The problem seems to be to be the tradeoff between going deep and going wide, with the added complexity that going deep on the wrong thing seems strictly worse than going wide, and so we’re defaulting to going wide where there’s uncertainty.
Put another way, it’s unlikely that any of those specific skills are going to be particularly important to any of our longest-term goals, but it also seems counterproductive to just sit there thinking about which direction to go in. I’m usually not the biggest expert in the room, but I usually am the most generally competent in terms of being able to fill holes or solve whatever problem crops up, and it’s because I have a habit of just constantly churning and picking up new skills and methods and heuristics wherever I go. I suspect that others would benefit from a similar habit, in particular because once “the right skill” does come along, you have both the affordance to start learning it and a variety of experiences allowing you to learn quickly and efficiently.
That’s a claim. Not necessarily supported, but reasonable, I think, and worth trying out.
I note that I disagree that it’s easy to break averages in all of these things at once. People who don’t actually check their abilities against a standard tend to be wildly overconfident, and people tend to underestimate how long it will take them to learn X or accomplish Y; these things are solidly documented. And while competence does tend to cluster (e.g. “G”), so the picture’s not quite as bleak as the second half of this sentence, once you’ve got a dozen different domains and shooting to be above the 50% mark in all of them, you’re looking at a person who’s approximating one in four thousand, and when you try to get a whole group to hit that mark, the challenge is pretty real. I wouldn’t be surprised if most people have most of this easy, but I think you’re not fully grokking the difficulty of making everybody baseline competent in all of these domains. For instance, you note that many of these skills require only a few weeks, but I don’t know if you added up all of those weeks, compared them to the time commitment, and noted that they’re all being practiced off-hours and people have their own jobs and lives as well.
It’s a floor, though, not a ceiling—we’re aiming at “world class skill,” we’re just not naively expecting that getting there is going to be easy, and initial expectations are meant to be exceeded.
Various additional points …
The trade skill goal got scaled back in response to another comment; it was the hardest/sketchiest one to begin with.
We will have some ability to practice trade skills at the house, and are adopting a norm of going and seeking professional instruction outside from time to time.
I buy that you meet a large number of these criteria; I meet most of them myself. But the ones I don’t have are sticky/tricky.
. And while competence does tend to cluster (e.g. “G”), so the picture’s not quite as bleak as the second half of this sentence, once you’ve got a dozen different domains and shooting to be above the 50% mark in all of them, you’re looking at a person who’s approximating one in four thousand,
I don’t think these skills are anywhere near independent. It’s also not obvious that they’re normally distributed. And, being above the 50% mark in a dozen skills by coincidence being unlikely does not at all tell you how hard it is to gain skills if you put in some deliberate work.
I generally am sympathetic to the argument that stuff can be harder than one assumes, but I also am generally cynical about the “average” level of most of these skills. Most people probably don’t even know what “calibration” means precisely enough to test their own level of calibration. I’m not trying to be arrogant here, I pretty much have only heard about the idea of writing down your confidence level of a bunch of predictions and seeing what comes true from the rationalist community and rationalist-adjacent ones.
For the sake of avoiding this issue, and because rather than using terms like “above-average,” I would attempt to pin down ahead of time requirements that are as specific as possible to measure progress in each of the areas you care about.
For instance, you note that many of these skills require only a few weeks, but I don’t know if you added up all of those weeks, compared them to the time commitment, and noted that they’re all being practiced off-hours and people have their own jobs and lives as well.
I don’t think it should take a few weeks each to exceed average in most of these skills. I expect it to take a few weeks total (or 1 day a week for a few months).
I’m plausibly interested in betting a few hundred dollars against you, especially if (as seems likely, given your confidence) you were to bet $1000 against my $250 or something like that. If I imagine the hundred closest people I know uttering the above, I think all but one or two of them are wrong/overconfident.
What statement, specifically, would we be betting on? It’s certainly plausible that I’m underestimating the difficulty in getting an entire group to above these standards in comparison to getting one person. Though, I think the main issue may be a difference in what we perceive as average, rather than a model of how hard learning these skills is.
