(At the time of this comment) 27 karma for a $20k donation, 13 karma for $250, 9 karma for $20 (and a joke) … something’s amiss with the karma-$ currency exchange rate!
Under the assumption that being rewarded with karma can motivate someone to make a donation, but if they make a donation, they do not respond to karma as an incentive when deciding how much to donate, then upvoting any donation is the best policy for maximizing money to SI. I’m not sure how realistic that model is, but it seems intuitive to me.
What do you expect to happen? We don’t have enough users giving karma for donation to sustain a linear exchange rate in the [$20, $20000] range. Unless, I suppose, we give up any attempt at fine resolution over the [$1, $500] range.
In practice, what most people are probably doing is picking a threshold (possibly $0) beyond which they give karma for a donation. This could be improved: you could pick a large threshold beyond which you give 1 karma, and give fractional karma (by flipping a biased coin) below that threshold. However, if the large threshold were anywhere close to $20000, and your fractional karma scales linearly, then you would pretty much never give karma to the other donations.
Edit: after doing some simulations, I’m no longer sure the fractional approach is an improvement. It gives interesting graphs, though!
If we knew the Singularity Institute’s approximate budget, we could fix this by assuming log-utility in money, but this is complicated.
(At the time of this comment) 27 karma for a $20k donation, 13 karma for $250, 9 karma for $20 (and a joke) … something’s amiss with the karma-$ currency exchange rate!
Under the assumption that being rewarded with karma can motivate someone to make a donation, but if they make a donation, they do not respond to karma as an incentive when deciding how much to donate, then upvoting any donation is the best policy for maximizing money to SI. I’m not sure how realistic that model is, but it seems intuitive to me.
It might motivate someone to donate $20 rather than $5 if there is a karma difference; probably not $20000 rather than $20, though.
What do you expect to happen? We don’t have enough users giving karma for donation to sustain a linear exchange rate in the [$20, $20000] range. Unless, I suppose, we give up any attempt at fine resolution over the [$1, $500] range.
In practice, what most people are probably doing is picking a threshold (possibly $0) beyond which they give karma for a donation. This could be improved: you could pick a large threshold beyond which you give 1 karma, and give fractional karma (by flipping a biased coin) below that threshold. However, if the large threshold were anywhere close to $20000, and your fractional karma scales linearly, then you would pretty much never give karma to the other donations.
Edit: after doing some simulations, I’m no longer sure the fractional approach is an improvement. It gives interesting graphs, though!
If we knew the Singularity Institute’s approximate budget, we could fix this by assuming log-utility in money, but this is complicated.
Reversed scope insensitivity?