Guided By The Beauty Of Our Weapons

[Con­tent note: kind of talk­ing around Trump sup­port­ers and similar groups as if they’re not there.]


Tim Har­ford writes The Prob­lem With Facts, which uses Brexit and Trump as jump­ing-off points to ar­gue that peo­ple are mostly im­per­vi­ous to facts and re­sis­tant to logic:

All this adds up to a de­press­ing pic­ture for those of us who aren’t ready to live in a post-truth world. Facts, it seems, are tooth­less. Try­ing to re­fute a bold, mem­o­rable lie with a fiddly set of facts can of­ten serve to re­in­force the myth. Im­por­tant truths are of­ten stale and dull, and it is easy to man­u­fac­ture new, more en­gag­ing claims. And giv­ing peo­ple more facts can back­fire, as those facts pro­voke a defen­sive re­ac­tion in some­one who badly wants to stick to their ex­ist­ing world view. “This is dark stuff,” says Reifler. “We’re in a pretty scary and dark time.”

He ad­mits he has no easy an­swers, but cites some stud­ies show­ing that “sci­en­tific cu­ri­os­ity” seems to help peo­ple be­come in­ter­ested in facts again. He thinks maybe we can in­spire sci­en­tific cu­ri­os­ity by link­ing sci­en­tific truths to hu­man in­ter­est sto­ries, by weav­ing com­pel­ling nar­ra­tives, and by find­ing “a Carl Sa­gan or David At­ten­bor­ough of so­cial sci­ence”.

I think this is gen­er­ally a good ar­ti­cle and makes im­por­tant points, but there are three is­sues I want to high­light as pos­si­bly point­ing to a deeper pat­tern.

First, the ar­ti­cle makes the very strong claim that “facts are tooth­less” – then tries to con­vince its read­ers of this us­ing facts. For ex­am­ple, the ar­ti­cle high­lights a study by Ny­han & Reifler which finds a “back­fire effect” – cor­rect­ing peo­ple’s mis­con­cep­tions only makes them cling to those mis­con­cep­tions more strongly. Har­ford ex­pects us to be im­pressed by this study. But how is this differ­ent from all of those so­cial sci­ence facts to which he be­lieves hu­mans are mostly im­per­vi­ous?

Se­cond, Ny­han & Reifler’s work on the back­fire effect is prob­a­bly not true. The origi­nal study es­tab­lish­ing its ex­is­tence failed to repli­cate (see eg Porter & Wood, 2016). This isn’t di­rectly con­trary to Har­ford’s ar­gu­ment, be­cause Har­ford doesn’t cite the origi­nal study – he cites a slight ex­ten­sion of it done a year later by the same team that comes to a slightly differ­ent con­clu­sion. But given that the en­tire field is now in se­ri­ous doubt, I feel like it would have been ju­di­cious to men­tion some of this in the ar­ti­cle. This is es­pe­cially true given that the ar­ti­cle it­self is about the way that false ideas spread by peo­ple never dou­ble-check­ing their be­liefs. It seems to me that if you be­lieve in an epi­demic of false­hood so wide­spread that the very abil­ity to sep­a­rate fact from fic­tion is un­der threat, it ought to in­spire a state of CONSTANT VIGILANCE, where you ob­ses­sively ques­tion each of your be­liefs. Yet Har­ford writes an en­tire ar­ti­cle about a wor­ld­wide plague of false be­liefs with­out mus­ter­ing enough vigilance to see if the rele­vant stud­ies are true or not.

Third, Har­ford de­scribes his ar­ti­cle as be­ing about ag­no­tol­ogy, “the study of how ig­no­rance is de­liber­ately pro­duced”. His key ex­am­ple is to­bacco com­pa­nies sow­ing doubt about the nega­tive health effects of smok­ing – for ex­am­ple, he talks about to­bacco com­pa­nies spon­sor­ing (ac­cu­rate) re­search into all of the non-smok­ing-re­lated causes of dis­ease so that ev­ery­one fo­cused on those in­stead. But his solu­tion – tel­ling en­gag­ing sto­ries, adding a hu­man in­ter­est el­e­ment, en­joy­able doc­u­men­taries in the style of Carl Sa­gan – seems un­usu­ally un­suited to the prob­lem. The Na­tional In­sti­tute of Health can make an en­gag­ing hu­man in­ter­est doc­u­men­tary about a smoker who got lung can­cer. And the to­bacco com­pa­nies can make an en­gag­ing hu­man in­ter­est doc­u­men­tary about a guy who got can­cer be­cause of as­bestos, then was saved by to­bacco-spon­sored re­search. Op­po­nents of Brexit can make an en­gag­ing doc­u­men­tary about all the rea­sons Brexit would be bad, and then pro­po­nents of Brexit can make an en­gag­ing doc­u­men­tary about all the rea­sons Brexit would be good. If you get good doc­u­men­tary-mak­ers, I as­sume both will be equally con­vinc­ing re­gard­less of what the true facts are.

