I interpret that as referring to one’s justification for making claims of more-than-base-rate-likelihood-for-weird-off-the-wall-suggestions—especially if you’re going to take up a lot of people’s time based on anecdotes, you should at least be presenting decent anecdotes.
If you dismissed or didn’t bring up anecdotes, I imagine there’d also be a positive response. For example, if one argued for using bright lights like thus, I imagine it’d go over well: ‘I found studies X Y and Z where bright lights increased alertness over a day; applying the usual meta-analytical considerations, I guesstimate there’s a 20% chance it’ll work on ordinary people for a payoff of $Z; and on Amazon these lights cost $A and to run for a year uses $B in eletricity; and turns out, 0.20 * Z > $A+$B. Also I have a rubbishy personal anecdote: after doing this calculation, I tried out the lights and it seems to be working.’
(My own opinion is that lights are stupid easy to test so there’s no excuse for not doing a self-experiment. Flip a coin each morning to turn the light on or keep it off, and write down which & how much you think you got done at the end of the day! But if you can’t do that, at least trying to get past the honeymoon period is a start.)
I agree completely, but the voting patterns here made me think LW thought differently.
I interpret that as referring to one’s justification for making claims of more-than-base-rate-likelihood-for-weird-off-the-wall-suggestions—especially if you’re going to take up a lot of people’s time based on anecdotes, you should at least be presenting decent anecdotes.
If you dismissed or didn’t bring up anecdotes, I imagine there’d also be a positive response. For example, if one argued for using bright lights like thus, I imagine it’d go over well: ‘I found studies X Y and Z where bright lights increased alertness over a day; applying the usual meta-analytical considerations, I guesstimate there’s a 20% chance it’ll work on ordinary people for a payoff of $Z; and on Amazon these lights cost $A and to run for a year uses $B in eletricity; and turns out, 0.20 * Z > $A+$B. Also I have a rubbishy personal anecdote: after doing this calculation, I tried out the lights and it seems to be working.’
(My own opinion is that lights are stupid easy to test so there’s no excuse for not doing a self-experiment. Flip a coin each morning to turn the light on or keep it off, and write down which & how much you think you got done at the end of the day! But if you can’t do that, at least trying to get past the honeymoon period is a start.)