ESRogs

• ESRogs 17 Jan 2020 0:11 UTC

  7 points

  in reply to: Ben Pace's comment on: Real­ity-Re­veal­ing and Real­ity-Mask­ing Puzzles

  I should say, these shifts have not been any­thing like an un­miti­gated failure, and I don’t now be­lieve were worth it just be­cause they caused me to be more so­cially con­nected to x-risk things.

  Had a lit­tle trou­ble pars­ing this, es­pe­cially the sec­ond half. Here’s my at­tempted para­phrase:

  I take you to be say­ing that: 1) the shifts that re­sulted from en­gag­ing with x-risk were not all bad, de­spite lead­ing to the di­s­ori­ent­ing events listed above, and 2) in par­tic­u­lar, you think the shifts were (par­tially) benefi­cial for rea­sons other than just that they led you to be more so­cially con­nected to x-risk peo­ple.

  Is that right?

• ESRogs 12 Jan 2020 18:34 UTC

  5 points

  in reply to: Elizabeth's comment on: Cir­cling as Cousin to Rationality

  This seems like use­ful ad­vice for how to en­gage with Cir­cling, etc., but I’m not sure how it re­sponds to what Said wrote in the par­ent com­ment.

  Is the idea that it would be okay if Cir­cling asks the wrong ques­tions when deal­ing with cases of po­ten­tial be­trayal (my quick sum­mary of Said’s point), be­cause Cir­cling is just prac­tice, and in real life you would still han­dle a po­ten­tial be­trayal in the same way?

  But if Cir­cling is just prac­tice, isn’t it im­por­tant what it trains you to do? (And that it not train you to do the wrong things?)

  (FWIW, I don’t share the ob­jec­tion that Said raises in the par­ent com­ment, but my re­sponse would be more like Rae­mon’s here, and not that Cir­cling is just prac­tice.)

• ESRogs 12 Jan 2020 17:15 UTC

  2 points

  in reply to: paulfchristiano's comment on: Of ar­gu­ments and wagers

  If the to­tal will­ing­ness to risk of peo­ple who be­lieve “Judy will be­lieve X on re­flec­tion” is lower than the value of Alice’s time

  Judy’s time?

• ESRogs 12 Jan 2020 4:33 UTC

  10 points

  on: CFAR’s 2019 Fundraiser

  Just donated!

• ESRogs 12 Jan 2020 2:34 UTC

  7 points

  in reply to: ESRogs's comment on: Book re­view: Re­think­ing Consciousness

  I guess my re­quest of philoso­phers (and the rest of us) is this: when you are us­ing an ev­ery day term like “free will” or “con­scious­ness”, please don’t define it to mean one very spe­cific thing that bakes in a bunch of philo­soph­i­cal as­sump­tions. Be­cause then any­one who ques­tions some of those as­sump­tions ends up ar­gu­ing whether the thing ex­ists. Rather than just say­ing it’s a lit­tle differ­ent than we thought be­fore.

  It’d be like if we couldn’t talk about “space” or “time” any­more af­ter Ein­stein. Or if half of us started call­ing our­selves “illu­sion­ists” w.r.t. space or time. They’re not illu­sions! They ex­ist! They’re just a lit­tle differ­ent than we thought be­fore.

  (See also this com­ment, and re­mem­ber that all ab­strac­tions are leaky!)

• ESRogs 12 Jan 2020 2:28 UTC

  7 points

  in reply to: romeostevensit's comment on: Book re­view: Re­think­ing Consciousness

  Semi-re­lat­edly, I’m get­ting frus­trated with the term “illu­sion­ist”. Peo­ple seem to use it in differ­ent ways. Within the last few weeks I listened to the 80k pod­cast with David Chalmers and the Ra­tion­ally Speak­ing pod­cast with “illu­sion­ist” Keith Frank­ish.

  Chalmers seemed to use the term to mean that con­scious­ness was an illu­sion, such that it means we don’t re­ally have con­scious­ness. Which seems very du­bi­ous.

  Frank­ish seemed to use the term to mean that many of the prop­er­ties that other philoso­phers think our con­scious­ness has are illu­sory, but that of course we are con­scious.

  From listen­ing to the lat­ter in­ter­view, it’s not clear to me that Frank­ish (who, ac­cord­ing to Wikipe­dia, is “known for his ‘illu­sion­ist’ stance in the the­ory of con­scious­ness”) be­lieves any­thing differ­ent from the view de­scribed in this post (which I as­sume you’re class­ing as “rep­re­sen­ta­tion­al­ism”).

  Maybe I’m just con­fused. But it seems like lead­ing philoso­phers of to­day still haven’t ab­sorbed the les­son of Wittgen­stein and are still talk­ing past each other with con­fus­ing words.