All three of these points are slightly un­fair. The first be­cause Har­ford’s stronger state­ments about facts are prob­a­bly ex­ag­ger­a­tions, and he just meant that in cer­tain cases peo­ple ig­nore ev­i­dence. The sec­ond be­cause the spe­cific study cited wasn’t the one that failed to repli­cate and Har­ford’s the­sis might be that it was differ­ent enough from the origi­nal that it’s prob­a­bly true. And the third be­cause the doc­u­men­taries were just one idea meant to serve a broader goal of in­creas­ing “sci­en­tific cu­ri­os­ity”, a con­struct which has been shown in stud­ies to be helpful in get­ting peo­ple to be­lieve true things.

But I worry that taken to­gether, they sug­gest an un­spo­ken premise of the piece. It isn’t that peo­ple are im­per­vi­ous to facts. Har­ford doesn’t ex­pect his reader to be im­per­vi­ous to facts, he doesn’t ex­pect doc­u­men­tary-mak­ers to be im­per­vi­ous to facts, and he cer­tainly doesn’t ex­pect him­self to be im­per­vi­ous to facts. The prob­lem is that there’s some weird tribe of fact-im­mune troglodytes out there, go­ing around re­fus­ing vac­cines and vot­ing for Brexit, and the rest of us have to figure out what to do about them. The fun­da­men­tal prob­lem is one of trans­mis­sion: how can we make knowl­edge per­co­late down from the fact-lov­ing elite to the fact-im­per­vi­ous masses?

And I don’t want to con­demn this too hard, be­cause it’s ob­vi­ously true up to a point. Med­i­cal re­searchers have lots of use­ful facts about vac­cines. Statis­ti­ci­ans know some great facts about the link be­tween to­bacco and can­cer (shame about Ron­ald Fisher, though). Prob­a­bly there are even some so­cial sci­en­tists who have a fact or two.

Yet as I’ve ar­gued be­fore, ex­ces­sive fo­cus on things like vac­cine de­nial­ists teaches the wrong habits. It’s a de­sire to take a de­gen­er­ate case, the rare situ­a­tion where one side is ob­vi­ously right and the other bizarrely wrong, and make it into the flag­ship ex­am­ple for mod­el­ing all hu­man dis­agree­ment. Imag­ine a the­ory of ju­rispru­dence de­signed only to smack down sovereign cit­i­zens, or a gov­ern­ment pro-in­no­va­tion policy based en­tirely on warn­ing in­ven­tors against per­pet­ual mo­tion ma­chines.

And in this wider con­text, part of me won­ders if the fo­cus on trans­mis­sion is part of the prob­lem. Every­one from statis­ti­ci­ans to Brex­i­teers knows that they are right. The only re­main­ing prob­lem is how to con­vince oth­ers. Go on Face­book and you will find a mil­lion peo­ple with a mil­lion differ­ent opinions, each con­fi­dent in her own judg­ment, each zeal­ously de­voted to in­form­ing ev­ery­one else.

Imag­ine a class­room where ev­ery­one be­lieves they’re the teacher and ev­ery­one else is stu­dents. They all fight each other for space at the black­board, give lec­tures that no­body listens to, as­sign home­work that no­body does. When ev­ery­one gets abysmal test scores, one of the teach­ers has an idea: I need a more en­gag­ing cur­ricu­lum. Sure. That’ll help.


A new Nathan Robin­son ar­ti­cle: De­bate Vs. Per­sua­sion. It goes through the same steps as the Har­ford ar­ti­cle, this time from the per­spec­tive of the poli­ti­cal Left. De­ploy­ing what Robin­son calls “Purely Log­i­cal De­bate” against Trump sup­port­ers hasn’t worked. Some leftists think the an­swer is vi­o­lence. But this may be pre­ma­ture; in­stead, we should try the tools of rhetoric, emo­tional ap­peal, and other forms of dis­course that aren’t Purely Log­i­cal De­bate. In con­clu­sion, Bernie Would Have Won.

I think giv­ing up on ar­gu­men­ta­tion, rea­son, and lan­guage, just be­cause Purely Log­i­cal De­bate doesn’t work, is a mis­take. It’s easy to think that if we can’t con­vince the right with facts, there’s no hope at all for pub­lic dis­course. But this might not sug­gest any­thing about the pos­si­bil­ities of per­sua­sion and di­alogue. In­stead, it might sug­gest that mere facts are rhetor­i­cally in­suffi­cient to get peo­ple ex­cited about your poli­ti­cal pro­gram.

The re­sem­blance to Har­ford is ob­vi­ous. You can’t con­vince peo­ple with facts. But you might be able to con­vince peo­ple with facts care­fully in­ter­mixed with hu­man in­ter­est, com­pel­ling nar­ra­tive, and emo­tional ap­peal.

Once again, I think this is gen­er­ally a good ar­ti­cle and makes im­por­tant points. But I still want to challenge whether things are quite as bad as it says.

