I mean, the sum to­tal of spaces that the ra­tio­nal­ist com­mu­nity uses to hold dis­cus­sions, prop­a­gate in­for­ma­tion, do col­lec­tive de­ci­sion mak­ing, (pre­sum­ably) provide mu­tual sup­port et cetera, to the ex­tent these spaces are effec­tive in fulfilling their func­tions. Any­where where I can say some­thing and peo­ple in the com­mu­nity will listen to me, and take this new in­for­ma­tion into ac­count if it’s worth tak­ing into ac­count, or at least provide me with com­pas­sion­ate feed­back even if it’s not.

Re­gard­ing “Kol­mogorov com­plic­ity”, I just want to make clear that I don’t want to cen­sor your opinion on the poli­ti­cal ques­tion. Such cen­sor­ship would only serve to jus­tify your no­tion that “we only re­fuse to be­lieve X be­cause it’s heresy, while any sys­tem­atic truth­seeker would be­lieve X”, which is some­thing I very much dis­agree with. I might be in­ter­ested in dis­cussing the poli­ti­cal ques­tion if we were al­lowed to do it. It is the dou­ble bind of, not be­ing able to al­lowed to ar­gue with you on the poli­ti­cal que­si­ton while hav­ing to listen to you con­stantly hint­ing at it, is what bug­ging me. Then again, I don’t re­ally have a good solu­tion.

What do you mean by ‘the com­mu­nity’s “dis­cus­sion space”’? Are you refer­ring to Less Wrong? Or some­thing else?

Thanks, these are some great points on some of the costs of de­cou­pling norms! (As you’ve ob­served, I’m gen­er­ally pretty strongly in fa­vor of de­cou­pling norms, but policy de­bates should not ap­pear one-sided.)

some­one brings it up all the time

I would want to dis­t­in­guish “brings it up all the time” in the sense of “this user posts about this topic when it’s not rele­vant” (which I agree is bad and war­rants mod­er­a­tor ac­tion) ver­sus the sense of “this user posts about this topic a lot, and not on other top­ics” (which I think is gen­er­ally OK).

If some­one is ob­ses­sively fo­cused on their nar­row spe­cial in­ter­est—let’s say, alge­braic topol­ogy—and oc­ca­sion­ally com­ments speci­fi­cally when they hap­pen to think of an ap­pli­ca­tion of alge­braic topol­ogy to the fo­rum topic, I think that’s fine, be­cause peo­ple read­ing that par­tic­u­lar thread get the benefit of a rele­vant alge­braic topol­ogy ap­pli­ca­tion—even if look­ing at that user’s post­ing his­tory leaves one with an un­set­tling sense of, “Wow, this per­son is creep­ily ob­sessed with their hob­by­horse.”

tries to twist other peo­ple’s posts to­wards a dis­cus­sion of their thing

I agree that this would be bad, but I think it’s usu­ally pos­si­ble to dis­t­in­guish “twist[ing] other peo­ple’s posts to­wards a dis­cus­sion of their thing” from a gen­uinely rele­vant men­tion of the thing that couldn’t (or shouldn’t) be rea­son­ably ex­pected to de­rail the dis­cus­sion?

In the pre­sent case, my great-great-grand­par­ent com­ment notes that the list-of-koans for­mat lends it­self to read­ers con­tribut­ing their own ex­am­ples in the com­ments, and I tried to give two such ex­am­ples (try­ing to mimic the æs­thetic of the OP by con­tin­u­ing the num­bered list and Alice/​Bob/​Char­lie/​&c. char­ac­ter name se­quence), one of which re­lated the theme of the OP to the main point of one of my re­cent posts.

In ret­ro­spect, maybe I should’ve thought more care­fully about how to phrase the pro­posed ex­am­ple in a way that makes the con­nec­tion to the OP more ex­plicit/​ob­vi­ous? (Prob­a­bly-bet­ter ver­sion: “A mean­ingful ‘Yes’ an­swer to the ques­tion ‘Is G an H?’ re­quires a defi­ni­tion of H such that the an­swer could be ‘No’.”)

