It’s been about six months since the end of Rationality Mega-Camp. Since then, what’s been going on in my brain?
Well, I did a semester of graduate school, and started a company as well. I did a reasonably passable job on my class and research projects, while building the core of what is now a funded startup. I was quite busy with work, but I even maintained a reasonable social & hobby life, brewing 60 gallons of beer, starting an indoor tomato garden, getting a decent level of exercise, and spending social time with friends.
Blink. Wow. Now that I am stepping back to consider it, I can’t remember a time when I felt even half as effective as I have consistently felt over the last six months. There have been ups and downs, but overall I’ve been both happy and productive – I am pretty damn satisfied with my effectiveness.
The real question I am interested in is: how much of this level of effectiveness was due to RBC / rationality training?
I don’t know how to directly make progress on answering this question, so I’m going to consider what skills from RBC I use often.
The most important one seems to be that RBC taught me to produce better models of the world, and rely on them more. The catchphrase here is “nothing works or fails as if by magic,” but that’s not really very descriptive.
I’ll give an example. Before, when evaluating a business idea, I would run down a mental checklist of “is it a real thing that you could produce? does it have a market? can you get people to hear about it and pay for it?” But that procedure wasn’t very effective at answering the question, because my brain would often answer “yes” or “no” based on some confused process: for example, comparing that strategy against similar strategies that had worked for others in the past, without attempting to understand why they worked or failed.
Now, when I hear the same idea, I don’t start with the checklist. Instead I actually reason about all the causal processes involved: “OK, so you want to build a widget. People want this widget because of X, which is a real problem that I agree people have. You expect them to hear about it because of Y, but I don’t understand how it will spread to more people after that.” Instead of a rote process, I am following a much more analytical process which relies on my models of the world, and I expect those models to be accurate and to lead to correct answers. When I am surprised by a result, I propagate that back to the models, so they are constantly improving.
Before, I was basically doing Cargo Cult Science. This is an essay by Richard Feynman where he describes people going through the motions – they saw someone else achieving something and want the same, so they mimic their actions. My brain was mimicking the actions of successful strategies in the past, without understanding why they worked. Now, I don’t do this anymore.
I’m not sure if this even quite captures the depth of the insight I’m trying to describe. It might not be possible to convey in a single blog post. Your best bet for gaining it might be to read a lot of Robin Hanson and Tyler Cowen.
Anyway, nothing works or fails by magic, and most practically relevant causal processes seem to be grokkable by humans. So it’s a matter of constantly learning and grokking more models. My brain does the learning part automatically, and as soon as it had enough of a framework to build on, it started filling in the holes. So the upshot is that I feel like I am a lot better at predicting the success or failure of various things than I used to be.
RBC did not directly attempt to teach this skill, but it was an attitude that a lot of people in the community had, and I am pretty sure I can credit RBC with teaching it to me – simply by participating in discussions with SI folks and the surrounding community. People who I associate with very high levels of this skill are Michael Vassar, Anna Salamon, and Geoff Anders, all of whom taught at RBC.
Asking for examples! This was pretty much the first thing they taught at RBC and it’s incredibly important to my day-to-day communication. I think of Anna every time I ask for an example.
Metadiscussion and metastrategy: these are things that my fellow campers and instructors loved to talk about. Example – “at the beginning of a brainstorming session, remind people to discuss the problem thoroughly before proposing any solutions”; “during a meeting, regularly ask if the meeting is achieving the goals it was designed for”.
Not trusting my brain
I set timers, I put things on calendars, I write things down, I block myself from the internet. All these things I do regularly and it helps. I admit that my brain is not perfect and that it will do the wrong thing if I don’t take simple steps to correct it. So I take the damn steps. I learned like a hundred ways that brains can go wrong at RBC, and I apply bits of that knowledge regularly.
So did RBC help?
I still don’t know. This post is quite unfinished but I am going to post it and get back to work. I intend to take the other angle later, arguing that despite these skills, RBC is not mainly responsible for my seemingly higher levels of effectiveness. Stay tuned…