Inadequacy and Modesty


The fol­low­ing is the be­gin­ning of Inad­e­quate Equil­ibria, a new se­quence/​book on a gen­er­al­iza­tion of the no­tion of effi­cient mar­kets, and on this no­tion’s im­pli­ca­tions for prac­ti­cal de­ci­sion-mak­ing and epistemic ra­tio­nal­ity.

This is a book about two in­com­pat­i­ble views on the age-old ques­tion: “When should I think that I may be able to do some­thing un­usu­ally well?”

Th­ese two view­points tend to give wildly differ­ent, nearly cog­ni­tively nonover­lap­ping analy­ses of ques­tions like:

  • My doc­tor says I need to eat less and ex­er­cise, but a lot of ed­u­cated-sound­ing eco­nomics blog­gers are talk­ing about this thing called the “Shangri-La Diet.” They’re say­ing that in or­der to lose weight, all you need to do is con­sume large quan­tities of fla­vor­less, high-calorie foods at par­tic­u­lar times of day; and they claim some amaz­ing re­sults with this diet. Could they re­ally know bet­ter than my doc­tor? Would I be able to tell if they did?

  • My day job is in ar­tifi­cial in­tel­li­gence and de­ci­sion the­ory. And I re­call the dark days be­fore 2015, when there was plenty of effort and at­ten­tion go­ing into ad­vanc­ing the state of the art in AI ca­pa­bil­ities, but al­most none go­ing into AI al­ign­ment: bet­ter un­der­stand­ing AI de­signs and goals that can safely scale with ca­pa­bil­ities. Though in­ter­est in the al­ign­ment prob­lem has since in­creased quite a bit, it still makes sense to ask whether at the time I should have in­ferred from the lack of aca­demic ac­tivity that there was no pro­duc­tive work to be done here; since if there were reach­able fruits, wouldn’t aca­demics be tak­ing them?

  • Should I try my hand at be­com­ing an en­trepreneur? Whether or not it should be difficult to spot promis­ing ideas in a sci­en­tific field, it cer­tainly can’t be easy to think up a prof­itable idea for a new startup. Will I be able to find any good ideas that aren’t already taken?

  • The effec­tive al­tru­ism com­mu­nity is a net­work of philan­thropists and re­searchers that try to find the very best ways to benefit oth­ers per dol­lar, in full gen­er­al­ity. Where should effec­tive al­tru­ism or­ga­ni­za­tions like GiveWell ex­pect to find low-hang­ing fruit—ne­glected in­ter­ven­tions ripe with po­ten­tial? Where should they look to find things that our civ­i­liza­tion isn’t already do­ing about as well as can be done?

When I think about prob­lems like these, I use what feels to me like a nat­u­ral gen­er­al­iza­tion of the eco­nomic idea of effi­cient mar­kets. The goal is to pre­dict what kinds of effi­ciency we should ex­pect to ex­ist in realms be­yond the mar­ket­place, and what we can de­duce from sim­ple ob­ser­va­tions. For lack of a bet­ter term, I will call this kind of think­ing in­ad­e­quacy anal­y­sis.

Toward the end of this book, I’ll try to re­fute an al­ter­na­tive view­point that is in­creas­ingly pop­u­lar among some of my friends, one that I think is ill-founded. This view­point is the one I’ve pre­vi­ously termed “mod­esty,” and the mes­sage of mod­esty tends to be: “You can’t ex­pect to be able to do X that isn’t usu­ally done, since you could just be de­lud­ing your­self into think­ing you’re bet­ter than other peo­ple.”

I’ll open with a cherry-picked ex­am­ple that I think helps high­light the differ­ence be­tween these two view­points.


I once wrote a re­port, “In­tel­li­gence Ex­plo­sion Microe­co­nomics,” that called for an es­ti­mate of the eco­nomic growth rate in a fully de­vel­oped coun­try—that is, a coun­try that is no longer able to im­prove pro­duc­tivity just by im­port­ing well-tested in­no­va­tions. A foot­note of the pa­per re­marked that even though Ja­pan was the coun­try with the most ad­vanced tech­nol­ogy—e.g., their cel­l­phones and vir­tual re­al­ity tech­nol­ogy were five years ahead of the rest of the world’s—I wasn’t go­ing to use Ja­pan as my es­ti­ma­tor for de­vel­oped eco­nomic growth, be­cause, as I saw it, Ja­pan’s mon­e­tary policy was ut­terly de­ranged.

Roughly, Ja­pan’s cen­tral bank wasn’t cre­at­ing enough money. I won’t go into de­tails here.