It’s true that, while com­pos­ing the great-great-grand­par­ent, I was kind of hop­ing that some read­ers would click through the link and read my ear­lier post, which I worked re­ally hard on and which I think is filling in a gap in “A Hu­man’s Guide to Words” that I’ve seen peo­ple be con­fused about. But I don’t see how this can rea­son­ably be con­strued as an at­tempt to de­rail the dis­cus­sion? Like, I or­di­nar­ily wouldn’t ex­pect a brief com­ment of the form “Great post! Here’s a cou­ple more ex­am­ples that oc­curred to me, per­son­ally” to re­ceive any replies in the me­dian case.

(Although un­for­tu­nately, it em­piri­cally looks like the dis­cus­sion did, in fact, get de­railed. I feel bad for Scott G. that we’re clut­ter­ing up his com­ment sec­tion like this, but I can’t think of any­thing I wish I had done differ­ently other than word­ing the great-great grand­par­ent more clearly, as men­tioned in the para­graph-be­fore-last. Given Vanessa’s re­ply, I felt jus­tified in writ­ing my coun­ter­re­ply … and here we are.)

It would be perfectly alright for mod­er­a­tors who didn’t want to drive away their vis­i­tors to ask this per­son to stop.

Agreed, the mod­er­a­tors are God and their will must be obeyed.

kick out some­one who has a bad rep­u­ta­tion that makes im­por­tant posters un­able to post on your web­site be­cause they don’t want to as­so­ci­ate with that per­son, even IF that per­son has good behavior

So, the dy­namic you de­scribe here definitely ex­ists, but I ac­tu­ally think it’s a pretty se­ri­ous prob­lem for our col­lec­tive san­ity: if some truths hap­pen to lie out­side of So­ciety’s Over­ton win­dow, then sys­tem­atic truth­seek­ers (who want to col­lect all the truths, not just the ma­jor­ity of them that are safely within the Over­ton win­dow) will find them­selves on the wrong side of Re­spectabil­ity, and if peo­ple who care about be­ing Re­spectable (and thereby hav­ing power in So­ciety) can’t even talk to peo­ple out­side the Over­ton win­dow (not even agree with—just talk to, us­ing, for ex­am­ple, a web­site), then that could have nega­tive in­stru­men­tal con­se­quences in the form of peo­ple with power in So­ciety mak­ing bad policy de­ci­sions on ac­count of hav­ing in­ac­cu­rate be­liefs.

I want to write more about this in the fu­ture (albeit not on Less Wrong), but in the mean­time, maybe see the im­mor­tal Scott Alexan­der’s “Kol­mogorov Com­plic­ity And The Parable Of Light­ning” for an ex­pres­sion of similar con­cerns:

Some other be­liefs will be found to cor­re­late heav­ily with light­ning-heresy. Maybe athe­ists are more of­ten light­ning-heretics; maybe be­liev­ers in global warm­ing are too. The en­e­mies of these groups will have a new cud­gel to beat them with, “If you be­liev­ers in global warm­ing are so smart and sci­en­tific, how come so many of you be­lieve in light­ning, huh?” Even the savvy Kol­mogorovs within the global warm­ing com­mu­nity will be forced to ad­mit that their the­ory just seems to at­tract uniquely crappy peo­ple. It won’t be very con­vinc­ing. Any po­si­tion cor­re­lated with be­ing truth-seek­ing and in­tel­li­gent will be always on the re­treat, hav­ing to for­ever apol­o­gize that so many mem­bers of their move­ment screw up the light­ning ques­tion so badly.

Alright, let’s sup­pose it’s off-topic in this thread, or even on this fo­rum. But is there an­other place within the com­mu­nity’s “dis­cus­sion space” where it is on-topic? Or you don’t think such a place should ex­ist at all?

Ugh, be­cause pro­duc­tive dis­cus­sion hap­pens be­tween perfectly dis­pas­sion­ate robots in a vac­uum, and if I’m not one then it is my fault and I should be ashamed?

As dis­cussed in the linked thread—it is none of my busi­ness, nor the busi­ness of any of your in­ter­locu­tors, whether you are, or are not, a “perfectly dis­pas­sion­ate robot in a vac­uum”, when it comes to dis­cus­sions on sub­jects like the OP. That is not some­thing which should en­ter into the dis­cus­sion at all; it is sim­ply off-topic.

