quanticle

• quanticle 10 Feb 2020 19:28 UTC

  2 points

  in reply to: Chris_Leong's comment on: Eukryt Wrts Blg

  I think there is a happy medium in be­tween hav­ing zero jar­gon (and limit­ing your­self to the style of Sim­ple English Wikipe­dia) and hav­ing so much jar­gon that your ideas are im­pen­e­tra­ble to any­one with­out a Ph.D in the field.

  I would also note that not all jar­gon is cre­ated equal. Some­times a new word is nec­es­sary as short­hand to en­cap­su­late a com­plex topic. How­ever, be­fore we cre­ate the word, we should know what the topic is, and have a short, clear defi­ni­tion for the topic. All too of­ten, I see peo­ple cre­at­ing words for top­ics where there isn’t a short, clear defi­ni­tion. I would ar­gue that jar­gon cre­ated with­out a clear, shared, ex­plicit defi­ni­tion hurts the abil­ity to build com­plex ideas even more so than not hav­ing jar­gon at all. It is only be­cause of this form of jar­gon that we need to have the prac­tice of taboo­ing words.

• quanticle 6 Feb 2020 6:53 UTC

  4 points

  in reply to: Dagon's comment on: Money isn’t real. When you donate money to a char­ity, how does it ac­tu­ally help?

  Per­sonal fa­vors have last­ing im­pact on the re­la­tion­ship, and can­not be ac­counted perfectly so there’s always some resi­d­ual debt (IMO, this is a fea­ture).

  First, have you read Grae­ber’s Debt: The First 5000 Years? It speaks to ex­actly this situ­a­tion and makes points similar to the ones you’re mak­ing.

  Se­cond, the real ad­van­tage of money is that it scales. Ex­chang­ing fa­vors is great when we’re all liv­ing in tribes or small towns. But if I, as many EAs be­lieve, think that the real prob­lem is not that my neigh­bor needs a new jacket but that a per­son 15,000 miles away needs a bed­net to pro­tect them from malaria, it’s not at all clear how I can donate that bed­net by do­ing fa­vors for peo­ple.

  While fa­vors might have a last­ing effect on the re­la­tion­ship, the prob­lem with fa­vors is that they re­quire the re­la­tion­ship to already ex­ist. His­tor­i­cally, this wasn’t a prob­lem be­cause peo­ple lived in small, eco­nom­i­cally iso­lated com­mu­ni­ties, and didn’t move. You got a set of pre-ex­ist­ing re­la­tion­ships sim­ply by virtue of be­ing born into that com­mu­nity, and you were stuck with those, whether you liked it or not. In that en­vi­ron­ment, peo­ple could keep and set­tle tabs, of­ten with­out re­sort­ing to any sort of phys­i­cal cur­rency at all. But now, thanks to global­ized trade net­works and the busi­nesses that built them, we have the abil­ity to make use of far­away re­sources (like prod­ucts made in China) and help peo­ple all over the globe (by donat­ing bed­nets in Africa). In ad­di­tion, the fact that peo­ple have the abil­ity to move means that there needs to be a way for peo­ple to trans­act with oth­ers im­per­son­ally. I can’t rely on trust­ing my IOUs when I’m new in town.

  So while I do think that gift and fa­vor based sys­tems have some ad­van­tages, there is no way I’d trade our mod­ern money-based fi­nan­cial sys­tem for a gift econ­omy.

• quanticle 6 Feb 2020 3:24 UTC

  2 points

  in reply to: shminux's comment on: Money isn’t real. When you donate money to a char­ity, how does it ac­tu­ally help?

  I would al­ter that ever so slightly to say that money is as real as math­e­mat­ics. Math is an un­rea­son­ably effec­tive ab­strac­tion that al­lows us to effi­ciently and com­pre­hen­sively de­scribe a wide range of nat­u­ral phe­nom­ena.

  Similarly, money is an ab­strac­tion that al­lows us to trade a wide range of goods and ser­vices.