A friend of mine, and one of the most care­ful thinkers I know—let’s call him “John”—made a com­ment on my draft to this effect:

How do you claim to know this? I can think of plenty of other rea­sons why Ja­pan could be in a slump: the coun­try’s shrink­ing and ag­ing pop­u­la­tion, its low fe­male work­place par­ti­ci­pa­tion, its high lev­els of product mar­ket reg­u­la­tion, etc. It looks like you’re ven­tur­ing out­side of your area of ex­per­tise to no good end.

“How do you claim to know this?” is a very rea­son­able ques­tion here. As John later elab­o­rated, macroe­co­nomics is an area where data sets tend to be thin and pre­dic­tive perfor­mance tends to be poor. And John had pre­vi­ously ob­served me mak­ing con­trar­ian claims where I’d turned out to be badly wrong, like en­dors­ing Gary Taubes’ the­o­ries about the causes of the obe­sity epi­demic. More re­cently, John won money off of me by bet­ting that AI perfor­mance on cer­tain met­rics would im­prove faster than I ex­pected; John has a good track record when it comes to spot­ting my mis­takes.

It’s also easy to imag­ine rea­sons an ob­server might have been skep­ti­cal. I wasn’t mak­ing up my cri­tique of Ja­pan my­self; I was read­ing other economists and de­cid­ing that I trusted the ones who were say­ing that the Bank of Ja­pan was do­ing it wrong… … Yet one would ex­pect the gov­ern­ing board of the Bank of Ja­pan to be com­posed of ex­pe­rienced economists with spe­cial­ized mon­e­tary ex­per­tise. How likely is it that any out­sider would be able to spot an ob­vi­ous flaw in their policy? How likely is it that some­one who isn’t a pro­fes­sional economist (e.g., me) would be able to judge which eco­nomic cri­tiques of the Bank of Ja­pan were cor­rect, or which crit­ics were wise?

How likely is it that an en­tire coun­try—one of the world’s most ad­vanced coun­tries—would forego trillions of dol­lars of real eco­nomic growth be­cause their mon­e­tary con­trol­lers—not poli­ti­ci­ans, but ap­poin­tees from the pro­fes­sional elite—were do­ing some­thing so wrong that even a non-pro­fes­sional could tell? How likely is it that a non-pro­fes­sional could not just sus­pect that the Bank of Ja­pan was do­ing some­thing badly wrong, but be con­fi­dent in that as­sess­ment?

Surely it would be more re­al­is­tic to search for pos­si­ble rea­sons why the Bank of Ja­pan might not be as stupid as it seemed, as stupid as some econ­blog­gers were claiming. Pos­si­bly Ja­pan’s ag­ing pop­u­la­tion made growth im­pos­si­ble. Pos­si­bly Ja­pan’s mas­sive out­stand­ing gov­ern­ment debt made even the slight­est in­fla­tion too dan­ger­ous. Pos­si­bly we just aren’t think­ing of the com­pli­cated rea­son­ing go­ing into the Bank of Ja­pan’s de­ci­sion.

Surely some hu­mil­ity is ap­pro­pri­ate when crit­i­ciz­ing the elite de­ci­sion-mak­ers gov­ern­ing the Bank of Ja­pan. What if it’s you, and not the pro­fes­sional economists mak­ing these de­ci­sions, who have failed to grasp the rele­vant eco­nomic con­sid­er­a­tions?

I’ll re­fer to this genre of ar­gu­ments as “mod­est episte­mol­ogy.”

In con­ver­sa­tion, John clar­ified to me that he re­jects this genre of ar­gu­ments; but I hear these kinds of ar­gu­ments fairly of­ten. The head of an effec­tive al­tru­ism or­ga­ni­za­tion once gave voice to what I would con­sider a good ex­am­ple of this mode of think­ing:

I find it helpful to ad­mit to un­pleas­ant facts that will nec­es­sar­ily be true in the ab­stract, in or­der to be more will­ing to ac­knowl­edge them in spe­cific cases. For in­stance, I should ex­pect a pri­ori to be be­low av­er­age at half of things, and be 50% likely to be of be­low av­er­age tal­ent over­all; to know many peo­ple who I re­gard as bet­ter than me ac­cord­ing to my val­ues; to reg­u­larly make de­ci­sions that look silly ex post, and also ex ante; to be mis­taken about is­sues on which there is ex­pert dis­agree­ment about half of the time; to perform badly at many things I at­tempt for the first time; and so on.

The Dun­ning-Kruger effect shows that un­skil­led in­di­vi­d­u­als of­ten rate their own skill very highly. Speci­fi­cally, al­though there does tend to be a cor­re­la­tion be­tween how com­pe­tent a per­son is and how com­pe­tent they guess they are, this cor­re­la­tion is weaker than one might sup­pose. In the origi­nal study, peo­ple in the bot­tom two quar­tiles of ac­tual test perfor­mance tended to think they did bet­ter than about 60% of test-tak­ers, while peo­ple in the top two quar­tiles tended to think they did bet­ter than 70% of test-tak­ers.

