Simple Rules of Law

Link post

Re­sponse To: Who Likes Sim­ple Rules?

Epistemic Sta­tus: Work­ing through ex­am­ples with vary­ing de­grees of con­fi­dence, to help us be con­crete and even­tu­ally gen­er­al­ize.

Robin Han­son has, in his words, “some puz­zles” that I will be an­a­lyz­ing. I’ve added let­ters for refer­ence.

  • A] Peo­ple are of­ten okay with hav­ing ei­ther policy A or policy B adopted as the stan­dard policy for all cases. But then they ob­ject greatly to a policy of ran­domly pick­ing A or B in par­tic­u­lar cases in or­der to find out which one works bet­ter, and then adopt it for ev­ery­one.
  • B] Peo­ple don’t like speed and red-light cam­eras; they pre­fer hu­man cops who will use dis­cre­tion. On av­er­age peo­ple don’t think that speed­ing en­force­ment dis­cre­tion will be used to benefit so­ciety, but 3 out of 4 ex­pect that it will benefit them per­son­ally. More gen­er­ally peo­ple seem to like a crime law sys­tem where at least a dozen differ­ent peo­ple are au­tho­rized to in effect par­don any given per­son ac­cused of any given crime; most peo­ple ex­pect to benefit per­son­ally from such dis­cre­tion.

  • C] In many Euro­pean na­tions cit­i­zens send their tax info into the gov­ern­ment who then tells them how much tax they owe. But in the US and many other na­tions, too many peo­ple op­pose this policy. The most vo­cal op­po­nents think they benefit per­son­ally from be­ing able to pay less than what the gov­ern­ment would say they owe.

  • D] The Bri­tish Na­tional Health Ser­vice gets a lot of crit­i­cism from choos­ing treat­ments by es­ti­mat­ing their cost per qual­ity-ad­justed-life-year. US folks wouldn’t tol­er­ate such a policy. Crit­ics lob­by­ing to get ex­cep­tional treat­ment say things like “one can­not as­sume that some­one who is wheel-chair bound can­not live as or more hap­pily. … [set] both false limits on health­care and re­duc­ing free­dom of choice. … re­flects an overly util­i­tar­ian ap­proach”

  • E] There’s long been op­po­si­tion to us­ing an offi­cial value of life pa­ram­e­ter in de­cid­ing gov­ern­ment poli­cies. Juries have also severely pun­ished firms for us­ing such pa­ram­e­ters to make firm de­ci­sions.

  • F] In aca­demic de­part­ments like mine, we tell new pro­fes­sors that to get tenure they need to pub­lish enough pa­pers in good jour­nals. But we re­fuse to say how many is enough or which jour­nals count as how good. We’d keep the flex­i­bil­ity to make what­ever de­ci­sion we want at the last minute.

  • G] Peo­ple who hire lawyers rarely know their track record and win­ning vs. los­ing court cases. The info is pub­lic, but so few are in­ter­ested that it is rarely col­lected or con­sulted. Peo­ple who hire do know the pres­tige of their schools and em­ploy­ers, and de­cide based on that.

  • H] When gov­ern­ment leases its land to pri­vate par­ties, some­times it uses cen­tral­ized, for­mal mechanisms, like auc­tions, and some­times it uses de­cen­tral­ized and in­for­mal mechanisms. Peo­ple seem to in­tu­itively pre­fer the lat­ter sort of mechanism, even though the former seems to works bet­ter. In one study “auc­tioned leases gen­er­ate 67% larger up-front pay­ments … [and were] 44% more pro­duc­tive”.

  • I] Peo­ple con­sis­tently in­vest in man­aged in­vest­ment funds, which af­ter the man­age­ment fee con­sis­tently re­turn less than in­dex funds, which fol­low a sim­ple clear rule. In­vestors seem to en­joy brag­ging about per­sonal con­nec­tions to peo­ple run­ning pres­ti­gious in­vest­ment funds.

  • J] When firms go pub­lic via an IPO, they typ­i­cally pay a bank 7% of their value to man­age the pro­cess, which is sup­pos­edly spent on lob­by­ing oth­ers to buy. Google fa­mously used an auc­tion to cut that fee, but banks have suc­ceed in squash­ing that re­bel­lion. When firms try to sell them­selves to other firms to ac­quire, they typ­i­cally pay 10% if they are priced at less than $1M, 6-8% if priced $10-30M, and 2-4% if priced over $100M.

  • K] Most elite col­leges de­cide who to ad­mit via opaque and fre­quently chang­ing crite­ria, crite­ria which al­low much dis­cre­tion by ad­mis­sions per­son­nel, and crite­ria about which some com­mu­ni­ties learn much more than oth­ers. Many elites learn to game such sys­tems to give their kids big ad­van­tages. While some com­plain, the sys­tem seems sta­ble.

  • L] In a Twit­ter poll, the main com­plaints about my fire-the-CEO de­ci­sions mar­kets pro­posal are that they don’t want a sim­ple clear me­chan­i­cal pro­cess to fire CEOs, and they don’t want to ex­plic­itly say that the firm makes such choices in or­der to max­i­mize prof­its. They in­stead want some peo­ple to have dis­cre­tion on CEO firing, and they want firm goals to be im­plicit and am­bigu­ous.

Some of these ex­am­ples don’t re­quire ex­pla­na­tion be­yond why they are bad ideas for rules. They wouldn’t work. Others would, or are even ob­vi­ously cor­rect, and re­quire more ex­pla­na­tion. I do think there is a enough of a pat­tern to be worth try­ing to ex­plain. Peo­ple re­li­ably dis­like the rule of law, and pre­fer to sub­sti­tute the rule of man.

