Overconfident talking down, humble or hostile talking up

A Distri­bu­tion of Knowlege

If one were to make a dis­tri­bu­tion of the amount of knowl­edge differ­ent peo­ple have about, say, macroe­co­nomics, I would sus­pect the dis­tri­bu­tion to be some­what log­nor­mal; they would have tails to both ends, but be very skewed to the right. Most peo­ple have al­most no knowl­edge of macroe­co­nomics, some have a bit, then there is a long tail of fewer and fewer peo­ple who make up the ex­perts.

The above graph doesn’t ex­actly re­sem­ble what I’d ex­pect for macroe­co­nomics but acts as a rough heuris­tic. The large num­bers rep­re­sent halv­ings of the re­main­ing per­centiles (3/​4th, 7/​8th, 15/​16th, etc).[1]

I’m go­ing to posit the fol­low­ing claims:[2]

Claim 1: It’s easy to judge where on the curve peo­ple are who are lower than you.

Claim 2: It’s difficult to judge where on the curve peo­ple are who are higher than you, ab­sent of so­phis­ti­cated sys­tems to sup­port this.

Given these, let’s imag­ine a few situ­a­tions:

  1. Say you’re the lo­cal economist for a state gov­ern­ment. You have some ac­tions you re­ally would like the gov­ern­ment to take, even though your col­leagues wouldn’t typ­i­cally ap­prove. You’re around a +4 on the macroe­co­nomic scale, and your most knowl­edge­able col­leagues are around a +2. Could you get away with pre­tend­ing that macroe­co­nomics has a very con­fi­dent stance that hap­pens to al­ign with what you want to see hap­pen? How would you do so?

  2. You’re a lo­cal ra­dio in­tel­lec­tual who’s a +3 on the macroe­co­nomic scale. Al­most all of your listen­ers are be­low a +1.5. You’ve been es­sen­tially ly­ing to them for some time about macroe­co­nomic the­ory be­cause it helps your poli­ti­cal mes­sage. A pro­fes­sor who’s a +5 starts writ­ing a few ar­ti­cles that call you out on your lies. How do you re­spond?

  3. You’re a col­lege stu­dent who’s a +3 on the macroe­co­nomic scale. Your pro­fes­sors are all a good deal higher and will be the pri­mary ones eval­u­at­ing your stud­ies. You want to le­gi­t­i­mately learn macroe­co­nomics. How do you treat your pro­fes­sors?

Over­con­fi­dent talk­ing down, hum­ble or hos­tile talk­ing up

I think the an­swers I’d ex­pect from these ques­tions can be sum­ma­rized in the phrase “Over­con­fi­dent talk­ing down, hum­ble or hos­tile talk­ing up.”

When you’re com­mu­ni­cat­ing with peo­ple who know less than you, and you have lit­tle ac­countabil­ity from peo­ple who know more, then you gen­er­ally have the op­tion of claiming to be more knowl­edge­able than you are, and ly­ing in ways that are use­ful to you.

When you’re com­mu­ni­cat­ing with peo­ple who know more than you, you have two op­tions. You can ac­cept their greater state of knowl­edge, caus­ing you to speak more hon­estly about the per­ti­nent top­ics. Or, you could re­ject their cred­i­bil­ity, claiming that they re­ally don’t know more than you. Many peo­ple who know less than you both may be­lieve you over them.

There are many ex­am­ples of this. One par­tic­u­larly good one may be the his­tory of schisms in re­li­gious or­ga­ni­za­tions. Reli­gious au­thor­i­ties gen­er­ally know a lot more about their re­spec­tive re­li­gions than the ma­jor­ity of cit­i­zens. Each au­thor­ity has a choice; they could ei­ther ac­cept the knowl­edge of the higher au­thor­i­ties, or they could re­ject the higher au­thor­i­ties. If they re­ject above au­thor­ity, they would be in­cen­tivized to dis­credit that au­thor­ity and ex­press over­con­fi­dence in their own new be­liefs. If they suc­ceed, some fol­low­ers would be­lieve them, giv­ing them both the as­sump­tion of ex­per­tise and also the flex­i­bil­ity of not hav­ing to be ac­countable to other knowl­edge­able groups. If they both defect on their pre­vi­ous au­thor­i­ties and fail, then they may wind up in a very poor po­si­tion, so af­ter defect­ing it’s very im­por­tant to en­sure that their ex­ist­ing au­di­ence gives them full sup­port.

The Eco­nomics of Knowl­edge Signaling

In slightly more eco­nomic terms, one could say that there are strong sig­nals go­ing up the chain of knowl­edge (from the non­ex­perts to the ex­perts), and weak sig­nals go­ing down it. The mar­ket for knowl­edge­able ex­per­tise is one with rel­a­tively low trans­parency and typ­i­cal in­cen­tives to lie and de­ceive, similar to the mar­kets for lemons.

I’m not claiming with this that all of the over­con­fi­dence and dis­cred­it­ing is know­ingly dishon­est.[3] I’m also not claiming that this is origi­nal; much is quite ob­vi­ous and parts are definitely stud­ied. That said, I do get the im­pres­sion that the sci­ence of sig­nal­ing is still pretty over­looked (much of this is from Robin Han­son), and this is one area I think may not be well un­der­stood as a holis­tic eco­nomic sys­tem.

Fi­nally, I’m re­minded of the old joke:

“Once I saw this guy on a bridge about to jump. I said, “Don’t do it!” He said, “No­body loves me.” I said, “God loves you. Do you be­lieve in God?”
He said, “Yes.” I said, “Are you a Chris­tian or a Jew?” He said, “A Chris­tian.” I said, “Me, too! What fran­chise?” He said, “Protes­tant.” I said, “Me, too! North­ern Bap­tist or South­ern Bap­tist?” He said, “North­ern Bap­tist.” I said, “Me, too! North­ern Con­ser­va­tive Bap­tist or North­ern Liberal Bap­tist?” He said, “North­ern Con­ser­va­tive Bap­tist.” I said, “Me, too! North­ern Con­ser­va­tive Bap­tist Great Lakes Re­gion, or North­ern Con­ser­va­tive Bap­tist Eastern Re­gion?” He said, “North­ern Con­ser­va­tive Bap­tist Great Lakes Re­gion.”
I said, “Me, too!” “North­ern Con­ser­va­tive Bap­tist Great Lakes Re­gions Coun­cil of 1879 or North­ern Con­ser­va­tive Bap­tist Great Lakes Re­gion Coun­cil of 1912?” He said “North­ern Con­ser­va­tive Bap­tist Great Lakes Coun­cil of 1912.” I said, “Die, heretic!” And I pushed him over.

One may won­der what in­cen­tives seem to lead to such heart­felt but pre­dictably fre­quent di­vi­sions.


  1. This is similar to the log-odds scale. Stan­dard de­vi­a­tion could also be used, but I find it a bit un­in­tu­itive, es­pe­cially for non-nor­mal dis­tri­bu­tions.

  2. Th­ese mostly come from lots of anec­do­tal ev­i­dence, some gen­eral rea­son­ing, and my mem­o­ries of a few re­search stud­ies. I’ve spent around 40 min­utes at­tempted to lo­cate use­ful stud­ies for this post, but haven’t, though I’m quite sure I re­mem­ber read­ing about re­lated ones sev­eral years ago. If you have any recom­mended links, please post in the com­ments.

  3. The first few chap­ters of The Elephant in the Brain go into this.