This sug­gests that a typ­i­cal per­son’s guesses about how they did on a test are ev­i­dence, but not par­tic­u­larly pow­er­ful ev­i­dence: the top quar­tile is un­der­con­fi­dent in how well they did, and the bot­tom quar­tiles are highly over­con­fi­dent.

Given all that, how can we gain much ev­i­dence from our be­lief that we are skil­led? Wouldn’t it be more pru­dent to re­mind our­selves of the base rate—the prior prob­a­bil­ity of 50% that we are be­low av­er­age?

Rea­son­ing along similar lines, soft­ware de­vel­oper Hal Fin­ney has en­dorsed “aban­don­ing per­sonal judg­ment on most mat­ters in fa­vor of the ma­jor­ity view.” Fin­ney notes that the av­er­age per­son’s opinions would be more ac­cu­rate (on av­er­age) if they sim­ply deferred to the most pop­u­lar po­si­tion on as many is­sues as they could. For this rea­son:

I choose to adopt the view that in gen­eral, on most is­sues, the av­er­age opinion of hu­man­ity will be a bet­ter and less bi­ased guide to the truth than my own judg­ment.

[…] I would sug­gest that al­though one might not always want to defer to the ma­jor­ity opinion, it should be the de­fault po­si­tion. Rather than start­ing with the as­sump­tion that one’s own opinion is right, and then look­ing to see if the ma­jor­ity has good rea­sons for hold­ing some other view, one should in­stead start off by fol­low­ing the ma­jor­ity opinion; and then only adopt a differ­ent view for good and con­vinc­ing rea­sons. On most is­sues, the de­fault of defer­ring to the ma­jor­ity will be the best ap­proach. If we ac­cept the prin­ci­ple that “ex­traor­di­nary claims re­quire ex­traor­di­nary ev­i­dence”, we should de­mand a high de­gree of jus­tifi­ca­tion for de­part­ing from the ma­jor­ity view. The mere fact that our own opinion seems sound would not be enough.1

In this way, Fin­ney hopes to cor­rect for over­con­fi­dence and ego­cen­tric bi­ases.

Fin­ney’s view is an ex­treme case, but helps illus­trate a pat­tern that I be­lieve can be found in some more mod­er­ate and widely en­dorsed views. When I speak of “mod­esty,” I have in mind a fairly di­verse set of po­si­tions that rest on a similar set of ar­gu­ments and mo­ti­va­tions.

I once heard an Oxford effec­tive al­tru­ism pro­po­nent crisply sum­ma­rize what I take to be the cen­tral ar­gu­ment for this per­spec­tive: “You see that some­one says X, which seems wrong, so you con­clude their epistemic stan­dards are bad. But they could just see that you say Y, which sounds wrong to them, and con­clude your epistemic stan­dards are bad.”2 On this line of think­ing, you don’t get any in­for­ma­tion about who has bet­ter epistemic stan­dards merely by ob­serv­ing that some­one dis­agrees with you. After all, the other side ob­serves just the same fact of dis­agree­ment.

Ap­ply­ing this ar­gu­ment form to the Bank of Ja­pan ex­am­ple: I re­ceive lit­tle or no ev­i­dence just from ob­serv­ing that the Bank of Ja­pan says “X” when I be­lieve “not X.” I also can’t be get­ting strong ev­i­dence from any ob­ject-level im­pres­sion I might have that I am un­usu­ally com­pe­tent. So did my pri­ors im­ply that I and I alone ought to have been born with awe­some pow­ers of dis­cern­ment? (Modest peo­ple have posed this ex­act ques­tion to me on more than one oc­ca­sion.)

It should go with­out say­ing that this isn’t how I would ex­plain my own rea­son­ing. But if I re­ject ar­gu­ments of the form, “We dis­agree, there­fore I’m right and you’re wrong,” how can I claim to be cor­rect on an eco­nomic ques­tion where I dis­agree with an in­sti­tu­tion as rep­utable as the Bank of Ja­pan?

The other view­point, op­posed to mod­esty—the view that I think is pre­scribed by nor­ma­tive episte­mol­ogy (and also by more or less main­stream microe­co­nomics)—re­quires a some­what longer in­tro­duc­tion.


By an­cient tra­di­tion, ev­ery ex­pla­na­tion of the Effi­cient Mar­kets Hy­poth­e­sis must open with the fol­low­ing joke:

Two economists are walk­ing along the street, and one says, “Hey, some­one dropped a $20 bill!” and the other says, “Well, it can’t be a real $20 bill be­cause some­one would have picked it up already.”