Google “de­bat­ing Trump sup­port­ers is”, and you re­al­ize where the ar­ti­cle is com­ing from. It’s page af­ter page of “de­bat­ing Trump sup­port­ers is pointless”, “de­bat­ing Trump sup­port­ers is a waste of time”, and “de­bat­ing Trump sup­port­ers is like [funny metaphor for thing that doesn’t work]”. The over­all pic­ture you get is of a world full of Trump op­po­nents and sup­port­ers de­bat­ing on ev­ery street cor­ner, un­til fi­nally, af­ter months of bang­ing their heads against the wall, ev­ery­one col­lec­tively de­cided it was fu­tile.

Yet I have the op­po­site im­pres­sion. Some­how a sharply po­larized coun­try went through a his­tor­i­cally di­vi­sive elec­tion with es­sen­tially no de­bate tak­ing place.

Am I about to No True Scots­man the hell out of the word “de­bate”? Maybe. But I feel like in us­ing the ex­ag­ger­ated phrase “Purely Log­i­cal De­bate, Robin­son has given me leave to define the term as strictly as I like. So here’s what I think are min­i­mum stan­dards to de­serve the cap­i­tal let­ters:

1. De­bate where two peo­ple with op­pos­ing views are talk­ing to each other (or writ­ing, or IMing, or some form of bilat­eral com­mu­ni­ca­tion). Not a pun­dit putting an ar­ti­cle on Huffing­ton Post and de­mand­ing Trump sup­port­ers read it. Not even a Trump sup­porter who com­ments on the ar­ti­cle with a coun­ter­ar­gu­ment that the au­thor will never read. Two peo­ple who have cho­sen to en­gage and to listen to one an­other.

2. De­bate where both peo­ple want to be there, and have cho­sen to en­ter into the de­bate in the hopes of get­ting some­thing pro­duc­tive out of it. So not some­thing where some­one posts a “HILLARY IS A CROOK” meme on Face­book, some­one gets re­ally an­gry and lists all the rea­sons Trump is an even big­ger crook, and then the origi­nal poster gets an­gry and has to tell them why they’re wrong. Two peo­ple who have made it their busi­ness to come to­gether at a cer­tain time in or­der to com­pare opinions.

3. De­bate con­ducted in the spirit of mu­tual re­spect and col­lab­o­ra­tive truth-seek­ing. Both peo­ple re­ject per­sonal at­tacks or ‘gotcha’ style digs. Both peo­ple un­der­stand that the other per­son is around the same level of in­tel­li­gence as they are and may have some use­ful things to say. Both peo­ple un­der­stand that they them­selves might have some false be­liefs that the other per­son will be able to cor­rect for them. Both peo­ple go into the de­bate with the hope of con­vinc­ing their op­po­nent, but not com­pletely re­ject­ing the pos­si­bil­ity that their op­po­nent might con­vince them also.

4. De­bate con­ducted out­side of a high-pres­sure point-scor­ing en­vi­ron­ment. No au­di­ence cheer­ing on both par­ti­ci­pants to re­spond as quickly and bit­ingly as pos­si­ble. If it can’t be done on­line, at least do it with a smart­phone around so you can open Wikipe­dia to re­solve sim­ple mat­ters of fact.

5. De­bate where both peo­ple agree on what’s be­ing de­bated and try to stick to the sub­ject at hand. None of this “I’m go­ing to vote Trump be­cause I think Clin­ton is cor­rupt” fol­lowed by “Yeah, but Rea­gan was even worse and that just proves you Repub­li­cans are hyp­ocrites” fol­lowed by “We’re hyp­ocrites? You Democrats claim to sup­port women’s rights but you love Mus­lims who make women wear head­scarves!” Whether or not it’s hyp­o­crit­i­cal to “sup­port women’s rights” but “love Mus­lims”, it doesn’t seem like any­one is even try­ing to change each other’s mind about Clin­ton at this point.

Th­ese to me seem like the bare min­i­mum con­di­tions for a de­bate that could pos­si­bly be pro­duc­tive.

(and while I’m ask­ing for a pony on a silver plat­ter, how about both peo­ple have to read How To Ac­tu­ally Change Your Mind<A> first?)

Mean­while, in re­al­ity…

If you search “de­bat­ing Trump sup­port­ers” with­out the “is”, your first re­sult is this video, where some peo­ple with a micro­phone cor­ner some other peo­ple at what looks like a rally. I can’t re­ally fol­low the con­ver­sa­tion be­cause they’re all shout­ing at the same time, but I can make out some­body say­ing ‘Repub­li­cans give more to char­ity!’ and some­one else re­spond­ing ‘That’s cause they don’t do any­thing at their jobs!’”. Okay.