• ESRogs 12 Jan 2020 2:08 UTC

  LW: 2 AF: 1

  AF

  in reply to: rohinmshah's comment on: [AN #81]: Univer­sal­ity as a po­ten­tial solu­tion to con­cep­tual difficul­ties in in­tent alignment

  Hmm, maybe I’m miss­ing some­thing ba­sic and should just go re-read the origi­nal posts, but I’m con­fused by this state­ment:

  So what we do here is say “be­lief set A is strictly ‘bet­ter’ if this par­tic­u­lar ob­server always trusts be­lief set A over be­lief set B”, and “trust” is defined as “what­ever we think be­lief set A be­lieves is also what we be­lieve”.

  In this, be­lief set A and be­lief set B are analo­gous to A[C] and C (or some c in C), right? If so, then what’s the analogue of “trust… over”?

  If we re­place our be­liefs with A[C]’s, then how is that us trust­ing it “over” c or C? It seems like it’s us trust­ing it, full stop (with­out refer­ence to any other thing that we are trust­ing it more than). No?

• ESRogs 11 Jan 2020 20:39 UTC

  17 points

  on: Vot­ing Phase of 2018 LW Re­view (Dead­line: Sun 19th Jan)

  I voted!

• ESRogs 11 Jan 2020 6:23 UTC

  4 points

  on: Book re­view: Re­think­ing Consciousness

  Now put the two to­gether, and you get an “at­ten­tion schema”, an in­ter­nal model of the ac­tivity of the GNW, which he calls at­ten­tion.

  To clar­ify, he calls “the ac­tivity of the GNW” at­ten­tion, or he calls “an in­ter­nal model of the ac­tivity of the GNW” at­ten­tion?

  My best guess in­ter­pre­ta­tion of what you’re say­ing is that it’s the former, and when you add “an in­ter­nal model of” on the front, that makes it a schema. Am I read­ing that right?

• ESRogs 11 Jan 2020 6:08 UTC

  LW: 2 AF: 1

  AF

  on: [AN #81]: Univer­sal­ity as a po­ten­tial solu­tion to con­cep­tual difficul­ties in in­tent alignment

  Notably, we need to trust A[C] even over our own be­liefs, that is, if A[C] be­lieves some­thing, we dis­card our po­si­tion and adopt A[C]’s be­lief.

  To clar­ify, this is only if we (or the pro­cess that gen­er­ated our be­liefs) fall into class C, right?

• ESRogs 9 Jan 2020 21:04 UTC

  14 points

  in reply to: reallyeli's comment on: Are “su­perfore­cast­ers” a real phe­nomenon?

  I definitely imag­ine look­ing at a graph of ev­ery­one’s perfor­mance on the pre­dic­tions and notic­ing a cluster who are dis­con­tin­u­ously much bet­ter than ev­ery­one else. I would be sur­prised if the au­thors of the piece didn’t imag­ine this as well.

  Some ev­i­dence against this is that they de­scribed it as be­ing a “power law” dis­tri­bu­tion, which is con­tin­u­ous and doesn’t have these kinds of clusters. (It just goes way way up as you move to the right.)

  If you had a power law dis­tri­bu­tion, it would still be ac­cu­rate to say that “a few are bet­ter than most”, even though there isn’t a dis­con­tin­u­ous break any­where.

  EDIT: It seems to me that most things like this fol­low ap­prox­i­mately con­tin­u­ous dis­tri­bu­tions. And so when­ever you hear some­one talk­ing about some­thing like this you should as­sume it’s con­tin­u­ous un­less it’s su­per clear that it’s not (and that should be a sur­pris­ing fact in need of ex­pla­na­tion!). But note that peo­ple will of­ten talk about it in mis­lead­ing ways, be­cause for the sake of dis­cus­sion it’s of­ten sim­pler to talk about it as if there are these dis­crete groups. So just be­cause peo­ple are talk­ing about it as if there are dis­crete groups does not mean they ac­tu­ally think there are dis­crete groups. I think that’s what hap­pened here.

• ESRogs 9 Jan 2020 20:55 UTC

  14 points

  in reply to: reallyeli's comment on: Are “su­perfore­cast­ers” a real phe­nomenon?

  If you graphed “jump height” of the pop­u­la­tion and 2% of the pop­u­la­tion is Su­per­man, there would be a clear dis­con­ti­nu­ity at the higher end.

  But note that the sec­tion you quote from Vox doesn’t say that there’s any dis­con­ti­nu­ity:

  Tet­lock and his col­lab­o­ra­tors have run stud­ies in­volv­ing tens of thou­sands of par­ti­ci­pants and have dis­cov­ered that pre­dic­tion fol­lows a power law dis­tri­bu­tion.

