jsalvatier

• jsalvatier 14 Apr 2020 18:36 UTC

  1 point

  in reply to: CellBioGuy’s comment on: Coron­avirus: Jus­tified Key In­sights Thread

  Is there a good source for many things we know from the Di­a­mond Princess data? Or even just the num­bers so far from DP? I’m not sure how to find that data.

• jsalvatier 23 Mar 2018 8:45 UTC

  23 points

  on: Nam­ing the Nameless

  Dou­ble crux is hard enough with ar­gu­ments, and here I’m try­ing to ad­vo­cate some­thing like dou­ble-crux­ing aes­thetic prefer­ences, which sounds ab­surdly am­bi­tious. But: imag­ine if we could talk about why things seem beau­tiful and ap­peal­ing, or ugly and un­ap­peal­ing.

  My work is ba­si­cally about this; ex­tract­ing aes­thetic prefer­ences from peo­ple (and S1 based in­side views more gen­er­ally).

  I haven’t done speci­fi­cally artis­tic aes­thet­ics, but most think­ing re­lies heav­ily on aes­thet­ics about which prob­lems are in­ter­est­ing or im­por­tant, ways of be­hav­ing, ways of think­ing about the world, what counts as ‘sim­ple’, etc. If you want to re­solve dis­agree­ments about big things, you’re go­ing to have to wade into aes­thet­ics.

• jsalvatier 19 Mar 2018 18:06 UTC

  15 points

  on: Idea Inoc­u­la­tion + In­fer­en­tial Distance

  Thanks for writ­ing this. In­fer­en­tial dis­tance + in­oc­u­la­tion is a huge prob­lem for trans­mit­ting large bod­ies of un­der­stand­ing in do­mains that pre­vi­ously didn’t look like do­mains to the stu­dent. The stu­dent fre­quently get­ting a smaller ver­sion of the ideas be­fore they can get the full ver­sion and that shuts off fur­ther in­ter­est be­cause they’ve “got it”.

• jsalvatier 5 Dec 2017 3:24 UTC

  11 points

  in reply to: sarahconstantin’s comment on: List of civil­i­sa­tional inadequacy

  >no­body has done stud­ies mea­sur­ing hor­mone lev­els over time and fit­ting a differ­en­tial-equa­tion model of how hor­mones af­fect each other’s levels

  What in the ev­er­lov­ing fuck? That re­ally seems like the first thing you should do. Has that at least been done for the shared hor­mones?

• jsalvatier 10 Nov 2017 17:53 UTC

  11 points

  in reply to: avturchin’s comment on: Liv­ing in an Inad­e­quate World

  The point was to raise nom­i­nal prices in the first place

• jsalvatier 10 Nov 2017 17:52 UTC

  3 points

  in reply to: avturchin’s comment on: Liv­ing in an Inad­e­quate World

  That is not how it works.

• jsalvatier 10 Nov 2017 8:48 UTC

  7 points

  on: Moloch’s Toolbox (2/​2)

  I don’t buy the idea that vot­ers are not the main source of prob­lem and that its vot­ing sys­tems. Vot­ers don’t have good in­cen­tives to have sen­si­ble opinions that go against nat­u­ral prej­u­dices.

  It seems to me that if you found the me­dian poli­ti­cal opinion of peo­ple, you would have much worse poli­cies in a lot of ar­eas. Prob­a­bly some would be bet­ter, but I would be sur­prised if it were many.

• jsalvatier 10 Nov 2017 8:39 UTC

  9 points

  on: Liv­ing in an Inad­e­quate World

  This is great, I love it

• jsalvatier 6 Nov 2017 6:54 UTC

  23 points

  in reply to: bwest’s comment on: Moloch’s Toolbox (1/​2)

  Do you know if they nor­mal­ize for case difficulty? If a hos­pi­tal pa­tients seems like it will get worse out­comes.

• jsalvatier 6 Nov 2017 5:50 UTC

  21 points

  in reply to: MakerOfErrors’s comment on: Moloch’s Toolbox (1/​2)

  I just did this and it was pretty easy! And in fact I de­cided to change the hos­pi­tal I go to by de­fault.

• jsalvatier 29 Oct 2017 0:07 UTC

  27 points

  in reply to: evolution-is-just-a-theorem’s comment on: Inad­e­quacy and Modesty

  Gen­uinely held aus­tar­ity-type ide­olo­gies are pop­u­lar among peo­ple who care a lot about cen­tral bank­ing (pos­si­bly due to the great de­pres­sion?), and I’m guess­ing that’s what hap­pened at the BoJ. It seems to be what hap­pened in the US, which made similar mis­takes, though less badly.