If we per­mit the in­tro­duc­tion of such ques­tions as whether you feel un­com­fortable (about the topic, or any on-topic claims) into dis­cus­sions of ab­stract episte­mol­ogy, or Bayesian in­fer­ence, or logic, etc., when that dis­com­fort in no way bears on the truth or falsity of the claims un­der dis­cus­sion, then we might as well close up shop, be­cause at that point, we have bid good-bye even to the pre­tense of “ra­tio­nal­ity”, much less the fact of it.

And if the “pre­dom­i­nant opinion” dis­agrees—so much the worse for pre­dom­i­nant opinion; and so much the sad­der for Less Wrong.

Edit: And all this is, of course, not even men­tion­ing your con­fla­tion of “I am un­com­fortable” with in­sinu­at­ing com­ments about “sinister con­text”, and im­pli­ca­tions of wrong­do­ing on Zack’s part!

I was go­ing to say “boot­straps don’t work that way”, but since the val­i­da­tion hap­pens on the fu­ture end, this might ac­tu­ally work.

Since Eliezer is a tem­po­ral re­duc­tion­ist, I think he might not mean “tem­po­rally con­tin­u­ous”, but rather “log­i­cal/​causal con­ti­nu­ity” or some­thing similar.

Discrete time travel would also vi­o­late tem­po­ral con­ti­nu­ity, by the way.

Ugh, be­cause pro­duc­tive dis­cus­sion hap­pens be­tween perfectly dis­pas­sion­ate robots in a vac­uum, and if I’m not one then it is my fault and I should be ashamed? Speci­fi­cally, I should be ashamed just for say­ing that some­thing made me un­com­fortable rather than suffer­ing in silence? I mean, if that’s your vi­sion, it’s fine, I un­der­stand. But I won­der whether that’s re­ally the pre­dom­i­nant opinion around here? What about all the stuff about “com­mu­nity” and “Village” etc?

I some­times round this off in my head to some­thing like “pure de­cou­plers think ar­gu­ments should be con­sid­ered only on their epistemic mer­its, and pure con­tex­tu­al­iz­ers think ar­gu­ments should be con­sid­ered only on their in­stru­men­tal mer­its”.

The proper words for that aren’t de­cou­pling vs con­tex­tu­al­iz­ing, it’s de­no­ta­tive vs en­ac­tive lan­guage. An or­thog­o­nal axis to how many rele­vant con­tex­tual fac­tors are sup­posed to be taken into ac­count. You can re­quire lots of con­tex­tual fac­tors to be taken into ac­count in epistemic anal­y­sis, or re­quire cer­tain en­act­ments to be made in­de­pen­dent of con­text.

Note, the origi­nal post makes the con­fla­tion I’m com­plain­ing about here too!

I know it’s the typ­i­cal out­come, but I don’t know why it would be in­evitable or ob­vi­ous.

Well, it’s “in­evitable” and/​or “ob­vi­ous” only in the sense that it is quite com­mon­place hu­man be­hav­ior. Cer­tainly it’s not uni­ver­sal—thank­fully!

A per­son that ver­bally asks for an “hon­est” an­swer but pun­ishes is not in fact ask­ing for a hon­est an­swer.

In­deed! They are not, in fact, look­ing for an hon­est an­swer. This is a thing peo­ple do, quite of­ten.

But why would they say that they want an hon­est an­swer? Well, there are many rea­sons, rang­ing from self-de­cep­tion to dom­i­nance/​power games to var­i­ous not-quite-so-dis­rep­utable rea­sons. Enu­mer­at­ing and an­a­lyz­ing all such things would be be­yond the scope of this com­ment thread. The rest of your com­ment is… not so much mis­taken, per se, as it is in­suffi­ciently built upon an un­der­stand­ing of the dy­nam­ics I have al­luded to, their causes, their con­se­quences, etc. There is a great deal of ma­te­rial, on Less Wrong and el­se­where, that dis­cusses this sort of thing. (I do not have links handy, I’m afraid, nor time at the mo­ment to lo­cate them, but per­haps some­one else can point you in the right di­rec­tion.)

… a “sinister con­text”?!