  A power law dis­tri­bu­tion is not a dis­con­ti­nu­ity! Some peo­ple are way way bet­ter than oth­ers. Other peo­ple are merely way bet­ter than oth­ers. And still oth­ers are only bet­ter than oth­ers.

  “Philip Tet­lock dis­cov­ered that 2% of peo­ple are su­perfore­cast­ers.” When I read this sen­tence, it reads to me like “2% of peo­ple are su­per­heroes”

  I think the sen­tence is mis­lead­ing (as per Scott Alexan­der). A bet­ter sen­tence should give the im­pres­sion that, by way of anal­ogy, some bas­ket­ball play­ers are NBA play­ers. They may seem su­per­hu­man in their bas­ket­ball abil­ity com­pared to the Aver­age Joe. And there are a com­bi­na­tion of in­nate traits as well as honed skills that got them there. Th­ese would be in­ter­est­ing to study if you wanted to know how to play bas­ket­ball well. Or if you were putting to­gether a team to play against the Mon­stars.

  But there’s no dis­con­ti­nu­ity. Go­ing down the curve from NBA play­ers, you get to pro­fes­sional play­ers in other leagues, and then to di­vi­sion 1 col­lege play­ers, and then di­vi­sion 2, etc. Some­where af­ter bench warmer on their high school bas­ket­ball team, you get to Aver­age Joe.

  So SSC and Vox are both right. Some peo­ple are way way bet­ter than oth­ers (with a power law-like dis­tri­bu­tion), but there’s no dis­con­ti­nu­ity.

• ESRogs 6 Jan 2020 19:25 UTC

  5 points

  on: Do­minic Cum­mings: “we’re hiring data sci­en­tists, pro­ject man­agers, policy ex­perts, as­sorted weirdos”

  Looks like he’s get­ting some push­back:

  Boris John­son’s chief ad­viser, Do­minic Cum­mings, will not be al­lowed to by­pass White­hall’s usual re­cruit­ment pro­cesses when re­cruit­ing “weirdos” and “mis­fits” for Down­ing Street jobs, No 10 has said.

  Cum­mings has been crit­i­cised by em­ploy­ment lawyers and unions af­ter post­ing a ram­bling 2,900-word blog­post call­ing for peo­ple with “odd skills” to cir­cum­vent the usual rules in ap­ply­ing for jobs as spe­cial ad­visers and offi­cials in gov­ern­ment.

  [...]

  The prime minister’s spokesman in­sisted the post was aimed only at seek­ing “ex­pres­sions of in­ter­est” and that civil ser­vants would still be ap­pointed within the usual tight pro­ce­dures of the civil ser­vice.

• ESRogs 5 Jan 2020 3:25 UTC

  9 points

  in reply to: Viliam's comment on: Do­minic Cum­mings: “we’re hiring data sci­en­tists, pro­ject man­agers, policy ex­perts, as­sorted weirdos”

  also con­fused why their list of achieve­ments con­tains (or con­sists en­tirely of) Brexit

  He also had a hand in keep­ing Bri­tain on the pound in­stead of the euro, back in 1999-2002.

  To me it seems like the origi­nal strat­egy be­hind Brexit refer­en­dum was sim­ply “let’s make a refer­en­dum that will lose, but it will give us power to con­vert any fu­ture com­plaints into poli­ti­cal points by say­ing ‘we told you’”.

  My un­der­stand­ing is that this was David Cameron’s strat­egy. But oth­ers, like Daniel Han­nan and Do­minic Cum­mings, ac­tu­ally wanted the UK out of the EU.

  In Cum­mings’ case, his (stated) rea­son was that he thought the UK gov­ern­ment was in need of ma­jor re­form, and the best odds for re­form­ing it seemed to re­quire first with­draw­ing from the EU. (See the sec­tion in this blog post la­beled, “Why do it?”)

• ESRogs 4 Jan 2020 23:37 UTC

  6 points

  in reply to: Zack_M_Davis's comment on: Meta-dis­cus­sion from “Cir­cling as Cousin to Ra­tion­al­ity”

  Driv­ing away other writ­ers with an­noy­ing­ness also con­strains the flow of in­for­ma­tion. Trade-offs abound!

• ESRogs 4 Jan 2020 19:03 UTC

  11 points

  in reply to: Zack_M_Davis's comment on: Meta-dis­cus­sion from “Cir­cling as Cousin to Ra­tion­al­ity”

  Why should Said be the one to change, though?