• jsalvatier 24 May 2017 2:42 UTC

  1 point

  in reply to: Daniel_Burfoot’s comment on: Prob­a­bil­is­tic Pro­gram­ming and Bayesian Meth­ods for Hackers

  There’s not that many that I know of. I do think its much more in­tu­itive and lets you build more nu­anced mod­els that are use­ful for so­cial sci­ences. You can fit the ex­act model that you want in­stead of need­ing to fit your case in a pre­ex­ist­ing box. How­ever, I don’t know of too many ex­am­ples where this is hugely prac­ti­cally im­por­tant.

  The lack of ob­vi­ously valuable use cases is part of why I stopped be­ing that in­ter­ested in MCMC, even though I in­vested a lot in it.

  There is one im­por­tant in­dus­trial ap­pli­ca­tion of MCMC: hy­per­pa­ram­e­ter sam­pling in Bayesian op­ti­miza­tion (Gaus­sian Pro­cesses + pri­ors for hy­per pa­ram­e­ters). And the hy­per­pa­ram­e­ter sam­pling does sub­stan­tially im­prove things.

• jsalvatier 23 May 2017 3:45 UTC

  7 points

  on: Prob­a­bil­is­tic Pro­gram­ming and Bayesian Meth­ods for Hackers

  Funny enough, as a di­rect re­sult of read­ing the se­quences, I got su­per ob­sessed with Bayesian stats and that even­tu­ally re­sulted in writ­ing PyMC3 (which is the soft­ware used in the book).

• jsalvatier 14 May 2017 22:57 UTC

  0 points

  in reply to: Raemon’s comment on: Real­ity has a sur­pris­ing amount of detail

  Likewise

• jsalvatier 14 May 2017 8:35 UTC

  1 point

  on: Real­ity has a sur­pris­ing amount of detail

  If you want to see a billion ex­am­ples of de­tails mat­ter­ing, watch any­thing about ship­build­ing by this guy: https://​​www.youtube.com/​​watch?v=jM6R81SiKgA

• jsalvatier 14 May 2017 8:32 UTC

  2 points

  in reply to: RomeoStevens’s comment on: Real­ity has a sur­pris­ing amount of detail

  Great de­scrip­tion. Yes, I think that’s ex­actly why peo­ple are re­luc­tant to see other peo­ple’s points.

• jsalvatier 14 May 2017 6:41 UTC

  2 points

  in reply to: Raemon’s comment on: Real­ity has a sur­pris­ing amount of detail

  Yeah, I wasn’t too spe­cific on that. I do en­dorse the piece that jb55 quotes be­low, but I’m still figur­ing out what to tell peo­ple to do. I’ll hope­fully have more to say in the com­ing months.

• jsalvatier 13 May 2017 20:31 UTC

  3 points

  on: Real­ity has a sur­pris­ing amount of detail

  John Maxwell posted this quote:

  The mys­tery is how a con­cep­tion of the util­ity of out­comes that is vuln­er­a­ble to such ob­vi­ous coun­terex­am­ples sur­vived for so long. I can ex­plain it only by a weak­ness of the schol­arly mind that I have of­ten ob­served in my­self. I call it the­ory-in­duced blind­ness: once you have ac­cepted a the­ory and used it as a tool in your think­ing, it is ex­traor­di­nar­ily difficult to no­tice its flaws. If you come upon an ob­ser­va­tion that does not seem to fit the model, you as­sume that there must be a perfectly good ex­pla­na­tion that you are some­how miss­ing. You give the the­ory the benefit of the doubt, trust­ing the com­mu­nity of ex­perts who have ac­cepted it.

  -- Daniel Kahneman

  What links here?

  • John_Maxwell's comment on LW 2.0 Strate­gic Overview by habryka (17 Sep 2017 2:05 UTC; 2 points)

Real­ity has a sur­pris­ing amount of detail

• jsalvatier 30 Mar 2017 1:48 UTC

  2 points

  in reply to: SquirrelInHell’s comment on: Sub­mis­sion and dom­i­nance among friends

  I want you to come up to me, put your arm around me, ask me how I am and start tel­ling me about the idea you’ve got. Show me you ought to be in charge, be­cause right now I’m a lit­tle lost and you’re not.

  My de­sire is not for some per­ma­nent power struc­ture, but for other peo­ple to some­times and tem­porar­ily take lead­er­ship with the ex­pec­ta­tion that I will prob­a­bly do so in the fu­ture as well. I think one of the most valuable things I do is sit peo­ple down and say ‘look, there’s this prob­lem you have that you don’t see, but I think its fix­able. You’re stuck think­ing of things as X, but ac­tu­ally Y.’ And I wish peo­ple would re­turn the fa­vor more of­ten.

  In ret­ro­spect, I should have way more clear about this.

  What links here?

  • Viliam's comment on Sub­mis­sion and dom­i­nance among friends by jsalvatier (30 Mar 2017 9:38 UTC; 0 points)