Also by an­cient tra­di­tion, the next step of the ex­pla­na­tion is to re­mark that while it may make sense to pick up a $20 bill you see on a rel­a­tively de­serted street, if you think you have spot­ted a $20 bill ly­ing on the floor of Grand Cen­tral Sta­tion (the main sub­way nexus of New York City), and it has stayed there for sev­eral hours, then it prob­a­bly is a fake $20 bill, or it has been glued to the ground.

In real life, when I asked a group of twenty rel­a­tively young peo­ple how many of them had ever found a $20 bill on the street, five raised their hands, and only one per­son had found a $20 bill on the street on two sep­a­rate oc­ca­sions. So the em­piri­cal truth about the joke is that while $20 bills on the street do ex­ist, they’re rare.

On the other hand, the im­plied policy is that if you do find a $20 bill on the street, you should go ahead and pick it up, be­cause that does hap­pen. It’s not that rare. You cer­tainly shouldn’t start ag­o­niz­ing over whether it’s too ar­ro­gant to be­lieve that you have bet­ter eye­sight than ev­ery­one else who has re­cently walked down the street.

On the other other hand, you should start ag­o­niz­ing about whether to trust your own men­tal pro­cesses if you think you’ve seen a $20 bill stay put for sev­eral hours on the floor of Grand Cen­tral Sta­tion. Espe­cially if your ex­pla­na­tion is that no­body else is ea­ger for money.

Is there any other do­main such that if we think we see an ex­ploitable pos­si­bil­ity, we should sooner doubt our own men­tal com­pe­tence than trust the con­clu­sion we rea­soned our way to?

If I had to name the sin­gle epistemic feat at which mod­ern hu­man civ­i­liza­tion is most ad­e­quate, the peak of all hu­man power of es­ti­ma­tion, I would un­hesi­tat­ingly re­ply, “Short-term rel­a­tive pric­ing of liquid fi­nan­cial as­sets, like the price of S&P 500 stocks rel­a­tive to other S&P 500 stocks over the next three months.” This is some­thing into which hu­man civ­i­liza­tion puts an ac­tual effort.

  • Millions of dol­lars are offered to smart, con­scien­tious peo­ple with physics PhDs to in­duce them to en­ter the field.

  • Th­ese peo­ple are then offered huge ad­di­tional pay­outs con­di­tional on ac­tual perfor­mance—es­pe­cially out­perfor­mance rel­a­tive to a baseline.3

  • Large cor­po­ra­tions form to spe­cial­ize in nar­row as­pects of price-tun­ing.

  • They have enor­mous com­put­ing clusters, vast his­tor­i­cal datasets, and com­pe­tent ma­chine learn­ing pro­fes­sion­als.

  • They re­ceive re­peated news of suc­cess or failure in a fast feed­back loop.4

  • The knowl­edge ag­gre­ga­tion mechanism—namely, prices that equil­ibrate sup­ply and de­mand for the fi­nan­cial as­set—has proven to work beau­tifully, and acts to sum up the wis­dom of all those highly mo­ti­vated ac­tors.

  • An ac­tor that spots a 1% sys­tem­atic er­ror in the ag­gre­gate es­ti­mate is re­warded with a billion dol­lars—in a pro­cess that also cor­rects the es­ti­mate.

  • Bar­ri­ers to en­try are not zero (you can’t get the loans to make a billion-dol­lar cor­rec­tive trade), but there are thou­sands of di­verse in­tel­li­gent ac­tors who are all in­di­vi­d­u­ally al­lowed to spot er­rors, cor­rect them, and be re­warded, with no cen­tral veto.

This is cer­tainly not perfect, but it is liter­ally as good as it gets on mod­ern-day Earth.

I don’t think I can beat the es­ti­mates pro­duced by that pro­cess. I have no sig­nifi­cant help to con­tribute to it. With study and effort I might be­come a de­cent hedge fundie and make a stan­dard re­turn. The­o­ret­i­cally, a liquid mar­ket should be just ex­ploitable enough to pay com­pe­tent pro­fes­sion­als the same hourly rate as their next-best op­por­tu­nity. I could po­ten­tially be­come one of those pro­fes­sion­als, and earn stan­dard hedge-fundie re­turns, but that’s not the same as sig­nifi­cantly im­prov­ing on the mar­ket’s effi­ciency. I’m not sure I ex­pect a huge hu­manly ac­cessible op­por­tu­nity of that kind to ex­ist, not in the thickly traded cen­ters of the mar­ket. Some­body re­ally would have taken it already! Our civ­i­liza­tion cares about whether Microsoft stock will be priced at $37.70 or $37.75 to­mor­row af­ter­noon.