• quanticle 5 Feb 2020 15:34 UTC

  5 points

  on: Look­ing for books about soft­ware en­g­ineer­ing as a field

  Another book that might be use­ful is Peter Seibel’s Coders at Work: Reflec­tions on the Craft of Pro­gram­ming. It is a col­lec­tion of in­ter­views with promi­nent soft­ware en­g­ineers (like Jamie Zaw­in­sky, Dou­glas Crock­ford, Joe Arm­strong, Ken Thomp­son, etc) in which they de­scribe how they work and what it feels like (sub­jec­tively) for them to write code.

  The benefit for prac­tic­ing soft­ware en­g­ineers is to read the re­sponses from other pro­gram­mers in or­der to gain the per­spec­tives of ac­com­plished pro­gram­mers on the act of pro­gram­ming. The benefit for you would be to look at how Seibel in­ter­views pro­gram­mers and how he can get them to speak about their ac­com­plish­ments with­out nec­es­sar­ily get­ting too deep into the de­tails of their work.

• quanticle 27 Jan 2020 23:39 UTC

  5 points

  in reply to: Richard_Kennaway's comment on: Tech­nol­ogy Changes Constraints

  Com­pos­ing is slower, true, but com­pos­ing a page for a print­ing press isn’t re­ally com­pa­rable to typ­ing on a type­writer. The cost of com­po­si­tion on a press is amor­tized across hun­dreds or thou­sands of pages printed, which isn’t the case for typ­ing. In fact, we know that com­po­si­tion was worth the cost be­cause the Chi­nese and the Ja­panese even­tu­ally did adopt the Guten­berg-style print­ing press in the mid-to-late 1800s as they in­dus­tri­al­ized and opened up to Western tech­nolo­gies.

• quanticle 27 Jan 2020 6:47 UTC

  6 points

  in reply to: Pattern's comment on: Tech­nol­ogy Changes Constraints

  That’s true, but it’s not re­ally the limit­ing fac­tor. If you have lead type, you can make your­self a set of Chi­nese char­ac­ters al­most as eas­ily as you can make your­self a set of Latin char­ac­ters. The limit­ing fac­tor is the fact that porce­lain type was a lot less durable in a press and needed more time and skil­led la­bor to make, whereas lead type can be made en-masse by met­al­smiths.

• quanticle 26 Jan 2020 22:24 UTC

  7 points

  on: Tech­nol­ogy Changes Constraints

  It’s illus­tra­tive to take your con­straint the­ory and ap­ply it to the press it­self. Pi Cheng’s press used porce­lain type, which re­quired skil­led la­bor to man­u­fac­ture and wore out quickly. Guten­berg’s in­no­va­tion was to re­place porce­lain (or wood, which had been used in pro­to­type print­ing presses in Europe) with lead. Build­ing on this, Guten­berg in­vented an in­ge­nious sys­tem of molds that al­lowed a sin­gle met­al­smith to cast a large num­ber of type blocks at once.

  Guten­berg’s in­no­va­tions (re­plac­ing porce­lain with lead and his in­ven­tion of a sys­tem to mass-pro­duce type blocks) re­laxed an­other con­straint on the print­ing press: the availa­bil­ity of let­ters. Th­ese in­no­va­tions were so suc­cess­ful, print­ing con­tinued to use vari­a­tions of lead type all the way into the early ’80s, when cast lead was fi­nally re­placed with elec­tronic sys­tems.

• quanticle 26 Jan 2020 9:03 UTC

  6 points

  in reply to: jessicata's comment on: On hid­ing the source of knowledge

  But these checks are in­suffi­cient to con­vince a skep­ti­cal au­di­ence, is the point.

  Yes, I see that as a fea­ture, whereas you see to see it as some­what of a bug. Given our propen­sity for self-de­cep­tion and the limits of our brains, we should gather ev­i­dence, even when our in­tu­ition is very strong, and we should be sus­pi­cious of oth­ers who have strong in­tu­itions, but don’t seem to have any sort of an­a­lyt­i­cal ev­i­dence to back their claims up.

