Douglas_Knight

• Douglas_Knight 6 Jan 2020 4:59 UTC

  19 points

  on: What is Life in an Im­moral Maze?

  in Mo­ral Mazes there are at least 25 (!) lev­els of management

  Ex­po­nen­tial growth makes that im­plau­si­ble.

  In the US mil­i­tary there are 25 ranks, but a hi­er­ar­chy of half that depth with a branch­ing fac­tor of 3. Com­mis­sioned officer ranks cor­re­spond pretty well to the hi­er­ar­chy, but there are only 3 lev­els of en­listed hi­er­ar­chy be­low them.

  You seem to be refer­ring to this pas­sage:

  A weed­ing-out pro­cess takes place among the lower ranks of man­agers dur­ing the first sev­eral years of their ex­pe­rience. The early ca­reers of promis­ing young man­agers are highly variegated; the more promise man­agers show, the more pro­ba­tions they must un­dergo. Take, for ex­am­ple, the case of a young man newly grad­u­ated in 1965 from one of the South’s lead­ing uni­ver­si­ties. He joined Weft Cor­po­ra­tion and spent the next two years in the com­pany’s pro­duc­tion man­age­ment train­ing pro­gram. Then he be­came a first-line su­per­vi­sor on the third shift at a small mill. Shortly there­after, he was pro­moted to the night su­per­in­ten­dent’s job of that mill and given over­all re­spon­si­bil­ity for the night shift. After six months, he be­came a de­part­ment head for weav­ing op­er­a­tions in an­other mill. After an­other six months, he was as­signed to head a larger weav­ing de­part­ment in yet an­other plant. After still an­other six months, he be­came as­sis­tant plant man­ager at a medium-sized mill and kept that job for four years. Then he moved to a still larger mill in the same ca­pac­ity for an­other two years. Then he be­came plant man­ager of a medium-sized mill for two years. Fi­nally, he was named one of two group man­agers with six plant man­agers re­port­ing to him. At the age of 36, he has reached grade 20, the “break­ing point” on a scale of 29, plac­ing him in the top 12.17 per­cent of man­age­ment in Weft Cor­po­ra­tion with, he hopes, a clear shot at be­com­ing vice-pres­i­dent of man­u­fac­tur­ing. Similarly variegated ca­reers are ev­i­dent for young mar­ket­ing and sales man­agers in Weft’s north­ern offices. In Alchemy Inc., whether in sales, mar­ket­ing, man­u­fac­tur­ing, or fi­nance, the “break­ing point” in the hi­er­ar­chy is gen­er­ally thought to be grade 13 out of 25 or the top 8.5 per­cent of man­age­ment.

  The low lev­els of these ranks prob­a­bly provide for recog­ni­tion of non-man­age­ment em­ploy­ees, like the en­listed and war­rant ranks in the US mil­i­tary. With a branch­ing fac­tor of 2, top 12% would mean 3 lev­els up from the bot­tom, not 20. With 9 lev­els above above 20, a to­tal of 13 lev­els of man­agers. The other com­pany, per­haps 15. But prob­a­bly the top of the hi­er­ar­chy is not ac­tu­ally 9 or 12 ranks, but sparser than they sug­gest, not as sparse as the lower ranks, but not com­pletely full like the US mil­i­tary.

• Douglas_Knight 5 Dec 2019 18:43 UTC

  4 points

  on: Could some­one please start a bright home light­ing com­pany?

  What do you want?

  Do you want to buy some­thing for your­self, or do you want a com­pany to change the world?

  Yes, there is room for a bet­ter product, but I think that off-the-shelf prod­ucts are pretty good and you should just get them. If you want to change the world, maybe you should just pro­mote these ex­ist­ing prod­ucts. In par­tic­u­lar, for your short term needs, just do it.

  I think that the right an­swer for most peo­ple and most pur­poses is Rae­mon’s in­struc­tions, $300 for 300 watts, same to­tal wat­tage as coelux. Why did you write this post already know­ing about Rae­mon’s in­struc­tions? What are they lack­ing? That they re­quire in­stal­la­tion? If you have 24 sep­a­rate bulbs spread around the room, in­stal­la­tion is un­avoid­able. Light strips may be a bet­ter solu­tion, but they re­quire even more in­stal­la­tion.