I can’t pre­dict a 5% move in Microsoft stock in the next two months, and nei­ther can you. If your un­cle tells an anec­dote about how he tripled his in­vest­ment in last year and he at­tributes this to his skill rather than luck, we know im­me­di­ately and out of hand that he is wrong. War­ren Buffett at the peak of his form couldn’t re­li­ably triple his money ev­ery year. If there is a strat­egy so sim­ple that your un­cle can un­der­stand it, which has ap­par­ently made him money—then we guess that there were just hid­den risks built into the strat­egy, and that in an­other year or with less fa­vor­able events he would have lost half as much as he gained. Any other pos­si­bil­ity would be the equiv­a­lent of a $20 bill stay­ing on the floor of Grand Cen­tral Sta­tion for ten years while a horde of physics PhDs searched for it us­ing naked eyes, micro­scopes, and ma­chine learn­ing.

In the thickly traded parts of the stock mar­ket, where the col­lec­tive power of hu­man civ­i­liza­tion is truly at its strongest, I doff my hat, I put aside my pride and kneel in true hu­mil­ity to ac­cept the mar­ket’s be­liefs as though they were my own, know­ing that any im­pulse I feel to sec­ond-guess and ev­ery in­de­pen­dent thought I have to ar­gue oth­er­wise is noth­ing but my own folly. If my per­cep­tions sug­gest an ex­ploitable op­por­tu­nity, then my per­cep­tions are far more likely mis­taken than the mar­kets. That is what it feels like to look upon a civ­i­liza­tion do­ing some­thing ad­e­quately.

The con­verse side of the effi­cient-mar­kets per­spec­tive would have said this about the Bank of Ja­pan:

con­ven­tional cyn­i­cal economist: So, Eliezer, you think you know bet­ter than the Bank of Ja­pan and many other cen­tral banks around the world, do you?

eliezer: Yep. Or rather, by read­ing econ­blogs, I be­lieve my­self to have iden­ti­fied which econ­blog­gers know bet­ter, like Scott Sum­ner.

c.c.e.: Even though liter­ally trillions of dol­lars of real value are at stake?

eliezer: Yep.

c.c.e.: How do you make money off this spe­cial knowl­edge of yours?

eliezer: I can’t. The mar­ket also col­lec­tively knows that the Bank of Ja­pan is pur­su­ing a bad mon­e­tary policy and has priced Ja­panese equities ac­cord­ingly. So even though I know the Bank of Ja­pan’s policy will make Ja­panese equities perform badly, that fact is already priced in; I can’t ex­pect to make money by short-sel­l­ing Ja­panese equities.

c.c.e.: I see. So ex­actly who is it, on this the­ory of yours, that is be­ing stupid and pass­ing up a pre­dictable pay­out?

eliezer: No­body, of course! Only the Bank of Ja­pan is al­lowed to con­trol the trend line of the Ja­panese money sup­ply, and the Bank of Ja­pan’s gov­er­nors are not paid any bonuses when the Ja­panese econ­omy does bet­ter. They don’t get a mil­lion dol­lars in per­sonal bonuses if the Ja­panese econ­omy grows by a trillion dol­lars.

c.c.e.: So you can’t make any money off know­ing bet­ter in­di­vi­d­u­ally, and no­body who has the ac­tual power and au­thor­ity to fix the prob­lem would gain a per­sonal fi­nan­cial benefit from fix­ing it? Then we’re done! No anoma­lies here; this sounds like a perfectly nor­mal state of af­fairs.

We don’t usu­ally ex­pect to find $20 bills ly­ing on the street, be­cause even though peo­ple some­times drop $20 bills, some­one else will usu­ally have a chance to pick up that $20 bill be­fore we do.

We don’t think we can pre­dict 5% price changes in S&P 500 com­pany stock prices over the next month, be­cause we’re com­pet­ing against dozens of hedge fund man­agers with enor­mous su­per­com­put­ers and physics PhDs, any one of whom could make mil­lions or billions on the pric­ing er­ror—and in do­ing so, cor­rect that er­ror.

We can ex­pect it to be hard to come up with a truly good startup idea, and for even the best ideas to in­volve sweat and risk, be­cause lots of other peo­ple are try­ing to think up good startup ideas. Though in this case we do have the ad­van­tage that we can pick our own bat­tles, seek out one good idea that we think hasn’t been done yet.

