unearned confidence: Nonlinear Function
Created: January 09, 2021
Modified: January 23, 2022

unearned confidence

This page is from my personal notes, and has not been specifically reviewed for public consumption. It might be incomplete, wrong, outdated, or stupid. Caveat lector.

Confidence is a positive feedback loop. You need a certain amount of confidence to try something hard. Success breeds more confidence, and so on---you update your beliefs about yourself, and also you tend to actually learn new skills in the process of succeeding at something. This means that confidence is something you both learn and earn from experience. But it's also something you have to start with a bit of, or else the feedback loop never engages. You need some unearned confidence.

You'd think that failure would mean you lose confidence, like a TCP sawtooth. But some people are able to frame failure as a temporary setback, as still consistent with the larger plan they're trying to execute. They have a sort of confidence that lets them believe that there's nothing fundamentally wrong with them that needs to be questioned or reexamined. And believing in yourself is an important part of success.

Of course, unearned confidence is also an important part of being an incompetent person at high levels. The existence of people like Trump (not him specifically, but he's the most salient example of this category now), who display confidence that they've never earned, and despite what are objectively speaking a ton of failures, scared me as a young person. I didn't want to become a confident bad person.

  • Even now, I see tons of other people in research who are 'overconfident' in my view. NameRedacted will confidently express opinions that are, in my view, often wrong. How can I take you seriously if you're so uncalibrated as to confidently say wrong things? Or if the things you confidently say are 'not even wrong', but just missing the point. A lot of professional AI researchers seem to me to fall into this category.

Believing that there's nothing fundamentally wrong with yourself is a privilege that not everybody's upbringing gives them. Growing up as a nerd in a non-nerdy world, and as gay in a straight world, I often felt unable to relate to the kids and the culture around me. I didn't know how to be funny or to make friends. I didn't know how to find a relationship or even what a relationship was. It even took me a long time to frame myself as being 'gay' rather than just bad with girls (the important difference is that I wasn't even motivated to try with girls, whereas I might have been motivated to try with boys if it didn't seem like such a big deal and if I knew boys who might have been receptive---but even in a perfectly accepting world, romance is twenty times harder for gay people).

  • Being around nerdy peers from a young age would have been a huge help to me. I'm jealous that SuccessfulFriend had a situation that made clear the necessity of going away to high school, and that he got to spend high school around a bunch of kids he could relate with.

Even now that I know that there was nothing wrong with me, the fact that I grew up thinking there was---or rather, that I grew up without being able to have concrete experiences of success in important areas---creates a problem. it's hard not to learn from experience: even if I believe at an abstract level that there was nothing wrong with me, I never had the concrete experiences of trying and succeeding at relating and relationships. Which means I didn't learn the skills that those experiences teach, and never gained the corresponding confidence.

And even if I think I've learned or am learning those skills now, I've lost a ton of opportunity. I've squandered my chance at grad school. I had an opportunity to join a research community or to at least (SuccessfulFriend-style) get to know the community well enough to feel superior to it, rather than inferior.

Can we distinguish qualities of 'good' unearned confidence from 'bad' unearned confidence?