  I don’t see any risk to hid­ing the ori­gins of one’s ideas, if one has ex­per­i­men­tal ev­i­dence con­firm­ing them. Similarly, I don’t see the benefit of dis­clos­ing the sources of un­con­firmed ideas. Where the idea comes from (a dream, an in­tu­itive leap, an LSD trip, a rea­soned in­fer­ence from a liter­a­ture re­view) is far less im­por­tant than ac­tu­ally do­ing the work to con­firm or dis­prove the idea.

• quanticle 26 Jan 2020 8:14 UTC

  1 point

  in reply to: jessicata's comment on: On hid­ing the source of knowledge

  Yes, this is a se­man­tic is­sue of what counts as “check­ing”, but that is ex­actly the is­sue at hand. Of course it’s pos­si­ble to check claims against mem­ory, in­tu­ition, men­tal calcu­la­tion, the In­ter­net, etc, but ev­ery such check has only limited re­li­a­bil­ity.

  That is cor­rect, but as Isaac Asi­mov pointed out in The Rel­a­tivity of Wrong, there is a big differ­ence be­tween say­ing, “Every such check has limited re­li­a­bil­ity,” and “Check­ing is the the same as not check­ing.” If some­one came to me to­mor­row and said, “You’re com­pletely wrong, quan­ti­cle, in fact Aus­tralia has a larger land mass than Asia,” I would be skep­ti­cal, and I would point out the mas­sive pre­pon­der­ance of ev­i­dence in my fa­vor. But if they man­aged to pro­duce the ex­traor­di­nary ev­i­dence re­quired for me to up­date my be­liefs, I would. How­ever, they would have to ac­tu­ally pro­duce that ev­i­dence. Sim­ply say­ing, “I in­tu­itively be­lieve it to be true with high prob­a­bil­ity,” is not ev­i­dence.

• quanticle 26 Jan 2020 7:25 UTC

  4 points

  in reply to: jessicata's comment on: On hid­ing the source of knowledge

  Ein­stein’s Ar­ro­gance isn’t as much of a coun­ter­point as you think it is. Yes, Ein­stein was ar­ro­gant, but we only re­mem­ber his ar­ro­gance be­cause he was right. Would we be hold­ing Ein­stein’s ar­ro­gance as such a good thing if Ed­ding­ton’s ex­pe­di­tion had dis­con­firmed Gen­eral Rel­a­tivity? What if the or­bital anoma­lies of Mer­cury had been ex­plained by the pres­ence of an­other planet even closer to the Sun? Without the nu­mer­ous ex­per­i­ments con­firm­ing Gen­eral Rel­a­tivity, Ein­stein would be just an­other kook, with a set of pa­pers that had in­ter­est­ing math­e­mat­ics, per­haps, but whose hy­pothe­ses were re­futed by ob­ser­va­tion.

  As far as Blind Em­piri­cism goes, I do find it tel­ling that Ja­pan did try the solu­tions that Eliezer pro­posed. How­ever, due to fac­tors that Eliezer did not con­sider, the Ja­panese gov­ern­ment was not able to go as far with those solu­tions as Eliezer pre­dicted, and as a re­sult, the perfor­mance of the Ja­panese econ­omy has re­mained lag­gardly. So per­haps Eliezer’s con­fi­dence in his abil­ity to figure out macroe­co­nomics from first prin­ci­ples isn’t as as great as he thought it was, and more em­piri­cism is re­quired.

  Fi­nally, with re­gards to The Sin of Un­der­con­fi­dence, while I agree that un­der­con­fi­dence leads one to pass up op­por­tu­ni­ties that one might have oth­er­wise taken, I would ar­gue that over­con­fi­dence is much worse. As Eliezer also stated:

  One of chief pieces of ad­vice I give to as­piring ra­tio­nal­ists is “Don’t try to be clever.” And, “Listen to those quiet, nag­ging doubts.” If you don’t know, you don’t know what you don’t know, you don’t know how much you don’t know, and you don’t know how much you needed to know.