  Some peo­ple want differ­ent things. David Chap­man seems to want to illu­mi­nate his desk, not his room, so he might not like Rae­mon’s setup. If you want to min­i­mize in­stal­la­tion, you might want a sin­gle light. This leads to Ben and Ashen’s sug­ges­tions. They prob­a­bly aren’t as nice as coelux, so, yes, it would be nice if some­one made nicer ver­sions (which should be pos­si­ble). Ashen’s out­door floodlights prob­a­bly have lousy CRI. Ben’s corn­cob isn’t the stan­dard res­i­den­tial fix­ture, and thus re­quired some as­sem­bly. Both prod­ucts prob­a­bly shine out­wards to illu­mi­nate an area, rather than the coelux which is in­tended to mimic the sun through a win­dow push­ing light in a sharp line. This illu­sion is prob­a­bly lux­u­ri­ous, but I’m skep­ti­cal that it is ac­tu­ally good for the goal.

  I was go­ing to fol­low up by say­ing that if you like the form fac­tor of coelux, there are similar prod­ucts on the mar­ket for maybe $2/​watt, only twice as ex­pen­sive as Rae­mon’s setup. They aren’t as bright as coelux, but you could get 5 or 10. There is the sec­ond product Ashen linked or light ther­apy boxes (ap­par­ently 72W) are prob­a­bly a good op­tion with full spec­trum and good lenses. But then I read more I heard a lot of ac­cu­sa­tions of poor qual­ity and fraud around light boxes, so I dunno.

• Douglas_Knight 5 Dec 2019 3:53 UTC

  2 points

  in reply to: gjm's comment on: CO2 Strip­per Post­mortem Thoughts

  You asked for an ex­pert con­sen­sus and I gave it to you. Naval re­searchers are the ex­perts.

  No, “ex­per­i­ments yield re­sults in differ­ent di­rec­tions” is not an ac­cu­rate sum­mary. Ex­per­i­ments with large in­ter­ven­tions trump ex­per­i­ments with small in­ter­ven­tions.

  But, it’s true, I left out the most con­vinc­ing ev­i­dence, which is back of the en­velope calcu­la­tions with gross anatomy.

• Douglas_Knight 4 Dec 2019 22:53 UTC

  5 points

  in reply to: Vaniver's comment on: CO2 Strip­per Post­mortem Thoughts

  When peo­ple say that ven­tila­tion helps them, I be­lieve them. They might even be far on an axis of re­sponse to pol­lu­tion. But how would they know that the par­tic­u­lar pol­lu­tant they re­spond to is CO2? They should be cau­tious in as­sign­ing blame and try­ing spe­cific in­ter­ven­tions. Gw­ern points out that one of the stud­ies that most im­pressed Paul about CO2 ac­tu­ally found larger effects from mold, which is a big prob­lem in the foggy slums of Berkeley. In the­ory there are ways to iso­late hu­man pol­lu­tion from house pol­lu­tion, such as vary­ing the num­ber of room­mates, but I doubt peo­ple are care­ful enough to dis­en­tan­gle that and CO2 isn’t even the only hu­man pol­lu­tant. [Added: but sub­marines are equally sub­ject to all hu­man pol­lu­tants, so that should limit the pos­si­bil­ities to the short list of what they scrub.]

  Are sub­mariners se­lected on that axis? I’m skep­ti­cal. In any event, the naval stud­ies don’t re­strict to sub­mariners.

• Douglas_Knight 4 Dec 2019 17:07 UTC

  2 points

  on: Long-last­ing Effects of Sus­pen­sions?

  It would be nicer if there were more ran­dom­iza­tion, but it would also be nicer if more in­for­ma­tion were ex­tracted from the few peo­ple who are ran­dom­ized. For ex­am­ple, I know some­one who par­ti­ci­pated in an RCT of breast­feed­ing/​for­mula. It was aimed at a spe­cific (acute, ad­verse) in­fant out­come. I’m not sure it even looked at other in­fant met­rics, but it cer­tainly did not have long-term fol­low-up, not even at 5 years. Not only did the study make a big in­vest­ment in per­suad­ing the sub­jects for such lit­tle mea­sure­ment, but it is now im­pos­si­ble to do a bet­ter ex­per­i­ment, be­cause RCTs of breast­feed­ing are now con­sid­ered un­eth­i­cal be­cause of the dam­age their null re­sults do to the au­thors’ ca­reers. (Similarly the Swedish and Aus­tralian twin reg­istries are the right way to do twin stud­ies.)

  On the other hand, some­times you can’t ran­dom­ize and you’d like to know how well you can do cor­re­la­tional stud­ies. If your em­ployer is so en­thu­si­as­tic about ex­per­i­ments, maybe it ap­ply that en­thu­si­asm to it­self and do an ex­per­i­ment to see how well its em­ploy­ees can do ob­ser­va­tional anal­y­sis?