But the Bank of Ja­pan is just one com­mit­tee, and it’s not pos­si­ble for any­one else to step up and make a billion dol­lars in the course of cor­rect­ing their er­ror. Even if you think you know ex­actly what the Bank of Ja­pan is do­ing wrong, you can’t make a profit on that. At least some hedge-fund man­agers also know what the Bank of Ja­pan is do­ing wrong, and the ex­pected con­se­quences are already priced into the mar­ket. Nor does this price move­ment fix the Bank of Ja­pan’s mis­taken be­hav­ior. So to the ex­tent the Bank of Ja­pan has poor in­cen­tives or some other sys­tem­atic dys­func­tion, their mis­take can per­sist. As a con­se­quence, when I read some econ­blog­gers who I’d seen be­ing right about em­piri­cal pre­dic­tions be­fore say­ing that Ja­pan was be­ing grotesquely silly, and the eco­nomic logic seemed to me to check out, as best I could fol­low it, I wasn’t par­tic­u­larly re­luc­tant to be­lieve them. Stan­dard eco­nomic the­ory, gen­er­al­ized be­yond the mar­kets to other facets of so­ciety, did not seem to me to pre­dict that the Bank of Ja­pan must act wisely for the good of Ja­pan. It would be no sur­prise if they were com­pe­tent, but also not much of a sur­prise if they were in­com­pe­tent. And know­ing this didn’t help me ei­ther—I couldn’t ex­ploit the knowl­edge to make an ex­cess profit my­self—and this too wasn’t a co­in­ci­dence.

This kind of think­ing can get quite a bit more com­pli­cated than the fore­go­ing para­graphs might sug­gest. We have to ask why the gov­ern­ment of Ja­pan didn’t put pres­sure on the Bank of Ja­pan (an­swer: they did, but the Bank of Ja­pan re­fused), and many other ques­tions. You would need to con­sider a much larger model of the world, and bring in a lot more back­ground the­ory, to be con­fi­dent that you un­der­stood the over­all situ­a­tion with the Bank of Ja­pan.

But even with­out that de­tailed anal­y­sis, in the episte­molog­i­cal back­ground we have a com­pletely differ­ent pic­ture from the mod­est one. We have a pic­ture of the world where it is perfectly plau­si­ble for an econ­blog­ger to write up a good anal­y­sis of what the Bank of Ja­pan is do­ing wrong, and for a so­phis­ti­cated reader to rea­son­ably agree that the anal­y­sis seems de­ci­sive, with­out a deep ag­o­niz­ing epi­sode of Dun­ning-Kruger-in­spired self-doubt play­ing any im­por­tant role in the anal­y­sis.


When we cri­tique a gov­ern­ment, we don’t usu­ally get to see what would ac­tu­ally hap­pen if the gov­ern­ment took our ad­vice. But in this one case, less than a month af­ter my ex­change with John, the Bank of Ja­pan—un­der the new lead­er­ship of Haruhiko Kuroda, and un­der un­prece­dented pres­sure from re­cently elected Prime Minister Shinzo Abe, who in­cluded mon­e­tary policy in his cam­paign plat­form—em­barked on an at­tempt to print huge amounts of money, with a stated goal of dou­bling the Ja­panese money sup­ply.5

Im­me­di­ately af­ter, Ja­pan ex­pe­rienced real GDP growth of 2.3%, where the pre­vi­ous trend was for fal­ling RGDP. Their econ­omy was op­er­at­ing that far un­der ca­pac­ity due to lack of money.6

Now, on the mod­est view, this was the un­fairest test imag­in­able. Out of all the times that I’ve ever sug­gested that a gov­ern­ment’s policy is sub­op­ti­mal, the rare time a gov­ern­ment tries my preferred al­ter­na­tive will se­lect the most main­stream, high­est-con­ven­tional-pres­tige poli­cies I hap­pen to ad­vo­cate, and those are the very policy pro­pos­als that mod­esty is least likely to dis­ap­prove of.

In­deed, if John had looked fur­ther into the is­sue, he would have found (as I found while writ­ing this) that No­bel lau­re­ates had also crit­i­cized Ja­pan’s mon­e­tary policy. He would have found that pre­vi­ous Ja­panese gov­ern­ments had also hinted to the Bank of Ja­pan that they should print more money. The view from mod­esty looks at this state of af­fairs and says, “Hold up! You aren’t so spe­cially blessed as your pri­ors would have you be­lieve; other aca­demics already know what you know! Civ­i­liza­tion isn’t so in­ad­e­quate af­ter all! This is how rea­son­able dis­sent from es­tab­lished in­sti­tu­tions and ex­perts op­er­ates in the real world: via op­po­si­tion by other main­stream ex­perts and in­sti­tu­tions, not via the heroic effort of a lone eco­nomics blog­ger.”