I spent five minutes trying to operationalize, but I couldn’t come up with anything that seemed workable. For now, we’ll just proceed knowing that at least one of us is wrong. =)
Either way is fine with me, but if you can express in any way what you think “average” is for some of these skills, I would like to know because now I’m really curious.
Thanks for taking so much time to keep responding to a fairly random commenter!
The problem seems to be to be the tradeoff between going deep and going wide, with the added complexity that going deep on the wrong thing seems strictly worse than going wide, and so we’re defaulting to going wide where there’s uncertainty.
Put another way, it’s unlikely that any of those specific skills are going to be particularly important to any of our longest-term goals, but it also seems counterproductive to just sit there thinking about which direction to go in. I’m usually not the biggest expert in the room, but I usually am the most generally competent in terms of being able to fill holes or solve whatever problem crops up, and it’s because I have a habit of just constantly churning and picking up new skills and methods and heuristics wherever I go. I suspect that others would benefit from a similar habit, in particular because once “the right skill” does come along, you have both the affordance to start learning it and a variety of experiences allowing you to learn quickly and efficiently.
That’s a claim. Not necessarily supported, but reasonable, I think, and worth trying out.
I note that I disagree that it’s easy to break averages in all of these things at once. People who don’t actually check their abilities against a standard tend to be wildly overconfident, and people tend to underestimate how long it will take them to learn X or accomplish Y; these things are solidly documented. And while competence does tend to cluster (e.g. “G”), so the picture’s not quite as bleak as the second half of this sentence, once you’ve got a dozen different domains and shooting to be above the 50% mark in all of them, you’re looking at a person who’s approximating one in four thousand, and when you try to get a whole group to hit that mark, the challenge is pretty real. I wouldn’t be surprised if most people have most of this easy, but I think you’re not fully grokking the difficulty of making everybody baseline competent in all of these domains. For instance, you note that many of these skills require only a few weeks, but I don’t know if you added up all of those weeks, compared them to the time commitment, and noted that they’re all being practiced off-hours and people have their own jobs and lives as well.
It’s a floor, though, not a ceiling—we’re aiming at “world class skill,” we’re just not naively expecting that getting there is going to be easy, and initial expectations are meant to be exceeded.
Various additional points …
The trade skill goal got scaled back in response to another comment; it was the hardest/sketchiest one to begin with.
We will have some ability to practice trade skills at the house, and are adopting a norm of going and seeking professional instruction outside from time to time.
I buy that you meet a large number of these criteria; I meet most of them myself. But the ones I don’t have are sticky/tricky.
I don’t think these skills are anywhere near independent. It’s also not obvious that they’re normally distributed. And, being above the 50% mark in a dozen skills by coincidence being unlikely does not at all tell you how hard it is to gain skills if you put in some deliberate work.
I generally am sympathetic to the argument that stuff can be harder than one assumes, but I also am generally cynical about the “average” level of most of these skills. Most people probably don’t even know what “calibration” means precisely enough to test their own level of calibration. I’m not trying to be arrogant here, I pretty much have only heard about the idea of writing down your confidence level of a bunch of predictions and seeing what comes true from the rationalist community and rationalist-adjacent ones.
For the sake of avoiding this issue, and because rather than using terms like “above-average,” I would attempt to pin down ahead of time requirements that are as specific as possible to measure progress in each of the areas you care about.
I don’t think it should take a few weeks each to exceed average in most of these skills. I expect it to take a few weeks total (or 1 day a week for a few months).
I’m plausibly interested in betting a few hundred dollars against you, especially if (as seems likely, given your confidence) you were to bet $1000 against my $250 or something like that. If I imagine the hundred closest people I know uttering the above, I think all but one or two of them are wrong/overconfident.
What statement, specifically, would we be betting on? It’s certainly plausible that I’m underestimating the difficulty in getting an entire group to above these standards in comparison to getting one person. Though, I think the main issue may be a difference in what we perceive as average, rather than a model of how hard learning these skills is.
I spent five minutes trying to operationalize, but I couldn’t come up with anything that seemed workable. For now, we’ll just proceed knowing that at least one of us is wrong. =)
Either way is fine with me, but if you can express in any way what you think “average” is for some of these skills, I would like to know because now I’m really curious.
Thanks for taking so much time to keep responding to a fairly random commenter!