• Douglas_Knight 2 Dec 2019 19:15 UTC

  3 points

  in reply to: gjm's comment on: CO2 Strip­per Post­mortem Thoughts

  Well, who are the ex­perts? Sub­marines rou­tinely have CO2 lev­els much higher than even Berkeley group homes. Naval re­searchers do ex­per­i­ments with higher lev­els still, show­ing lit­tle effect. There seems to be an illeg­ible LW con­sen­sus to the op­po­site, prob­a­bly from peo­ple pre­tend­ing to read this post. Peo­ple praise Gw­ern for his quan­tity, but they don’t ac­tu­ally read him.

  Again, most re­search is about ven­tila­tion and is thus con­founded by other pol­lu­tants. I don’t usu­ally speak up about this be­cause most dis­cus­sion of this doesn’t de­pend on the CO2 hy­poth­e­sis.

• Douglas_Knight 2 Dec 2019 5:09 UTC

  7 points

  in reply to: Donald Hobson's comment on: CO2 Strip­per Post­mortem Thoughts

  Ven­tila­tion has the ad­van­tage that it dumps all pol­lu­tants, not just CO2. In fact, the premise that CO2 af­fects cog­ni­tion is false.

• Douglas_Knight 30 Nov 2019 2:33 UTC

  4 points

  in reply to: maximkazhenkov's comment on: The un­ex­pected difficulty of com­par­ing AlphaS­tar to humans

  This is su­per tan­gen­tial, but I think you’re mak­ing a tech­ni­cal er­ror here. It’s true that poker is im­perfect in­for­ma­tion and it’s true that this makes it re­quire more com­pu­ta­tional re­sources, which matches the main text, but not this com­ment. But does im­perfect in­for­ma­tion sug­gest mixed strate­gies? Does op­ti­mal play in poker re­quire mixed strate­gies? I see this slo­gan re­peated a lot and I’m cu­ri­ous where you learned it. Was it in a tech­ni­cal con­text? Did you en­counter tech­ni­cal jus­tifi­ca­tion for it?

  Games where play­ers move si­mul­ta­neously, like rock-pa­per-scis­sors re­quire mixed strate­gies, and that ap­plies to SC. But I’m not sure that re­quires ex­tra com­pu­ta­tional re­sources. Whether they count as “im­perfect in­for­ma­tion” is sub­ject of con­flict­ing con­ven­tions. Whereas play al­ter­nates in poker. I sus­pect that this meme prop­a­gates be­cause of a spe­cific er­ror. Im­perfect in­for­ma­tion de­mands bluffing and peo­ple widely be­lieve that bluffing is a mixed strat­egy. But it isn’t. The sim­plest ver­sion of poker to in­duce bluffing is von Neu­mann poker, which has a unique (pure) Nash equil­ibrium in which one bets on a good hand or a bad hand and checks on a medium hand. I sus­pect that for poker based on a dis­crete deck that the op­ti­mal strat­egy is mixed, but close to be­ing de­ter­minis­tic and mixed only be­cause of dis­cretiza­tion er­ror.

• Douglas_Knight 29 Nov 2019 2:19 UTC

  2 points

  in reply to: gjm's comment on: Mar­ket Rate Food Is Lux­ury Food

  Maybe this ar­gu­ment is a straw man. That is, maybe it’s not ac­cu­rately de­scribing the ar­gu­ments that peo­ple use. But that is a very differ­ent prob­lem than say­ing this ar­gu­ment might be OK.

• Douglas_Knight 27 Nov 2019 23:30 UTC

  2 points

  in reply to: gjm's comment on: Mar­ket Rate Food Is Lux­ury Food

  If the ar­gu­ments are ac­tu­ally analo­gous, then this shows that one of them is wrong. Maybe there are im­por­tant differ­ences be­tween food and hous­ing, but if the ar­gu­ment doesn’t men­tion them, it is wrong. It’s that sim­ple.

  It is also strik­ing that when peo­ple claim that there are differ­ences and flail around look­ing for differ­ences, the differ­ences gen­er­ally sup­port the wrong side. It makes is pretty clear that they didn’t have any be­lief about the topic.

• Douglas_Knight 27 Nov 2019 15:49 UTC

  2 points

  in reply to: bendini's comment on: Is daily caf­feine con­sump­tion benefi­cial to pro­duc­tivity?

  You are now mak­ing a differ­ent claim than your first com­ment (which was prob­a­bly false and is definitely is con­tra­dicted by pa­pers).

