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.
“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.
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.
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.