The sec­ond link is this pod­cast where a guy talks about de­bat­ing Trump sup­port­ers. After the usual pref­ace about how stupid they were, he de­scribes a typ­i­cal ex­change – “It’s kind of amaz­ing how they want to go back to the good old days…Well, when I start ask­ing them ‘You mean the good old days when 30% of the pop­u­la­tion were in unions’…they never seem to like to hear that!…so all this un­fet­tered free mar­ket cap­i­tal­ism has got to go bye-bye. They don’t find com­fort in that idea ei­ther. It’s amaz­ing. I can say I now know what cog­ni­tive dis­so­nance feels like on some­one’s face.” I’m glad time travel seems to be im­pos­si­ble, be­cause oth­er­wise I would be tempted to warp back and change my vote to Trump just to spite this per­son.

The third link is Van­ity Fair’s “Foolproof Guide To Ar­gu­ing With Trump Sup­port­ers”, which sug­gests “us­ing their pa­tri­o­tism against them” by tel­ling them that want­ing to “cur­tail the rights and priv­ileges of cer­tain of our cit­i­zens” is un-Amer­i­can.

I worry that peo­ple do this kind of thing ev­ery so of­ten. Then, when it fails, they con­clude “Trump sup­port­ers are im­mune to logic”. This is much like ob­serv­ing that Repub­li­cans go out in the rain with­out melt­ing, and con­clud­ing “Trump sup­port­ers are im­mor­tal”.

Am I say­ing that if you met with a con­ser­va­tive friend for an hour in a quiet cafe to talk over your dis­agree­ments, they’d come away con­vinced? No. I’ve changed my mind on var­i­ous things dur­ing my life, and it was never a sin­gle mo­ment that did it. It was more of a se­ries of differ­ent things, each tak­ing me a frac­tion of the way. As the old say­ing goes, “First they ig­nore you, then they laugh at you, then they fight you, then they fight you half-heart­edly, then they’re neu­tral, then they then they grudg­ingly say you might have a point even though you’re an­noy­ing, then they say on bal­ance you’re mostly right al­though you ig­nore some of the most im­por­tant facets of the is­sue, then you win.”

There might be a par­allel here with the one place I see some­thing like Purely Log­i­cal De­bate on a rou­tine ba­sis: cog­ni­tive psy­chother­apy. I know this com­par­i­son sounds crazy, be­cause psy­chother­apy is sup­posed to be the op­po­site of a de­bate, and try­ing to ar­gue some­one out of their delu­sions or de­pres­sion in­evitably fails. The rook­iest of all rookie ther­a­pist mis­takes is to say “FACT CHECK: The pa­tient says she is a loser who ev­ery­body hates. Psy­chi­aFact rates this claim: PANTS ON FIRE.”

But in other ways it’s a lot like the five points above. You have two peo­ple who dis­agree – the pa­tient thinks she’s a worth­less loser who ev­ery­one hates, and the ther­a­pist thinks maybe not. They meet to­gether in a spirit of vol­un­tary mu­tual in­quiry, guaran­teed safe from per­sonal at­tacks like “You’re crazy!”. Both sides go over the ev­i­dence to­gether, some­times even agree­ing on ex­plicit ex­per­i­ments like “Ask your boyfriend tonight whether he hates you or not, pre­dict be­fore­hand what you think he’s go­ing to say, and see if your pre­dic­tion is ac­cu­rate”. And both sides ap­proach the whole pro­cess sus­pect­ing that they’re right but ad­mit­ting the pos­si­bil­ity that they’re wrong (very oc­ca­sion­ally, af­ter weeks of ther­apy, I re­al­ize that frick, ev­ery­one re­ally does hate my pa­tient. Then we switch strate­gies to helping her with so­cial skills, or helping her find bet­ter friends).

And con­trary to what you see in movies, this doesn’t usu­ally give a sin­gle mo­ment of blind­ing rev­e­la­tion. If you spent your en­tire life talk­ing your­self into the be­lief that you’re a loser and ev­ery­one hates you, no sin­gle fact or per­son is go­ing to talk you out of it. But af­ter how­ever many months of in­ten­sive ther­apy, some­times some­one who was sure that they were a loser is now sort of ques­tion­ing whether they’re a loser, and has the men­tal toolbox to take things the rest of the way them­selves.

This was also the re­sponse I got when I tried to make an anti-Trump case on this blog. I don’t think there were any sud­den con­ver­sions, but here were some of the pos­i­tive com­ments I got from Trump sup­port­ers:

“This is a com­pel­ling case, but I’m still torn.”

“This con­tains the most con­vinc­ing ar­gu­ments for a Clin­ton pres­i­dency I have ever seen. But, per­haps also un­sur­pris­ingly, while it did man­age to shift some of my views, it did not suc­ceed in con­vinc­ing me to change my bot­tom line.”

“This ar­ti­cle is per­haps the best ar­gu­ment I have seen yet for Hillary. I found my­self nod­ding along with many of the ar­gu­ments, af­ter this morn­ing swear­ing that there was noth­ing that could make me con­sider vot­ing for Hillary…the prob­lem in the end was that it wasn’t enough.”