• Douglas_Knight 27 Nov 2019 15:44 UTC

  2 points

  in reply to: Thrasymachus's comment on: Drown­ing chil­dren are rare

  US health out­comes are a suffi­ciently patholog­i­cal out­lier in the world

  I just want to reg­ister dis­agree­ment.

• Douglas_Knight 27 Nov 2019 15:24 UTC

  5 points

  in reply to: Viliam's comment on: Effect of Advertising

  You could pay youtube to buy out the ads. Have you con­sid­ered do­ing so?

  When youtube launched their sub­scrip­tion ser­vice, they now have two cus­tomers and thus di­vided loy­alties. They ad­justed their poli­cies to be more an­noy­ing, to put more pres­sure on the user to sub­scribe. So this is not quite the pure ad­ver­tis­ing ex­am­ple.

• Douglas_Knight 27 Nov 2019 15:15 UTC

  3 points

  on: Effect of Advertising

  I don’t dis­agree with any of the ob­ject-level claims, but I think the fram­ing is con­fused and could be greatly im­proved.

  One way to think about this is, what would the world would be like if we didn’t al­low ad­ver­tis­ing?

  I don’t think that is what you are do­ing in this es­say. In­stead you are propos­ing other meth­ods by way which ad­ver­tis­ing could work. That’s what Kevin does and I think his es­say is bet­ter be­cause he is ex­plicit that this is what he is do­ing. Once you have ex­plic­itly said that ads con­tain in­for­ma­tion, maybe then it is good to talk about the hy­po­thet­i­cal to ex­plain how im­por­tant in­for­ma­tion is. But as­sert­ing your hy­po­thet­i­cal us­ing your model of the world seems to me rhetor­i­cally poor. If you don’t un­der­stand how you dis­agree with other peo­ple, per­haps there is no other ap­proach, but in this case you do know.

  Ask­ing peo­ple to make open-ended in­vest­ment in hy­po­thet­i­cals could be use­ful, but how? If peo­ple have co­her­ent the­o­ries, then they should find it easy to think about hy­po­thet­i­cals with­out chang­ing their minds. If peo­ple have in­co­her­ent the­o­ries, maybe it is use­ful to get them to no­tice that by hav­ing them con­sider hy­po­thet­i­cals. But I don’t think that you’re do­ing that. Also, this seems very difficult, prob­a­bly only vi­able in an in­ter­ac­tive way. If the writer of a static es­say knows ex­actly how the au­di­ence the­o­ries are in­co­her­ent, eg, be­cause they hold two con­tra­dic­tory the­o­ries, then it is prob­a­bly bet­ter to write down the con­tra­dic­tion ex­plic­itly. For an ex­am­ple of this log­i­cal struc­ture, Kevin does this with trick­ing vs Homo eco­nomi­cus. But that’s not the same rhetor­i­cal struc­ture, be­cause his au­di­ence doesn’t ac­tu­ally be­lieve that. (I dub this rhetor­i­cal move the Robin Han­son.)

  I’ve re­cently had sev­eral con­ver­sa­tions around whether ad­ver­tis­ing is harm­ful, and speci­fi­cally whether ads pri­mar­ily work by trick­ing peo­ple into pur­chas­ing things they don’t need.

  It would prob­a­bly be bet­ter to ex­pand on this. There are sev­eral sep­a­rate ques­tions. Ads have two ob­vi­ous costs, the cash to the ad­ver­tiser and the at­ten­tion to the au­di­ence. Why do ad­ver­tisers buy ads? Is it to trick the au­di­ence, or to in­form? That is the topic of the es­say, ex­plic­itly brack­et­ing off of the at­ten­tion cost. But, read­ing the re­sponses, not ex­plicit enough.

• Douglas_Knight 19 Nov 2019 23:12 UTC

  2 points

  in reply to: JohnBuridan's comment on: Pre­dic­tion Mar­kets Don’t Re­veal The Territory

  What do I mean by “pes­simistic about our abil­ity to cre­ate a col­lec­tive map”? Maybe I should not have said “pes­simistic,” but in­stead used “cyn­i­cal.” There are lots of places where we claim to have con­sen­sus and I think that those claims are false. I gave lots of ex­am­ples of very small scale failures to com­mu­ni­cate, like one de­part­ment of a med­i­cal school ly­ing about the work of an­other med­i­cal de­part­ment. If we re­vere cer­tain peo­ple as ex­perts, it be­hooves us to find out what they claim. Find­ing that out would count as pro­mot­ing a col­lec­tive map.