  Good ques­tion. When there are con­flicts over norms, it’s not ob­vi­ous how to re­solve them in gen­eral. I sup­pose the easy, though less preferred, solu­tion would be some kind of ap­peal to the will of the ma­jor­ity, or to an au­thor­ity. The harder, but bet­ter, way would be an ap­peal to a deeper set of shared norms. I’m not sure how tractable that is in this case though.

  What hap­pens if you re­frame your re­ac­tion as, “He’s sur­prised, but sur­prise is the mea­sure of a poor hy­poth­e­sis—the fact that he’s so clue­lessly self-cen­tered as to not be able to pre­dict what other peo­ple know means that I should have higher sta­tus”?

  This is in fact of­ten my re­ac­tion. But I will note that nei­ther so­cial at­tacks nor the writ­ings of clue­less self-cen­tered peo­ple are par­tic­u­larly fun to read. (Espe­cially not when it seems to be both.)

  That may be stat­ing it overly harshly. I do think Said is an in­tel­li­gent per­son and of­ten has good points to make. And I find it valuable to learn that oth­ers are get­ting a lot of value from his com­ments.

  The sig­nal to noise (not ex­actly the right term) ra­tio has not seemed par­tic­u­larly fa­vor­able to me though. But per­haps there’s yet some differ­ent re­fram­ing that I could do to be less frus­trated (in ad­di­tion to what­ever changes Said might make).

• ESRogs 4 Jan 2020 9:32 UTC

  30 points

  in reply to: Mark_Friedenbach's comment on: Meta-dis­cus­sion from “Cir­cling as Cousin to Ra­tion­al­ity”

  Said is do­ing some­thing similar, so I see it as a valuable con­tri­bu­tion.

  I ap­pre­ci­ate hear­ing this coun­ter­point.

  I wish there was a way to get the benefit of Said’s pointed ques­tion­ing w/​o read­ers like me be­ing so frus­trated by the style. I sus­pect that rel­a­tively sub­tle tweaks to the style could make a big differ­ence. But I’m not ex­actly sure how to get there from here.

  For now all I can think of is to note that some users, like Wei Dai, ask lots of pointed and clar­ify­ing ques­tions and never pro­voke in me the same kind of frus­tra­tion that many of Said’s com­ments do.

• ESRogs 4 Jan 2020 8:21 UTC

  6 points

  in reply to: Zack_M_Davis's comment on: Meta-dis­cus­sion from “Cir­cling as Cousin to Ra­tion­al­ity”

  while defend­ing Said, who is a su­per valuable commenter

  Just wanted to note that, as a per­son who of­ten finds Said’s style off-putting, I ap­pre­ci­ate read­ing this coun­ter­point from you.

  EDIT: In my ideal world, Said can find a way to still be nit­pick-y and in­sis­tent on pre­ci­sion and rigor in a way that doesn’t frus­trate me (and other read­ers) so much. I am un­for­tu­nately not ex­actly sure how to get to there from here.

• ESRogs 4 Jan 2020 8:02 UTC

  2 points

  in reply to: Mark_Friedenbach's comment on: Meta-dis­cus­sion from “Cir­cling as Cousin to Ra­tion­al­ity”

  I agree, but as I put it in the great-grand­par­ent com­ment:

  I want to clar­ify that ask­ing about the mean­ings of par­tic­u­lar words is not the main thing I’m talk­ing about (even though that was the ex­am­ple at the top of this whole thread).

  It’s more a pat­tern of ex­press­ing sur­prise /​ in­dig­na­tion as a rhetor­i­cal move. Here is an ex­am­ple, where he’s not ask­ing for clar­ifi­ca­tion, but still do­ing the sur­prise /​ in­dig­na­tion thing.

  You might think that com­ment is perfectly fine, and even from my per­spec­tive in any one sin­gle com­ment, it’s of­ten no big deal. But when sur­prised in­dig­na­tion is ba­si­cally your tone in seem­ingly al­most ev­ery thread, I even­tu­ally find it an­noy­ing.

• ESRogs 4 Jan 2020 7:46 UTC

  4 points

  in reply to: FeepingCreature's comment on: Meta-dis­cus­sion from “Cir­cling as Cousin to Ra­tion­al­ity”

  May as well blame the dis­abled for das­tardly forc­ing us to waste money on wheelchair ramps

  I do not be­lieve that Said is un­able to gen­er­ate hy­pothe­ses in all the cases where he ex­presses baf­fle­ment /​ in­dig­na­tion. I be­lieve it is (at least par­tially) a rhetor­i­cal move.

  If peo­ple pre­tended to need wheelchairs to prove a point, we’d be right to blame them for forc­ing us to spend re­sources on them.