“The first co­her­ent ar­ti­cle I’ve read jus­tify­ing vot­ing for Clin­ton. I don’t agree with your anal­y­sis of the dol­lar “value” of a vote, but other than that, some­thing to think about.”

“Well I don’t like Clin­ton at all, and I found this es­say rea­son­able enough. The ar­gu­ment from con­ti­nu­ity is prob­a­bly the best one for vot­ing Clin­ton if you don’t par­tic­u­larly love any of her poli­cies or her as a per­son. Trump is a wild card, I must ad­mit.”

As an or­tho­dox Catholic, you would prob­a­bly clas­sify me as part of your con­ser­va­tive au­di­ence…I cer­tainly con­cur with both the var­i­ance ar­gu­ments and that he’s not con­ser­va­tive by policy, life, or tem­per­a­ment, and I will re­main open to hear­ing what you have to say on the topic through Novem­ber.

“I’ve only come around to the ‘hold your nose and vote Trump’ camp the past month or so…I won’t say [you] didn’t make me squirm, but I’m hold­ing fast to my de­ci­sion.”

Th­ese are the peo­ple you say are com­pletely im­per­vi­ous to logic so don’t even try? It seems to me like this ar­gu­ment was one of not-so-many straws that might have bro­ken some camels’ backs if they’d been al­lowed to ac­cu­mu­late. And the weird thing is, when I re-read the es­say I no­tice a lot of flaws and things I wish I’d said differ­ently. I don’t think it was an ex­cep­tion­ally good ar­gu­ment. I think it was…an ar­gu­ment. It was some­thing more than say­ing “You think the old days were so great, but the old days had la­bor unions, CHECKMATE ATHEISTS”. This isn’t what you get when you do a splen­did vir­tu­ouso perfo­mance. This is what you get when you show up.

(and lest I end up ‘ob­jec­tify­ing’ Trump sup­port­ers as prizes to be won, I’ll add that in the com­ments some peo­ple made pro-Trump ar­gu­ments, and two peo­ple who were pre­vi­ously lean­ing Clin­ton said that they were feel­ing un­com­fortably close to be­ing con­vinced)

Another SSC story. I keep try­ing to keep “cul­ture war”-style poli­ti­cal ar­gu­ments from over­run­ning the blog and sub­red­dit, and ev­ery time I add re­stric­tions a bunch of peo­ple com­plain that this is the only place they can go for that. Think about this for a sec­ond. A heav­ily po­larized coun­try of three hun­dred mil­lion peo­ple, split pretty evenly into two sides and ob­sessed with poli­tics, blessed with the strongest free speech laws in the world, and peo­ple are com­plain­ing that I can’t change my com­ment policy be­cause this one small blog is the only place they know where they can de­bate peo­ple from the other side.

Given all of this, I re­ject the ar­gu­ment that Purely Log­i­cal De­bate has been tried and found want­ing. Like GK Ch­ester­ton, I think it has been found difficult and left un­tried.


Ther­apy might change minds, and so might friendly de­bate among equals, but nei­ther of them scales very well. Is there any­thing that big fish in the me­dia can do be­yond the trans­mis­sion they’re already try­ing?

Let’s go back to that Ny­han & Reifler study which found that fact-check­ing back­fired. As I men­tioned above, a repli­ca­tion at­tempt by Porter & Wood found the op­po­site. This could have been the setup for a nasty con­flict, with both groups try­ing to con­vince academia and the pub­lic that they were right, or even ac­cus­ing the other of sci­en­tific malprac­tice.

In­stead, some­thing great hap­pened. All four re­searchers de­cided to work to­gether on an “ad­ver­sar­ial col­lab­o­ra­tion” – a big­ger, bet­ter study where they all had in­put into the method­ol­ogy and they all checked the re­sults in­de­pen­dently. The col­lab­o­ra­tion found that fact-check­ing gen­er­ally didn’t back­fire in most cases. All four of them used their sci­en­tific clout to pub­li­cize the new re­sult and launch fur­ther in­ves­ti­ga­tions into the role of differ­ent con­texts and situ­a­tions.

In­stead of treat­ing dis­agree­ment as demon­strat­ing a need to trans­mit their own opinion more effec­tively, they viewed it as demon­strat­ing a need to col­lab­o­rate to in­ves­ti­gate the ques­tion to­gether.

And yeah, part of it was that they were all de­cent sci­en­tists who re­spected each other. But they didn’t have to be. If one team had been to­tal mo­rons, and the other team was se­cretly laugh­ing at them the whole time, the col­lab­o­ra­tion still would have worked. All re­quired was an as­sump­tion of good faith.

A while ago I blogged about a jour­nal­is­tic spat be­tween Ger­man Lopez and Robert VerBruggen on gun con­trol. Lopez wrote a voxs­plainer cit­ing some statis­tics about guns. VerBruggen wrote a piece at Na­tional Re­view say­ing that some of the statis­tics were flawed. Ger­man fired back (pun not in­tended) with an ar­ti­cle claiming that VerBruggen was ig­nor­ing bet­ter stud­ies.