• Douglas_Knight 18 Nov 2019 23:30 UTC

  −1 points

  on: Matthew Walker’s “Why We Sleep” Is Rid­dled with Scien­tific and Fac­tual Errors

  Why do you care about de­bunk­ing this par­tic­u­lar book? If this book in­fluenced your de­ci­sions years ago, you should say so. If not, this prob­a­bly shouldn’t cause you to re­verse course. If you made your de­ci­sion 3 years ago and the book was only pub­lished 2 years ago, prob­a­bly not. Even 1 year ago, it sounded like you hadn’t read it (PS—you owe guzey $50).

  More gen­er­ally, why is this of in­ter­est to this com­mu­nity? Is this book pop­u­lar in this com­mu­nity? I only see it re­ally men­tioned twice on less­wrong.com, that re­view I linked and this from last month. Is it stan­dard in other chan­nels, but no one talks about these top­ics on this site?

• Douglas_Knight 17 Nov 2019 6:04 UTC

  4 points

  in reply to: habryka's comment on: [Site Up­date] Sub­scrip­tions, Book­marks, & Pingbacks

  I went to set­tings and turned on “auto-sub­scribe to replies to my com­ments” (and posts), the check­boxes shown in the image in this post. Since I knew that I wasn’t re­ceiv­ing no­tifi­ca­tions and had just figured out from this post that these set­ting ex­isted, I wasn’t sur­prised to find them off.

  In­ci­den­tally, the link to “User Set­tings” at the top of this post links to less­es­t­wrong. You should prob­a­bly be us­ing rel­a­tive links or some­thing.

• Douglas_Knight 16 Nov 2019 21:16 UTC

  2 points

  in reply to: habryka's comment on: [Site Up­date] Sub­scrip­tions, Book­marks, & Pingbacks

  What do you mean did I “ex­pe­rience them be­ing set pre­vi­ously”? Do you mean that they were pre­vi­ously op­tional? I didn’t know that and, clearly, was not pay­ing at­ten­tion.

  Not only did I used to re­ceive no­tifi­ca­tions for replies to my com­ments and then it stopped, but I still have a record of it. My bell shows the no­tifi­ca­tion from you and then a no­tifi­ca­tion from a month ago, miss­ing a cou­ple replies to re­cent com­ments.

• Douglas_Knight 16 Nov 2019 19:50 UTC

  2 points

  on: [Site Up­date] Sub­scrip­tions, Book­marks, & Pingbacks

  Bug re­port:

  When this rol­led out I was not sub­scribed to my own com­ments and posts. I as­sume that the in­ten­tion was that the de­fault was to auto-sub­scribe to them be­cause (1) that would mimic the old be­hav­ior and (2) that is sug­gested by the screen­shots in this post. But I’m also sur­prised that I’m the first per­son to re­port this. (Maybe Pat­tern, who did not de­scribe it as a bug.) Is it con­sis­tent across peo­ple?

• Douglas_Knight 16 Nov 2019 3:14 UTC

  2 points

  in reply to: jasoncrawford's comment on: In­stant stone (just add wa­ter!)

  Courland (and the many oth­ers who say the same thing) prob­a­bly means that there weren’t ce­ram­ics in the mideast be­fore lime. But there were ce­ram­ics el­se­where.

  What is your tech­ni­cal crite­rion? That it isn’t a pot? Maybe that is what peo­ple mean by “Pot­tery Ne­olithic,” but this seems to me a stupid crite­rion. Any­how, there were ce­ramic pots in many places in East Asia be­fore this. The gen­eral world-wide trend is that ce­ramic pots pre­date lo­cal agri­cul­ture, with the odd ex­cep­tion of the mideast. Here is pot from China, dated there 20-10kya, to­gether with a frag­ment said to have a more pre­cise dat­ing of 20kya. Here is a Ja­panese pot. Here is a Sibe­rian pot­sherd. I think South Amer­i­can pot­tery was pots, but I’m not sure. It’s pretty re­cent, but I think it pre­dated An­dean agri­cul­ture, al­though not Mex­i­can. Afri­can bowls pre­date agri­cul­ture and seem con­tem­po­ra­ne­ous with Çatalhöyük, per­haps Göbekli Tepe. [This is list is sim­ply the sec­ond para­graph of wikipe­dia on pot­tery, which was the ba­sis for my pre­vi­ous claim about “all around the world,” but now I’ve tracked down the in­di­vi­d­ual ex­am­ples.]

  The Venus figurine might be uniquely old, but Croa­tia had a bunch of pa­le­olithic ce­ramic figurines.