I left Eve.
After being consumed for so long by a single project it's hard to know what to do next. I'm sure that whatever thoughts I have right now are far too narrow and railroaded compared to the actual possibilities, so what follows is just thinking out loud.
Over the last two years I learned more than I ever expected to know about logic languages, relational databases and progressive systems, to the point of being able to do novel work in the field. Whether or not I use that knowledge again in future, it's the first time I've reached that depth in any field. More importantly, it's the first time I've properly come to grips with the idea that even relatively small investments of time/effort/organization can multiply into massive outcomes when steadily applied. One hour of reading per day for two years adds up to a lot of papers. I'm pretty excited to see what I could do with a more deliberate effort. (EDIT: My first deliberate effort is in progress here.)
I also learned a lot about managing research, mostly after being burned by my repeated mistakes. I've become much more reliant on empirical testing and on trying to falsify ideas early. I learned to keep a project diary to track what I've done, why I did it, how it turned out and how long I spent doing it. I learned to set concrete goals in advance and stop as soon as they are met rather than eagerly jump down every little rabbit hole.
On a parallel track, my programming has been heavily influenced data-oriented and compression-oriented design and by a dawning mechanical sympathy, all of which place the emphasis on solving simply problems with machines, rather than allowing instrumental values to become terminal values. I’ve also become much more willing and able to tear systems apart and read their entrails rather than blindly trusting to hearsay or documentation.
Outside of work, I was surprised to notice how much I read in the last year (largely due to buying a kindle) and how much of it revolved around rationality / epistemology / cognitive science. The combination of reading, experience and occasional exposure to some very effective people made me more epistemically cautious and more of a fox, which lead to some considerable frustration with the Silicon Valley culture.
The end result is that I've become much more confident in my ability to solve important problems, but much less convinced that I have any. Looking back at my career so far, I find that very little of the work I've done has created any lasting value (I hope Eve might be an exception, but it has a long way to go yet). Similarly, much of what I learned from being active in the tech community turned out to be noise, and the ideas that were useful and powerful largely came from people who had real problems to solve and only cared incidentally about the tech.
Inspired in part by Engelbart's system of A) do the work B) improve the doing C) improve the improving, I think the best way for me to pursue value is to switch fields entirely. I don't even have an A right now - I need exposure to fresh problems.
My general interest is still in improving/augmenting human intelligence. I saw and still see promise in the original Eve vision of tackling the end-to-end ergonomics of programming, but there are plenty of other avenues to pursue, technological and otherwise. Most of the improvements I've seen in my own effectiveness in the last few years have been from new mental tools rather than from software or biology.
The obvious choice of field is somewhere in the nexus around cognitive science / decision science / human error / behavioral economics. In addition to being a fresh source of problems, it's something that has captured my interest for years and also one of the more likely paths towards improving intelligence in it's own right.
A reasonable attack would be to move back to London (quitting derailed my US visa) and pick up a part-time programming job to fund part-time study. Going back to school is not the only way to learn, but I think I would benefit from being surrounded by other students.
As for the job, part-time work is hard to find so I may not be able to be too picky. The ideal would be something that is producing immediate value rather than another moonshot. Being London, the most likely case is I end up in a hedge fund and content myself with earning to give. Being a programmer is pretty cushy.
This somewhat resembles a plan. As I said at the beginning though, I'm aware that there are probably good options that I haven't even considered yet. If you have suggestions please don't hesitate to let me know.