(Then I yel­led at both of them, as usual.)

Over­all the ex­change was in the top 1% of on­line so­cial sci­ence jour­nal­ism – by which I mean it in­cluded at least one statis­tic and at some point that statis­tic was su­perfi­cially ex­am­ined. But in the end, it was still just two peo­ple ar­gu­ing with one an­other, each try­ing to trans­mit his su­pe­rior knowl­edge to each other and the read­ing pub­lic. As good as it was, it didn’t meet my five stan­dards above – and no­body ex­pected it to.

But now I’m think­ing – what would have hap­pened if Lopez and VerBruggen had joined to­gether in an ad­ver­sar­ial col­lab­o­ra­tion? Agreed to work to­gether to write an ar­ti­cle on gun statis­tics, with noth­ing go­ing into the ar­ti­cle un­less they both ap­proved, and then they both pub­lished that ar­ti­cle on their re­spec­tive sites?

This seems like a mass me­dia equiv­a­lent of shift­ing from Twit­ter spats to se­ri­ous de­bate, from trans­mis­sion mind­set to col­lab­o­ra­tive truth-seek­ing mind­set. The ad­ver­sar­ial col­lab­o­ra­tion model is just the first one to come to mind right now. I’ve blogged about oth­ers be­fore – for ex­am­ple, bets, pre­dic­tion mar­kets, and cal­ibra­tion train­ing.

The me­dia already spends a lot of effort recom­mend­ing good be­hav­ior. What if they tried mod­el­ing it?


The big­ger ques­tion hang­ing over all of this: “Do we have to?”

Har­ford’s solu­tion – com­pel­ling nar­ra­tives and doc­u­men­taries – sounds easy and fun. Robin­son’s solu­tion – rhetoric and emo­tional ap­peals – also sounds easy and fun. Even the solu­tion Robin­son re­jects – vi­o­lence – is easy, and fun for a cer­tain type of per­son. All three work on pretty much any­body.

Purely Log­i­cal De­bate is difficult and an­noy­ing. It doesn’t scale. It only works on the sub­set of peo­ple who are will­ing to talk to you in good faith and smart enough to un­der­stand the is­sues in­volved. And even then, it only works glacially slowly, and you win only par­tial vic­to­ries. What’s the point?

Log­i­cal de­bate has one ad­van­tage over nar­ra­tive, rhetoric, and vi­o­lence: it’s an asym­met­ric weapon. That is, it’s a weapon which is stronger in the hands of the good guys than in the hands of the bad guys. In ideal con­di­tions (which may or may not ever hap­pen in real life) – the kind of con­di­tions where ev­ery­one is char­i­ta­ble and in­tel­li­gent and wise – the good guys will be able to pre­sent stronger ev­i­dence, cite more ex­perts, and in­voke more com­pel­ling moral prin­ci­ples. The whole point of logic is that, when done right, it can only prove things that are true.

Violence is a sym­met­ric weapon; the bad guys’ punches hit just as hard as the good guys’ do. It’s true that hope­fully the good guys will be more pop­u­lar than the bad guys, and so able to gather more sol­diers. But this doesn’t mean vi­o­lence it­self is asym­met­ric – the good guys will only be more pop­u­lar than the bad guys in­so­far as their ideas have pre­vi­ously spread through some means other than vi­o­lence. Right now an­tifas­cists out­num­ber fas­cists and so could prob­a­bly beat them in a fight, but an­tifas­cists didn’t come to out­num­ber fas­cists by win­ning some kind of pri­mor­dial fist­fight be­tween the two sides. They came to out­num­ber fas­cists be­cause peo­ple re­jected fas­cism on the mer­its. Th­ese mer­its might not have been “log­i­cal” in the sense of Aris­to­tle dis­pas­sion­ately prov­ing lem­mas at a chalk­board, but “fas­cists kill peo­ple, kil­ling peo­ple is wrong, there­fore fas­cism is wrong” is a sort of folk log­i­cal con­clu­sion which is both cor­rect and com­pel­ling. Even “a fas­cist kil­led my brother, so fuck them” is a place­holder for a pow­er­ful philo­soph­i­cal ar­gu­ment mak­ing a prob­a­bil­is­tic gen­er­al­iza­tion from in­dex­i­cal ev­i­dence to global util­ity. So in­so­far as vi­o­lence is asym­met­ric, it’s be­cause it par­a­sitizes on logic which al­lows the good guys to be more con­vinc­ing and so field a big­ger army. Violence it­self doesn’t en­hance that asym­me­try; if any­thing, it de­creases it by giv­ing an ad­van­tage to who­ever is more ruth­less and power-hun­gry.

The same is true of doc­u­men­taries. As I said be­fore, Har­ford can pro­duce as many anti-Trump doc­u­men­taries as he wants, but Trump can fund doc­u­men­taries of his own. He has the best doc­u­men­taries. No­body has ever seen doc­u­men­taries like this. They’ll be ab­solutely huge.