  There is no sec­ond-or­der ra­tio­nal­ity. There is only a blind leap into what may or may not be a flam­ing lava pit. Once you know, it will be too late for blind­ness.

  While it’s cer­tainly pos­si­ble to be very con­fi­dent in em­piri­cal claims with­out hav­ing checked, I don’t think it’s cor­rect to do so. I am very con­fi­dent that Aus­tralia has a smaller land mass than Asia, but the rea­son I am so con­fi­dent is be­cause I have re­peat­edly seen maps and at­lases that show me that fact. If I did not, I would be as con­fi­dent of my an­swer as I would be of my an­swer to the ques­tion, “Which has the greater land­mass, the Bri­tish Isles or the Ja­panese Home Is­lands?” Similarly, I have ob­served nu­mer­ous flies and a few mice, and thus I can claim that the av­er­age fly is smaller than the av­er­age mouse. If I had not, I would be much less con­fi­dent of my an­swer, much like I have lit­tle con­fi­dence in my an­swer to, “Which is larger? The av­er­age spi­der or the av­er­age fly?” Fi­nally, I have ab­solutely no con­fi­dence in my in­tu­itive an­swer to, “Am I go­ing to get hit by a car when I cross the street?” This is why I look both ways be­fore step­ping out into the road, even when it’s a “quiet” street. As some­one who goes long-dis­tance run­ning, I have had enough un­pleas­ant sur­prises there that I dou­ble and some­times triple check be­fore step­ping out. Do you mean to sug­gest that you step out into roads with­out look­ing both ways?

• quanticle 26 Jan 2020 6:34 UTC

  5 points

  in reply to: jessicata's comment on: On hid­ing the source of knowledge

  I was fairly con­fi­dent of my an­swer, but I still used a calcu­la­tor to dou­ble-check my math. More­over, I have com­puted 15*5 in the past, and I was able to check against the mem­ory of those com­pu­ta­tions to en­sure that I had the cor­rect an­swer. Fi­nally, math is not a sci­ence. Con­fi­dence ap­plies to sci­en­tific re­sults, which rely on ex­per­i­men­tal and ob­ser­va­tional ev­i­dence about the world to sup­port or op­pose spe­cific hy­pothe­ses. The an­swer to 15*5 is not a hy­poth­e­sis. It is a fact, which can be proven to be cor­rect in the con­text of a par­tic­u­lar math­e­mat­i­cal sys­tem.

  A sci­en­tific hy­poth­e­sis, like the struc­ture of ben­zene, is not re­li­ant upon log­i­cal proof in the same way that a math­e­mat­i­cal re­sult is. If I have a proof of the an­swer to 15 * 5, then I know it is cor­rect, ab­sent any er­rors in my proof. How­ever, if I have a par­tic­u­lar hy­poth­e­sis re­gard­ing the struc­ture of ben­zene, or the na­ture of grav­ity, the log­i­cal sound­ness of the ar­gu­ment in fa­vor of my hy­poth­e­sis offers no ev­i­dence as to the ar­gu­ment’s cor­rect­ness. Only ev­i­dence can en­tan­gle my log­i­cal ar­gu­ment with the state of the world, and al­low me to use my log­i­cal ar­gu­ment to make pre­dic­tions.

• quanticle 26 Jan 2020 6:13 UTC

  2 points

  in reply to: jessicata's comment on: On hid­ing the source of knowledge

  Couldn’t it also be the case that the claim is already known through in­tu­ition, and prov­ing it is the main prob­lem? Of course, check­ing against more things will pro­duce higher con­fi­dence, but con­fi­dence can still ex­ceed 99 per­cent with­out do­ing other checks

  How can some­thing have any con­fi­dence be­hind it, much less greater than 99%, with­out ev­i­dence? 99% con­fi­dence on the ba­sis of in­tu­ition alone is re­li­gion, not ra­tio­nal­ity.

  What links here?

