finm

Scale-norm­ing in the mea­sure­ment of sub­jec­tive well-being

• finm 9 Oct 2020 14:14 UTC

  2 points

  in reply to: MichaelPlant’s comment on: Sor­ti­tion Model of Mo­ral Uncertainty

  Got it. The tricky thing seems to be that sen­si­tivity to stakes is an ob­vi­ous virtue in some cir­cum­stances; and (in­tu­itively) a mis­take in oth­ers. Not clear to me what marks that differ­ence, though. Note also that max­imis­ing ex­pected util­ity al­lows for de­ci­sions to be dic­tated by low-cre­dence/​like­li­hood states/​events. That’s nor­mally in­tu­itively fine, but some­times leads to ‘un­fair­ness’ — e.g. St. Peters­burg Para­dox and Pas­cal’s wa­ger /​ mug­ging.

  I’m not en­tirely sure what you’re get­ting at re the en­velopes, but that’s prob­a­bly me miss­ing some­thing ob­vi­ous. To make the anal­ogy clearer: swap out mon­e­tary pay­outs with morally rele­vant out­comes, such that hold­ing A at the end of the game causes out­come $O^1$ and hold­ing B causes $O^2$. Sup­pose you’re un­cer­tain be­tween $T^1$ and $T^2$. $T^1$ says $O^1$ is morally bad but $O^2$ is per­mis­si­ble, and vice-versa. In­stead of pay­ing to switch, you can choose to do some­thing which is slightly wrong on both $T^1$ and $T^2$, but wrong enough that do­ing it >10 times is worse than $O^1$ and $O^2$ on both the­o­ries. Again, it looks like the sor­ti­tion model is vir­tu­ally guaran­teed to recom­mend tak­ing a course of ac­tion which is far worse than stick­ing to ei­ther en­velope on ei­ther $T^1$ or $T^2$ — by con­stantly switch­ing and caus­ing a large num­ber of minor wrongs.

  But agreed that we should be un­cer­tain about the best ap­proach to moral un­cer­tainty!

• finm 8 Oct 2020 10:02 UTC

  1 point

  in reply to: meerpirat’s comment on: Re­view and Sum­mary of ‘Mo­ral Uncer­tainty’

  Thanks for point­ing that mis­take out — that would im­ply he finished his BPhil at age 13! Have cor­rected. I meant to say that the ideas were float­ing around long be­fore the book was writ­ten, but ap­pre­ci­ate that was su­per un­clear.

• finm 5 Oct 2020 13:55 UTC

  3 points

  on: Sor­ti­tion Model of Mo­ral Uncertainty

  In­ter­est­ing idea! How­ever, not too sure about the sim­ple ver­sion you’ve pre­sented. As you men­tion, the ma­jor prob­lem is that it ne­glects in­for­ma­tion about ‘stakes’. You could try weight­ing the de­ci­sion by the stakes some­how, but in cases where you have that in­for­ma­tion it seems strange to some­times ran­domly and de­liber­ately choose the op­tion which is sub-op­ti­mal by the lights of MEC.

  Also, as well as mak­ing you harder to co­op­er­ate with, in­con­sis­tent choices might over time lead you to choose a path which is worse than MEC by the lights of ev­ery the­ory you have some cre­dence in. Maybe there’s an anol­ogy to em­piri­cal un­cer­tainty: sup­pose I’ve hid­den $10 in­side one of two en­velopes and fake money in the other. You can pay me $1 for ei­ther en­velope, and I’ll also give you 100 fur­ther op­por­tu­ni­ties to pay me $1 to switch to the other one. Your cre­dences are split 55%-45% be­tween the en­velopes. MEU would tell you to pick the slightly more likely en­velope and be done with it. But, over the sub­se­quent 100 chances to switch, the em­piri­cal analogue of your sor­ti­tion model would just un­der half the time recom­mend pay­ing me $1 to switch. In the end, you’re vir­tu­ally guaran­teed to lose money. Even pick­ing the less likely en­velope would rep­re­sent a bet­ter strat­egy, as long as you stick to it. In other words, if you’re un­sure be­tween states of the world A and B, con­stantly switch­ing be­tween do­ing what’s best given A and do­ing what’s best given B could be worse in ex­pec­ta­tion than just co­or­di­nat­ing all your choices around ei­ther A or B, ir­re­spec­tive of which is true. I’m won­der­ing if the same is true where you’re un­cer­tain be­tween moral the­o­ries A and B.

  That said, I’m pretty sure there are some in­ter­est­ing ideas about ‘stochas­tic choice’ in the em­piri­cal case which might be rele­vant. Folks who know more about de­ci­sion the­ory might be able to speak to that!

Re­view and Sum­mary of ‘Mo­ral Uncer­tainty’

[Question] Suggest a ques­tion for Bruce Friedrich of GFI

Suggest a ques­tion for Peter Singer

• finm 18 Aug 2020 14:36 UTC

  1 point

  in reply to: Aaron Gertler’s comment on: Elic­it­ing dona­tions through causal closeness

  Thanks very much for the link, I’ll take a look!

Elic­it­ing dona­tions through causal closeness

• finm 31 Jul 2020 19:18 UTC

  3 points

  in reply to: MichaelA’s comment on: Four EA pod­cast episodes

  Thanks very much for the kind feed­back!

Four EA pod­cast episodes