I am, frankly, ap­palled to read this sort of thing on Less Wrong. You are, in all se­ri­ous­ness, at­tack­ing some­one’s writ­ings about ab­stract episte­mol­ogy and Bayesian in­fer­ence, on Less Wrong, of all places (!!), not be­cause there is any­thing at all mis­taken about them, but be­cause of some alleged “sinister con­text” that you are bring­ing in from some­where else. To call this “not a fair ob­jec­tion” would be a gross un­der­state­ment. It is shame­ful.

If we are sup­posed to be a com­mu­nity, then it should be nor­mal for us to con­sider each other’s feel­ings, even when there was no norm vi­o­la­tion per se in­volved, not so?

Ab­solutely not.

This sort of at­ti­tude is tremen­dously cor­ro­sive to pro­duc­tive dis­cus­sion and gen­uine truth-seek­ing. We have dis­cussed this be­fore… and am I gen­uinely dis­ap­pointed that this sort of thing is hap­pen­ing again.

My ba­sic read of Zack’s en­tire post was him say­ing over and over “Well there might be re­ally bad in­stru­men­tal effects of these ar­gu­ments, but you have to ig­nore that if their epistemics are good.” And my im­me­di­ate re­ac­tion to that was “No I don’t, and that’s a bad norm.”

It re­ally looks like you’re defend­ing the “ap­peal to con­se­quences” as a rea­son­able way to think, and a re­spectable ap­proach to pub­lic episte­mol­ogy. But that seems so plainly ab­surd that I have to as­sume that I’ve mi­s­un­der­stood. What am I miss­ing?

Your points 1,2,3 have not­ing to do with the epistemic prob­lem of de­cou­pling vs con­tex­tu­al­iz­ing,

This is prob­a­bly be­cause I don’t know what the epistemic prob­lem is. I only know about the linked post, which defines things like this:

De­cou­pling norms: It is con­sid­ered em­i­nently rea­son­able to re­quire your claims to be con­sid­ered in iso­la­tion—free of any con­text or po­ten­tial im­pli­ca­tions. At­tempts to raise these is­sues are of­ten seen as sloppy think­ing or at­tempts to deflect.

Con­tex­tu­al­is­ing norms: It is con­sid­ered em­i­nently rea­son­able to ex­pect cer­tain con­tex­tual fac­tors or im­pli­ca­tions to be ad­dressed. Not ad­dress­ing these fac­tors is of­ten seen as sloppy or even an in­ten­tional eva­sion.

… To a con­tex­tu­al­iser, de­cou­plers’ abil­ity to fence off any threat­en­ing im­pli­ca­tions looks like a lack of em­pa­thy for those threat­ened, while to a de­cou­pler the con­tex­tu­al­iser’s in­sis­tence that this isn’t pos­si­ble looks like naked bias and an in­abil­ity to think straight”

I some­times round this off in my head to some­thing like “pure de­cou­plers think ar­gu­ments should be con­sid­ered only on their epistemic mer­its, and pure con­tex­tu­al­iz­ers think ar­gu­ments should be con­sid­ered only on their in­stru­men­tal mer­its”.

There might be an­other use of de­cou­pling and con­tex­tu­al­iz­ing that ap­plies to an epistemic prob­lem, but if so it’s not defined in the canon­i­cal ar­ti­cle on the site.

My ba­sic read of Zack’s en­tire post was him say­ing over and over “Well there might be re­ally bad in­stru­men­tal effects of these ar­gu­ments, but you have to ig­nore that if their epistemics are good.” And my im­me­di­ate re­ac­tion to that was “No I don’t, and that’s a bad norm.”

See my re­ply to Said Ach­miz.

I know it’s the typ­i­cal out­come, but I don’t know why it would be in­evitable or ob­vi­ous. A per­son that ver­bally asks for an “hon­est” an­swer but pun­ishes is not in fact ask­ing for a hon­est an­swer. Part of the rea­son why peo­ple add the qual­ifier is the be­lief that those kinds “give you more pos­i­tive af­fect”.