How­ever helpful or un­helpful such re­marks may be for guard­ing against in­flated pride, how­ever, they don’t seem to re­fute (or even ad­dress) the cen­tral the­sis of civ­i­liza­tional in­ad­e­quacy, as I will define that term later. Roughly, the civ­i­liza­tional in­ad­e­quacy the­sis states that in situ­a­tions where the cen­tral bank of a ma­jor de­vel­oped democ­racy is car­ry­ing out a policy, and a num­ber of highly re­garded economists like Ben Ber­nanke have writ­ten pa­pers about what that cen­tral bank is do­ing wrong, and there are widely ac­cepted macroe­co­nomic the­o­ries for un­der­stand­ing what that cen­tral bank is do­ing wrong, and the gov­ern­ment of the coun­try has tried to put pres­sure on the cen­tral bank to stop do­ing it wrong, and liter­ally trillions of dol­lars in real wealth are at stake, then the over­all com­pe­tence of hu­man civ­i­liza­tion is such that we shouldn’t be sur­prised to find the pro­fes­sional economists at the Bank of Ja­pan do­ing it wrong.

We shouldn’t even be sur­prised to find that a de­ci­sion the­o­rist with­out all that much back­ground in eco­nomics can iden­tify which econ­blog­gers have cor­rectly stated what the Bank of Ja­pan is do­ing wrong, or which sim­ple im­prove­ments to their cur­rent poli­cies would im­prove the situ­a­tion.


It doesn’t make much differ­ence to my life whether I un­der­stand mon­e­tary policy bet­ter than, say, the Euro­pean Cen­tral Bank, which as of late 2015 was re­peat­ing the same text­book mis­take as the Bank of Ja­pan and caus­ing trillions of eu­ros of dam­age to the Euro­pean econ­omy. In­so­far as I have other Euro­pean friends in coun­tries like Italy, it might be im­por­tant to them to know that Europe’s econ­omy is prob­a­bly not go­ing to get any bet­ter soon; or the knowl­edge might be rele­vant to pre­dict­ing AI progress timelines to know whether Ja­pan ran out of low-hang­ing tech­nolog­i­cal fruit or just had bad mon­e­tary policy. But that’s a rather dis­tant rele­vance, and for most of my read­ers I would ex­pect this is­sue to be even less rele­vant to their lives.

But you run into the same im­plicit back­ground ques­tions of in­ad­e­quacy anal­y­sis when, for ex­am­ple, you’re mak­ing health care de­ci­sions. Cherry-pick­ing an­other anec­dote: My wife has a se­vere case of Sea­sonal Affec­tive Di­sor­der. As of 2014, she’d tried sit­ting in front of a lit­tle light­box for an hour per day, and it hadn’t worked. SAD’s effects were crip­pling enough for it to be worth our time to con­sider ex­treme op­tions, like her spend­ing time in South Amer­ica dur­ing the win­ter months. And in­deed, va­ca­tion­ing in Chile and re­ceiv­ing more ex­po­sure to ac­tual sun­light did work, where light­boxes failed.

From my per­spec­tive, the ob­vi­ous next thought was: “Em­piri­cally, dinky lit­tle light­boxes don’t work. Em­piri­cally, the Sun does work. Next step: more light. Fill our house with more lu­mens than light­boxes provide.” In short or­der, I had strung up sixty-five 60W-equiv­a­lent LED bulbs in the liv­ing room, and an­other sixty-five in her bed­room.

Ah, but should I as­sume that my civ­i­liza­tion is be­ing op­por­tunis­tic about seek­ing out ways to cure SAD, and that if putting up 130 LED light bulbs of­ten worked when light­boxes failed, doc­tors would already know about that? Should the fact that putting up 130 light bulbs isn’t a well-known next step af­ter light­boxes con­vince me that my bright idea is prob­a­bly not a good idea, be­cause if it were, ev­ery­one would already be do­ing it? Should I con­clude from my in­abil­ity to find any pub­lished stud­ies on the In­ter­net test­ing this ques­tion that there is some fatal flaw in my plan that I’m just not see­ing?

We might call this ar­gu­ment “Ch­ester­ton’s Ab­sence of a Fence.” The thought be­ing: I shouldn’t build a fence here, be­cause if it were a good idea to have a fence here, some­one would already have built it. The un­der­ly­ing ques­tion here is: How strongly should I ex­pect that this ex­tremely com­mon med­i­cal prob­lem has been thor­oughly con­sid­ered by my civ­i­liza­tion, and that there’s noth­ing new, effec­tive, and un­con­ven­tional that I can per­son­ally im­pro­vise?

Eye­bal­ling this ques­tion, my off-the-cuff an­swer—based mostly on the im­pres­sions re­lated to me by ev­ery friend of mine who has ever dealt with medicine on a re­search level—is that I wouldn’t nec­es­sar­ily ex­pect any med­i­cal re­searcher ever to have done a for­mal ex­per­i­ment on the first thought that popped into my mind for treat­ing this ex­tremely com­mon de­pres­sive syn­drome. Nor would I strongly ex­pect the in­ter­ven­tion, if ini­tial tests found it to be effec­tive, to have re­ceived enough at­ten­tion that I could Google it.

