= 27d567bc2d48d9f40e9ccc20ea370b15)
gjm comments on The Zeroth Skillset - Less Wrong Discussion
You are viewing a comment permalink. View the original post to see all comments and the full post content.
= 783df68a0f980790206b9ea87794c5b6)
Loading…= b715d02e85f7b3b4ae65ca3d3e450868)
Subscribe to RSS Feed
Powered by Reddit
= f037147d6e6c911a85753b9abdedda8d)
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (108)
Among the outcomes I care about is having other people not get screwed over. Shunning them because other things have gone badly for them contributes to not achieving such outcomes.
Another outcome I care about is associating with people who are interesting, good company, useful to me, etc. Shunning broadly-defined groups that contain many such people, when there are narrower groups whose shunning would be just as effective, contributes to not achieving such outcomes.
A policy of avoiding dangerous-seeming people seems very reasonable, especially if it's applied flexibly. (One might, e.g., have a close and dearly loved family member who is dangerous, or an important business connection with someone dangerous. Personal safety is good but not the only thing that matters.) I think it's very likely that "seems dangerous to me", fuzzy though it is, is a much more accurate heuristic for identifying dangerous people than "is poor", and that the same is true for most people here.
It's a heuristic, not an algorithm for safety.
"A heuristic, not an algorithm": what difference are you intending to convey? (I wasn't trying to suggest that you think avoiding poor people gives some kind of guarantee, or anything like that. A heuristic is what I took you to be saying it was. For me, at least, a heuristic is a kind of algorithm.)
"Not ... for safety": do you mean that there are other purposes to it besides safety? OK, fair enough (though your presentation of this "skill" here has been all about safety) but I don't think it makes a difference to what I'm saying: safety together with the other things you intend this to achieve are still not the only things that matter, and I gravely doubt that the broad-brush policy of avoiding poor people is a great way of achieving those other things (by comparison with less-simplistic heuristics) -- though on that point I'm prepared to be convinced.
A heuristic is a fuzzy set of principles that are correlated with the outcomes you want. An algorithm is a set of directions that give you the outcome you want. When I say "avoiding poor people is a heuristic" I mean that it is the high level abstraction of a bunch of low level behaviors in various situations.
Something can be an algorithm despite not necessarily giving you exactly the outcome you want. Hence approximation algorithms and probabilistic algorithms.
Edit: the boundaries between algorithms and heuristics are complicated. Colloquial usage referring to heuristics as something like "rules of thumb" and algorithms as "a set of directions" is what was intended.