I’m somewhat surprised at the notion of “just be trustworthy” being helpful for anyone, though maybe that’s because of an assumption that anyone who doesn’t already employ this tactic must have considered it and have solid reasons to not use it?
an assumption that anyone who doesn’t already employ this tactic must have considered it and have solid reasons to not use it?
I think by default, the main way rationalists become less trustworthy is not on purpose: it comes from a form of naive consequentialism where you do things like make plans with people and then abandon them for better plans without considering the larger effect of this sort of behavior on how much people trust you. One way to say it is that one of the main consequences of you taking an action is to update other people’s models of you and this is a consequence that naive consequentialists typically undervalue.
In my case, I find Romeo’s approximation to be a fairly good descriptor of how I operate. There’s obviously a whole bunch of conditionals, of course, and if I were to try and reduce it down, it might look like:
In general:
1) Talking to someone? Figure out what they’re interested in, and how you can offer resources to help them advance their goals.
2) If there’s tangential overlap, maybe mention your own goals.
3) Iterate back and forth for a while with questions.
If you’re generally interested in learning more about the other person, they tend to reciprocate in nice ways, e.g. giving attention to your own stuff. [I think this is enough of an approximation to “just be naively good”?]
Then there’s a few corollaries:
a) With a casual friend? Intersperse conversation with banter.
b) Talking to someone new? Use generally accepted stereotype phrases (comment on weather, etc.) and then introduce yourself. Maybe start by complimenting something of theirs.
But all of this is fairly black-boxed, and I tend not to operate by explicitly reasoning these things out, i.e. the above rules are a result of my applying introspection / reductionism to what are usually “hidden” rules.
The closest I get to any of the (weird-from-my-perspective) recursive modeling / explicit reasoning is when I have no idea what to say. In such a case, I might ask myself, “What would [insert socially adept friend] do?” which queries my inner simulator of them and often spits out passable suggestions.
‘try being naively good first’ has been a surprisingly useful thought.
I’m somewhat surprised at the notion of “just be trustworthy” being helpful for anyone, though maybe that’s because of an assumption that anyone who doesn’t already employ this tactic must have considered it and have solid reasons to not use it?
I think by default, the main way rationalists become less trustworthy is not on purpose: it comes from a form of naive consequentialism where you do things like make plans with people and then abandon them for better plans without considering the larger effect of this sort of behavior on how much people trust you. One way to say it is that one of the main consequences of you taking an action is to update other people’s models of you and this is a consequence that naive consequentialists typically undervalue.
That seems like a possible selection effect.
In my case, I find Romeo’s approximation to be a fairly good descriptor of how I operate. There’s obviously a whole bunch of conditionals, of course, and if I were to try and reduce it down, it might look like:
In general: 1) Talking to someone? Figure out what they’re interested in, and how you can offer resources to help them advance their goals. 2) If there’s tangential overlap, maybe mention your own goals. 3) Iterate back and forth for a while with questions.
If you’re generally interested in learning more about the other person, they tend to reciprocate in nice ways, e.g. giving attention to your own stuff. [I think this is enough of an approximation to “just be naively good”?]
Then there’s a few corollaries: a) With a casual friend? Intersperse conversation with banter. b) Talking to someone new? Use generally accepted stereotype phrases (comment on weather, etc.) and then introduce yourself. Maybe start by complimenting something of theirs.
But all of this is fairly black-boxed, and I tend not to operate by explicitly reasoning these things out, i.e. the above rules are a result of my applying introspection / reductionism to what are usually “hidden” rules.
The closest I get to any of the (weird-from-my-perspective) recursive modeling / explicit reasoning is when I have no idea what to say. In such a case, I might ask myself, “What would [insert socially adept friend] do?” which queries my inner simulator of them and often spits out passable suggestions.