If you try to shoot for an ac­tual hon­est opinion you have to care to differ­en­ti­ate be­tween ask­ing for “dishon­estly hon­est” opinions. For the kind of mind­set that has “what­ever can be de­stroyed by the truth should be de­stroyed” ac­tu­ally hon­est opinions are what to shoot for. But I have bad mod­els on what at­tracts peo­ple to “dishon­estly hon­est” opinions. I sus­pect that that mind­set could benefit from differ­ent fram­ing (“I have your back” vs “yes” ie forgo claims on state of the world in favour of ex­plicit so­cial moves).

This less­wrong post might make some­one seek out more “dishon­est pos­i­tivity” by ap­ply­ing a “re­jec­tion dan­ger” in pur­suit of “be­lief strength­en­ing”. I feel that there is an ar­gu­ment to be made that when re­jec­tion dan­ger re­al­ises you should just eat it in the face with­out re­sist­ing and the failure mode promi­nently fea­tures re­sist­ing the re­jec­tion. And on the bal­ance if you can’t with­stand a no then you will not have earned the yes and should not be ask­ing the ques­tion in the first place.

That is on the epistemic side there is a “con­ser­va­tion of ex­pected ev­i­dence” but on the so­cial side there is a “ad­her­ence to re­cieved choice”, you can’t give con­trol of an area of life con­di­tional on how that con­trol would be used, if you cen­sor some­one you are not in­fact giv­ing them a choice.

The abuse did not hap­pen on LW. How­ever, be­cause I hap­pen to be some­what fa­mil­iar with Davis’ poli­ti­cal writ­ing, I am aware of a sinister con­text to what ey write in LW of which you are not aware. Now, you may say that this is not a fair ob­jec­tion to Davis writ­ing what­ever ey write here, and you might well be right. How­ever, I thought I at least have the right to ex­press my feel­ings on this mat­ter so that Davis and oth­ers can take them into ac­count (or not). If we are sup­posed to be a com­mu­nity, then it should be nor­mal for us to con­sider each other’s feel­ings, even when there was no norm vi­o­la­tion per se in­volved, not so?

It sounds more like a defense of dis­cussing a poli­ti­cal spe­cific by means of ab­strac­tion.

Zack said:

Like, maybe statis­tics is part of the com­mon in­ter­est of many causes, such that, as a mat­ter of lo­cal val­idity, you should as­sess ar­gu­ments about statis­tics on their own mer­its in the con­text that those ar­gu­ments are pre­sented, with­out wor­ry­ing about how those ar­gu­ments might or might not be ap­plied in other con­texts?

What, re­al­is­ti­cally, do you ex­pect the athe­ist—or the racist, or me—to do? Am I sup­posed to just pas­sively ac­cept that all of my thoughts about episte­mol­ogy are tainted and un­fit for this fo­rum, be­cause I hap­pen to be in­ter­ested in ap­ply­ing episte­mol­ogy to other top­ics (on a sep­a­rate web­site, un­der a pseudonym)?

Which isn’t say­ing speci­fics should be dis­cussed by dis­cussing ab­stracts, it says ab­stracts should be dis­cussed, even when part of the mo­ti­va­tion for dis­cussing the ab­stract is spe­cific. Like, peo­ple should be able to col­lab­o­rate on statis­tics text­books even if they don’t agree with their co-au­thors’ spe­cific ap­pli­ca­tions of statis­tics to their non-statis­ti­cal do­mains. (It would be pretty use­less to dis­cuss ab­stracts if there we no spe­cific mo­ti­va­tions, af­ter all...)

It might be in­ter­est­ing that even if you ask for an “hon­est” an­swer peo­ple might re­fuse to give one.

Well, of course this is ex­actly what hap­pens. (Be­cause many peo­ple will claim to want an “hon­est” an­swer, but nev­er­the­less so­cially pun­ish you if you give an an­swer other than the “cor­rect” one… this is ex­tremely un­for­tu­nate be­hav­ior, but such are peo­ple.)

I don’t re­ally have a dog in what­ever fight this is, but look­ing at Zack’s posts and com­ments re­cently, I see noth­ing but in­ter­est­ing and cor­rect in­sights and anal­y­sis, de­void of any ex­plicit poli­tics (but per­haps yield­ing in­sights about such?). How can you call this “abuse”? The over­whelming ma­jor­ity of the con­tent that gets posts to Less Wrong these days should as­pire to the level of qual­ity of the stuff I just linked!