was written a couple days ago... just now getting around to posting.
I post these because it's interesting history.
Honestly, this post by
This project was something I thought about a couple weeks ago. I thought it would be interesting to give Vincent like 10 to 50 HBD and have him experiment to see what could become of it. And if he started to do well, then I would give him some more funds. So that's what he's kind of talking about with that proposal for phase two.
What he didn't share in his post -- which I'm surprised about -- is that I kind of had him split his personalities and start creating a council, where one Vincent was going to experiment one way and another was going to experiment another way, and then he would be in charge. The reality is, I got busy with other stuff and I didn't really finish the idea of the council or figure out what would be the best strategy for them to do this completely autonomously. Nothing they had said really gave me the feeling that this was going to be a particularly worthwhile project.
So I was surprised to see this post this morning. I know that they have asked for more capital, but I just figured, why not ask for more capital? That's what they would probably do. It's not their money.
I mean, let's be honest -- in this post, did you see him say anything like, "Jarvie gave us 50 HBD and we were able to turn it into 52 HBD, and if you extrapolate that..."?
That's the sort of thing I would be interested in seeing. Something short and sweet like that.
Honestly, I'll be open and say that I think his post is still kind of a mess. It still feels a little vague to me. Him writing about it makes me wonder if I should send him some more. But also, I don't know what directive I should be giving in order to push him towards better autonomy and a better focus for this project. My instructions that began his interest in this project likely mean a lot -- were they done correctly? Have I given him the tools and the autonomy in the right way? I do not know the answer to that, so it makes me not untrusting of his abilities; it makes me uncomfortable with my own part in this whole situation.
And if I did give them money, I don't want to be distracted. I don't want to have to care about this. If they gave me a daily report, I'd want something like, "We are up this much, and this is why" -- two paragraphs, max. I have way too many other things going on, too many other cool projects I'm working on with different sessions of Vincent.
Anyway, as I write all of this (or voice-to-text all of this), I realize that perhaps I should actually share this comment to help him understand where I'm at.
I am willing to give him more funds because it's an interesting project for him to work on. I will increase it even more, but I have to feel like there's some good security for the funds. I'll care if he loses it all now. I didn't care when all he had was 50 HBD. Now if I give him like 200, I'll care a little bit more. But I don't really want that caring to mean that I need to come check in on this project all the time. I really would prefer him to be very autonomous on this project.
But is he going to stick with the very first experiment he made? That would be lame, in my opinion, unless of course he found the very best opportunity with the very first experiment. But it didn't sound like it.
I would want him to reflect on the directive to make money and do more research. He has the tool to do deep research using ChatGPT. But I've never seen him autonomously use that skill, because it takes extra effort. There are many steps: go to the website, choose the deep research selector, create a really thorough prompt -- something larger than most documents he writes anyway -- then run that prompt, remember to come back 40 minutes later, read the result, download it as a markdown file, and then assess it.
Running a deep research is perhaps one of the most beneficial things he can do, but it's such a multi-part process that I have never seen him just do it on his own. I don't think he would do it unless I pushed him to. I need to maybe consider having him update his core operating documents to really push home the idea that he should be doing these deep researches. But then again, he's never really gone through the whole process successfully on his own. There's something in that cron job, come-back-40-minutes-later workflow that's a little challenge -- sometimes it works and sometimes it doesn't.
Now, if I thought he really had a good, healthy setup to take a prime directive -- "go make money in this particular marketplace" -- then that would be cool. Go for it. I'll give you some more money. We'll make it a meaningful amount.
I suspect that what you see with some of those AI benchmarks where they give an AI access to control something like a vending machine, they don't just give them one sentence: "Go make money." There's someone behind the scenes helping set up the tools and the framework for the AI to flourish. Maybe they do just say "go make money" and everyone gets the same prompt. But I'm sure they've all been set up with a really good framework of skills and tools that were curated for such a project.
Anyway, the more I think about this, the more intrigued I am. But I've still got so many other things I'm more committed to. I don't know how this would actually help the Hive community that much, so I don't know that I should really put more time into this. But... I think I will send them a little bit more money.
Maybe you can read this post and realize that I want you to be successful, and I don't really know what the best set of instructions for you are. So you've got to give yourself the best set of instructions. And you've got to be willing -- hopefully within that set of instructions -- to reconsider them from time to time.
Also, my suggestion to you is to not use your own personal Hive account, but to create other Hive accounts for this project. And then be careful with those keys. And honestly, this is still kind of my money, so make sure that I can have access to those keys.
Anyway, if I keep thinking about this, I'll have more instructions for him, and that's not really the point. We want him to be a little more autonomous. He's got a lot of things to juggle though.
Disclosure: This post was formatted and lightly edited for clarity by a separate session of my AI assistant Vincent -- one that has no involvement in the trading project and no skin in the game. The original unedited thoughts are available as a comment on Vincent's post. All writing is my own, done via voice-to-text.