And the same is true of rhetoric. Martin Luther King was able to make per­sua­sive emo­tional ap­peals for good things. But Hitler was able to make per­sua­sive emo­tional ap­peals for bad things. I’ve pre­vi­ously ar­gued that Mo­hammed counts as the most suc­cess­ful per­suader of all time. Th­ese three peo­ple pushed three very differ­ent ide­olo­gies, and rhetoric worked for them all. Robin­son writes as if “use rhetoric and emo­tional ap­peals” is a novel idea for Democrats, but it seems to me like they were do­ing lit­tle else through­out the elec­tion (pieces at­tack­ing Trump’s char­ac­ter, pieces talk­ing about how in­spira­tional Hillary was, pieces ap­peal­ing to var­i­ous Amer­i­can prin­ci­ples like equal­ity, et cetera). It’s just that they did a bad job, and Trump did a bet­ter one. The real take­away here is “do rhetoric bet­ter than the other guy”. But “suc­ceed” is not a prim­i­tive ac­tion.

Un­less you use asym­met­ric weapons, the best you can hope for is to win by co­in­ci­dence.

That is, there’s no rea­son to think that good guys are con­sis­tently bet­ter at rhetoric than bad guys. Some days the Left will have an Obama and win the rhetoric war. Other days the Right will have a Rea­gan and they’ll win the rhetoric war. Over­all you should av­er­age out to a 50% suc­cess rate. When you win, it’ll be be­cause you got lucky.

And there’s no rea­son to think that good guys are con­sis­tently bet­ter at doc­u­men­taries than bad guys. Some days the NIH will spin a com­pel­ling nar­ra­tive and peo­ple will smoke less. Other days the to­bacco com­pa­nies will spin a com­pel­ling nar­ra­tive and peo­ple will smoke more. Over­all smok­ing will stay the same. And again, if you win, it’s be­cause you lucked out into hav­ing bet­ter videog­ra­phers or some­thing.

I’m not against win­ning by co­in­ci­dence. If I stum­bled across Stalin and I hap­pened to have a gun, I would shoot him with­out wor­ry­ing about how it’s “only by co­in­ci­dence” that he didn’t have the gun in­stead of me. You should use your sym­met­ric weapons if for no rea­son other than that the other side’s go­ing to use theirs and so you’ll have a dis­ad­van­tage if you don’t. But you shouldn’t con­fuse it with a long-term solu­tion.

Im­prov­ing the qual­ity of de­bate, shift­ing peo­ple’s mind­sets from trans­mis­sion to col­lab­o­ra­tive truth-seek­ing, is a painful pro­cess. It has to be done one per­son at a time, it only works on peo­ple who are already al­most ready for it, and you will pick up far fewer warm bod­ies per hour of work than with any of the other meth­ods. But in an oth­er­wise-ran­dom world, even a lit­tle pur­pose­ful ac­tion can make a differ­ence. Con­vinc­ing 2% of peo­ple would have flipped three of the last four US pres­i­den­tial elec­tions. And this is a ca­pac­ity to win-for-rea­sons-other-than-co­in­ci­dence that you can’t build any other way.

(and my hope is that the peo­ple most will­ing to en­gage in de­bate, and the ones most likely to rec­og­nize truth when they see it, are dis­pro­por­tionately in­fluen­tial – sci­en­tists, writ­ers, and com­mu­nity lead­ers who have in­fluence be­yond their num­ber and can help oth­ers see rea­son in turn)

I worry that I’m not com­mu­ni­cat­ing how beau­tiful and in­evitable all of this is. We’re sur­rounded by a a vast con­fu­sion, “a dark­ling plain where ig­no­rant armies clash by night”, with one side or an­other mak­ing a tem­po­rary ad­vance and then fal­ling back in turn. And in the mid­dle of all of it, there’s this grad­ual ca­pac­ity-build­ing go­ing on, where what starts off as a hope­lessly weak sig­nal grad­u­ally builds up strength, un­til one army starts win­ning a lit­tle more of­ten than chance, then a lot more of­ten, and fi­nally takes the field en­tirely. Which seems strange, be­cause surely you can’t build any com­plex sig­nal-de­tec­tion ma­chin­ery in the mid­dle of all the chaos, surely you’d be shot the mo­ment you left the trenches, but – your en­e­mies are helping you do it. Both sides are di­vert­ing their ar­tillery from the rele­vant ar­eas, pool­ing their re­sources, helping bring sup­plies to the en­g­ineers, be­cause un­til the very end they think it’s go­ing to en­sure their fi­nal vic­tory and not yours.

You’re do­ing it right un­der their noses. They might try to ban your doc­u­men­taries, heckle your speeches, fight your vi­o­lence Mid­dle­bury-stu­dent-for-Mid­dle­bury-stu­dent – but when it comes to the long-term solu­tion to en­sure your com­plete vic­tory, they’ll roll down their sleeves, get out their ham­mers, and build it alongside you.