  • jessicata's comment on High-pre­ci­sion claims may be re­futed with­out be­ing re­placed with other high-pre­ci­sion claims by jessicata (31 Jan 2020 20:48 UTC; 6 points)

• quanticle 26 Jan 2020 5:18 UTC

  11 points

  on: On hid­ing the source of knowledge

  I’m not sure I agree that the in­tu­ition it­self is an ar­ti­fact of value. To use a con­crete ex­am­ple: Kekulé con­ceived of the struc­ture of ben­zene af­ter hav­ing a dream where he saw an ouroboros. But what does that give us as a way of fur­ther in­ves­ti­ga­tion? Should we ask chemists to take mela­tonin for more vivid dreams?

  I re­call a con­ver­sa­tion I had where some­one (call them A) com­mented that some other per­son (call them B) had de­vel­oped some ideas, then af­ter­wards found aca­demic sources agree­ing with these ideas (or at least, seem­ing com­pat­i­ble), and cited these as sources in the blog post write-ups of these ideas. Per­son A be­lieved that this was im­por­tantly bad in that it hides where the ac­tual ideas came from, and as­signed credit for them to a sys­tem that did not ac­tu­ally pro­duce the ideas.

  Per­son A is cor­rect that this is im­por­tantly bad, but in­cor­rect as to the rea­son. The rea­son this is bad is be­cause it is in­dica­tive of bot­tom-line think­ing. The prob­lem isn’t as­sign­ing credit to a sys­tem that didn’t ac­tu­ally pro­duce the ideas. The prob­lem is se­lec­tively scan­ning for con­fir­ma­tory ev­i­dence and dis­card­ing con­tra­dic­tory ev­i­dence be­cause one is so wed­ded to their in­tu­ition that they can’t ac­cept that their in­tu­ition might be wrong.

  How­ever, if Per­son B did the re­search, and sur­veyed all the ev­i­dence (in­clud­ing the ev­i­dence that dis­agreed with their in­tu­ition) and came to the con­clu­sion that their in­tu­ition was cor­rect, then I don’t see what the prob­lem is in say­ing, “I did a sur­vey of the ev­i­dence for X, and I came to the con­clu­sion that X is prob­a­bly true.” If any­one asks why you were in­ves­ti­gat­ing X in the first place, you can share that you had an in­tu­ition or a hunch. But at that point, the fact that you got the idea to study X from an in­tu­ition or a hunch no longer de­tracts from your ev­i­dence that X is true.

• quanticle 21 Jan 2020 1:56 UTC

  8 points

  in reply to: Davis_Kingsley's comment on: Why Do You Keep Hav­ing This Prob­lem?

  Oh, yes, that’s a good point, and I have up­dated in your di­rec­tion. I was think­ing more along the lines of things like product re­views and sur­vey feed­back, where the user is much more likely to take the time to com­plete the feed­back form if they’ve had a nega­tive ex­pe­rience than if they’ve had a pos­i­tive one.

  Edited to add: I won­der if there’s a dis­tinc­tion be­tween un­so­lic­ited feed­back and re­quest­ing feed­back, or giv­ing feed­back to an in­di­vi­d­ual vs. feed­back to a cor­po­rate en­tity.

• quanticle 20 Jan 2020 17:54 UTC

  5 points

  in reply to: Davis_Kingsley's comment on: Why Do You Keep Hav­ing This Prob­lem?

  I think what tends to be voiced vs. not voiced varies a lot based on both the field and the cul­ture in­volved. I’ve been in some en­vi­ron­ments where it seems like ev­ery­one loves to com­plain even when things are fine, but I’ve also been in some where peo­ple are very ret­i­cent to speak up even when there’s a prob­lem.

  I’m not sure that’s at odds with what mscottveach is say­ing. To put it in differ­ent words, while the amount of feed­back might vary, I don’t think the ra­tio of pos­i­tive vs. nega­tive feed­back varies. It’s the very rare situ­a­tion where the num­ber of mes­sages that say, “This was good, ev­ery­thing went as planned or in­tended,” out­num­bers the mes­sages that talk about how some­thing went wrong.