But this is just my per­sonal take on the ad­e­quacy of 21st-cen­tury med­i­cal re­search. Should I be ner­vous that this line of think­ing is just an ex­cuse? Should I fret about the ap­par­ently high es­ti­mate of my own com­pe­tence im­plied by my think­ing that I could find an ob­vi­ous-seem­ing way to rem­edy SAD when trained doc­tors aren’t talk­ing about it and I’m not a med­i­cal re­searcher? Am I go­ing too far out­side my own area of ex­per­tise and start­ing to think that I’m good at ev­ery­thing?

In prac­tice, I didn’t bother go­ing through an ag­o­niz­ing fit of self-doubt along those lines. The sys­tem­atic com­pe­tence of hu­man civ­i­liza­tion with re­spect to treat­ing mood di­s­or­ders wasn’t so ap­par­ent to me that I con­sid­ered it a bet­ter use of re­sources to quietly drop the is­sue than to just lay down the ~$600 needed to test my sus­pi­cion. So I went ahead and ran the ex­per­i­ment. And as of early 2017, with two win­ters come and gone, Brienne seems to no longer have crip­pling SAD—though it took a lot of light bulbs, in­clud­ing light bulbs in her bed­room that had to be timed to go on at 7:30am be­fore she woke up, to sus­tain the ap­par­ent cure.7

If you want to out­perform—if you want to do any­thing not usu­ally done—then you’ll need to con­cep­tu­ally di­vide our civ­i­liza­tion into ar­eas of lower and greater com­pe­tency. My view is that this is best done from a frame­work of in­cen­tives and the equil­ibria of those in­cen­tives—which is to say, from the stand­point of microe­co­nomics. This is the main topic I’ll cover here.

In the pro­cess, I will also make the case that mod­esty—the part of this pro­cess where you go into an ag­o­niz­ing fit of self-doubt—isn’t ac­tu­ally helpful for figur­ing out when you might out­perform some as­pect of the equil­ibrium.

But one should ini­tially pre­sent a pos­i­tive agenda in dis­cus­sions like these—say­ing first what you think is the cor­rect episte­mol­ogy, be­fore in­veigh­ing against a po­si­tion you think is wrong.

So with­out fur­ther ado, in the next chap­ter I shall pre­sent a very sim­ple frame­work for in­ad­e­quate equil­ibria.

Next chap­ter: An Equil­ibrium of No Free En­ergy.

The full book will be available Novem­ber 16th. You can go to equil­ibri­ to pre-or­der the book, or sign up for no­tifi­ca­tions about new chap­ters and other de­vel­op­ments.

  1. See Fin­ney, “Philo­soph­i­cal Ma­jori­tar­i­anism.”

  2. Note: They later said that I’d mi­s­un­der­stood their in­tent, so take this ex­am­ple with some grains of salt.

  3. This is why I speci­fied rel­a­tive prices: stock-trad­ing pro­fes­sion­als are usu­ally graded on how well they do com­pared to the stock mar­ket, not com­pared to bonds. It’s much less ob­vi­ous that bonds in gen­eral are priced rea­son­ably rel­a­tive to stocks in gen­eral, though this is still be­ing de­bated by economists.

  4. This is why I speci­fied near-term pric­ing of liquid as­sets.

  5. That is, the Bank of Ja­pan pur­chased huge num­bers of bonds with newly cre­ated elec­tronic money.

  6. See “How Ja­pan Proved Print­ing Money Can Be A Great Idea” for a more re­cent up­date.

    For read­ers who are won­der­ing, “Wait, how the heck can print­ing money pos­si­bly lead to real goods and ser­vices be­ing cre­ated?” I sug­gest Googling “sticky wages” and pos­si­bly con­sult­ing Scott Sum­ner’s his­tory of the Great De­pres­sion, The Mi­das Para­dox.

  7. Speci­fi­cally, Brienne’s symp­toms were mostly cured in the win­ter of 2015, and par­tially cured in the win­ter of 2016, when she spent most of her time un­der fewer lights. Brienne re­ports that she suffered a lot less even in the more re­cent win­ter, and ex­pe­rienced no suici­dal ideation, un­like in years prior to the light ther­apy.

    I’ll be mod­er­ately sur­prised if this treat­ment works re­li­ably, just be­cause most things don’t where de­pres­sion is con­cerned; but I would pre­dict that it works of­ten enough to be worth try­ing for other peo­ple ex­pe­rienc­ing se­vere treat­ment-re­sis­tant SAD.