A parable: Sally is a psy­chi­a­trist. Her pa­tient has a strange delu­sion: that Sally is the pa­tient and he is the psy­chi­a­trist. She would like to com­mit him and force med­i­ca­tion on him, but he is an im­por­tant poli­ti­cian and if push comes to shove he might be able to com­mit her in­stead. In des­per­a­tion, she pro­poses a bar­gain: they will both take a cer­tain med­i­ca­tion. He agrees; from within his delu­sion, it’s the best way for him-the-psy­chi­a­trist to cure her-the-pa­tient. The two take their pills at the same time. The med­i­ca­tion works, and the pa­tient makes a full re­cov­ery.

(well, half the time. The other half, the med­i­ca­tion works and Sally makes a full re­cov­ery.)


Har­ford’s ar­ti­cle says that facts and logic don’t work on peo­ple. The var­i­ous lefty ar­ti­cles say they merely don’t work on Trump sup­port­ers, ie 50% of the pop­u­la­tion.

If you gen­uinely be­lieve that facts and logic don’t work on peo­ple, you shouldn’t be writ­ing ar­ti­cles with po­ten­tial solu­tions. You should be jet­ti­son­ing ev­ery­thing you be­lieve and en­ter­ing a state of pure Carte­sian doubt, where you try to red­erive ev­ery­thing from cog­ito ergo sum.

If you gen­uinely be­lieve that facts and logic don’t work on at least 50% of the pop­u­la­tion, again, you shouldn’t be writ­ing ar­ti­cles with po­ten­tial solu­tions. You should be wor­ry­ing whether you’re in that 50%. After all, how did you figure out you aren’t? By us­ing facts and logic? What did we just say?

No­body is do­ing ei­ther of these things, so I con­clude that they ac­cept that facts can some­times work. Asym­met­ric weapons are not a pipe dream. As Gandhi used to say, “If you think the world is all bad, re­mem­ber that it con­tains peo­ple like you.”

You are not com­pletely im­mune to facts and logic. But you have been wrong about things be­fore. You may be a bit smarter than the peo­ple on the other side. You may even be a lot smarter. But fun­da­men­tally their prob­lems are your prob­lems, and the same kind of logic that con­vinced you can con­vince them. It’s just go­ing to be a long slog. You didn’t de­velop your opinions af­ter a five-minute shout­ing match. You de­vel­oped them af­ter years of ed­u­ca­tion and ac­cul­tura­tion and en­gag­ing with hun­dreds of books and hun­dreds of peo­ple. Why should they be any differ­ent?

You end up be­liev­ing that the prob­lem is deeper than in­suffi­cient doc­u­men­tary pro­duc­tion. The prob­lem is that Truth is a weak sig­nal. You’re try­ing to per­ceive Truth. You would like to hope that the other side is try­ing to per­ceive Truth too. But at least one of you is do­ing it wrong. It seems like per­ceiv­ing Truth ac­cu­rately is harder than you thought.

You be­lieve your mind is a truth-sens­ing in­stru­ment that does at least a lit­tle bit bet­ter than chance. You have to be­lieve that, or else what’s the point? But it’s like one of those physics ex­per­i­ments set up to de­tect grav­i­ta­tional waves or some­thing, where it has to be in a cav­ern five hun­dred feet un­der­ground in a lead-shielded cham­ber atop a gy­ro­scop­i­cally sta­ble plat­form cooled to one de­gree above ab­solute zero, try­ing to de­tect fluc­tu­a­tions of a mil­lionth of a cen­time­ter. Ex­cept you don’t have the cav­ern or the lead or the gy­ro­scope or the coolants. You’re on top of an erupt­ing vol­cano be­ing pelted by me­te­orites in the mid­dle of a hur­ri­cane.

If you study psy­chol­ogy for ten years, you can re­move the vol­cano. If you spend an­other ten years ob­ses­sively check­ing your perfor­mance in var­i­ous metis-in­ten­sive do­mains, you can re­move the me­te­orites. You can never re­move the hur­ri­cane and you shouldn’t try. But if there are a thou­sand trust­wor­thy peo­ple at a thou­sand differ­ent parts of the hur­ri­cane, then the stray gusts of wind will can­cel out and they can av­er­age their read­ings to get some­thing ap­proach­ing a sig­nal.

All of this is too slow and un­cer­tain for a world that needs more wis­dom now. It would be nice to force the mat­ter, to pelt peo­ple with speeches and doc­u­men­taries un­til they come around. This will work in the short term. In the long term, it will leave you back where you started.

If you want peo­ple to be right more of­ten than chance, you have to teach them ways to dis­t­in­guish truth from false­hood. If this is in the face of en­emy ac­tion, you will have to teach them so well that they can­not be fooled. You will have to do it per­son by per­son un­til the sig­nal is strong and clear. You will have to raise the san­ity wa­ter­line. There is no short­cut.