• quanticle 19 Jan 2020 4:45 UTC

  5 points

  in reply to: AlanXun's comment on: Steel­man­ning Divination

  Thank you for this com­ment. I was try­ing to point to some­thing similar in this com­ment I wrote about the im­por­tance of un­der­stand­ing the ori­gins of your ideas, but you’ve stated the point, in my opinion, much bet­ter than I did.

• quanticle 18 Jan 2020 3:42 UTC

  16 points

  on: In defense of deviousness

  You might be in­ter­ested in Fred Brooks’ sem­i­nal es­say, No Silver Bul­let—Essence and Ac­ci­dent in Soft­ware Eng­ineer­ing. In it, he dis­t­in­guishes be­tween es­sen­tial com­plex­ity and ac­ci­den­tal com­plex­ity. Essen­tial com­plex­ity is com­plex­ity that comes from the prob­lem do­main. It can­not be fac­tored out of the pro­gram, and any at­tempt to do so will likely in­tro­duce bugs. Ac­ci­den­tal com­plex­ity is com­plex­ity that arises from de­tails of the im­ple­men­ta­tion, and which can be sim­plified out of the im­ple­men­ta­tion.

  A good ex­am­ple of ac­ci­den­tal com­plex­ity is mem­ory man­age­ment. A good chunk of pro­gram­mer effort in lan­guages like C and C++ goes to­wards en­sur­ing that mem­ory is man­aged prop­erly, and that the pro­gram re­turns mem­ory to the op­er­at­ing sys­tem when it is finished us­ing it. Me­mory man­aged lan­guages take that bur­den away from the pro­gram­mer and place it ei­ther with the com­piler (in the case of Rust’s bor­row checker) or with the run­time en­vi­ron­ment (in the case of garbage col­lected lan­guages like Java, or Python). The effect of this has been a sig­nifi­cant re­duc­tion in ac­ci­den­tal com­plex­ity (at the cost of some perfor­mance), with a com­men­su­rate in­crease in pro­gram­mer pro­duc­tivity.

• quanticle 13 Jan 2020 7:44 UTC

  19 points

  on: “hu­man con­nec­tion” as col­lab­o­ra­tive epistemics

  Surely there are all kinds of other ways to co­op­er­ate. A friend can help you move your stuff. You can ex­change gifts. You can fend for each other. But ob­jec­tively none of these are worth the huge chunk of re­sources we al­lo­cate to main­tain­ing friend­ships and re­la­tion­ships. Only the up­grades to your wor­ld­view you get from in­ter­act­ing with other peo­ple is worth the trou­ble of in­ter­act­ing.

  I think this is a very nar­row, Straw Vul­can way of view­ing the world. The pri­mary value that other peo­ple have is the in­for­ma­tion and up­dates they can bring to your wor­ld­view? That seems like an awfully nar­row con­cep­tion of the value that other peo­ple can bring.

  More­over, how can you say, “ob­jec­tively” that these benefits are not worth the effort? Do you pre­sume to speak for ev­ery­one’s util­ity func­tion here?

• quanticle 11 Jan 2020 5:57 UTC

  2 points

  in reply to: Viliam's comment on: New edi­tion of “Ra­tion­al­ity: From AI to Zom­bies”

  User Said Ach­miz has take the old Ra­tion­al­ity: From AI To Zom­bies text and laid it out in a much more aes­thet­i­cally pleas­ing for­mat at Read The Se­quences. I definitely en­courage you to check it out.

• quanticle 10 Jan 2020 5:06 UTC

  2 points

  in reply to: Raemon's comment on: What is Life in an Im­moral Maze?

  That’s fair, though I do won­der how rep­re­sen­ta­tive 25-per­son-deep re­port­ing chains are. I’ve never worked in a com­pany that had a re­port­ing chain > 8 and my dad works in a com­pany with a re­port­ing chain of 12. 25 seems… in­cred­ibly painful.