Why do peo­ple op­pose the rule of law?

You can read his anal­y­sis in the origi­nal post. His main di­ag­no­sis is that peo­ple pro­mote dis­cre­tion for two rea­sons:

  1. They be­lieve they are un­usu­ally smart, at­trac­tive, charis­matic and well-liked, just the sort of peo­ple who tend to be fa­vored by in­for­mal dis­cre­tion.

  2. They want to sig­nal to oth­ers that they have con­fi­dence in elites who have dis­cre­tion, and that they ex­pect to benefit from that dis­cre­tion.

While I do think these are re­lated to some of the key rea­sons, I do not think these point at the cen­tral things go­ing on. Below I tackle all of these cases and their speci­fics. Over­all I think the fol­low­ing are the five key sto­ries:

  1. Good­hart’s Law. We see a lot of Re­gres­sional and Ad­ver­sar­ial Good­hart, and also some Ex­tremal and Causal as well. B, F, G, K and L all have large is­sues here.

  2. Copen­hagen In­ter­pre­ta­tion of Ethics. If you put a con­sid­er­a­tion ex­plic­itly into your rule, that makes you blame­wor­thy for in­ter­act­ing with it. And you’re pro­vid­ing all the in­for­ma­tion oth­ers need to scape­goat you not only for what you do, but what you would do in other sce­nar­ios, or why you are do­ing all of this. This is a fac­tor in A, B, D, E, F, H and K all have to worry about this.

  3. For­bid­den Con­sid­er­a­tions and Law­suit Pro­tec­tion. We can take the dan­ger of trans­parency a step fur­ther. There are things that one wants to take into con­sid­er­a­tion that are ille­gal to con­sider, or which you are con­sid­ered blame­wor­thy for con­sid­er­ing. The num­ber of ways one can be sued or boy­cotted for their de­ci­sion pro­cess goes up ev­ery year. You need a com­plex sys­tem in or­der to dis­guise hid­den fac­tors, and you cer­tainly can’t put them into ex­plicit rules. And in gen­eral, the less in­for­ma­tion oth­ers have, the less those that don’t like your de­ci­sion have to jus­tify a law­suit or mak­ing other trou­ble. B, D, E, F, K and L have this in play.

  4. Power. Discre­tion and com­plex­ity fa­vor those with power. If you have power, oth­ers worry that any­thing they do can in­fluence de­ci­sions, grant­ing you fur­ther power over ev­ery­thing in a vi­cious cy­cle. If you use a sim­ple rule, you give up your power, in­clud­ing over un­re­lated el­e­ments. Even if you don’t have power di­rectly over a de­ci­sion, any­one with power any­where can threaten to use that power when­ever any dis­cre­tion is available, so ev­ery­one with power by de­fault fa­vors com­plex dis­cre­tionary rules ev­ery­where. This shows up pretty much ev­ery­where other than A, but is es­pe­cially im­por­tant in B, F, H, K and L.

  5. Theft. Com­plex rules are of­ten noth­ing more than a method of ex­pro­pri­a­tion. This is es­pe­cially cen­tral in B, C, D, H, I and J.

I’ll now work through the ex­am­ples. I’ll start with C, since it seems out of place, then go in Robin’s or­der.

Long post is long. If a case doesn’t in­ter­est you, please do skip it, and it’s fine to stop here.

C] Tax Returns

The odd policy out here is C, the failure to have the gov­ern­ment tell cit­i­zens what it already knows about cit­i­zens’ tax re­turns. This re­sults in many lost hours search­ing down records, and much money lost pay­ing for tax prepa­ra­tion. As a com­men­ta­tor points out, the ar­gu­ment that ‘this al­lows one to pay less than they owe’ doesn’t ac­tu­ally make sense as an ex­pla­na­tion. The gov­ern­ment still knows what it knows, and still cross-checks that against what you say and pay. In other coun­tries one can still choose to ad­just the num­bers shared with you by the gov­ern­ment.

In the­ory, one could make a case similar to those I’ll make in other places, that tel­ling peo­ple what in­for­ma­tion the gov­ern­ment knows and doesn’t know al­lows peo­ple to hide any­thing that the gov­ern­ment doesn’t know about. But that seems quite minor.

What’s go­ing on here is sim­ple reg­u­la­tory cap­ture, cor­rup­tion, rent seek­ing and crim­i­nal theft. Robin’s link ex­plains this ex­plic­itly. Tax prepa­ra­tion cor­po­ra­tions like H&R Block are the pri­mary drivers here, be­cause the rules gen­er­ate more busi­ness for them. There is also a sec­ondary prob­lem that fa­nat­i­cal anti-tax con­ser­va­tives like that taxes an­noy peo­ple.

But I’ve never heard of a reg­u­lar per­son who thinks this policy is a good idea, and I never ex­pect to find one. We’re not this crazy. We have a dys­func­tional gov­ern­ment.

A] Ran­dom Experiments

Robin’s ex­pla­na­tions don’t fit case A. If you’re choos­ing ran­domly, no one can benefit from dis­cre­tion. If you choose the same thing for ev­ery­one, again no one can benefit from dis­cre­tion. If any­thing, the ran­dom sys­tem al­lows par­ti­ci­pants to po­ten­tially cheat or find a way around a se­lec­tion they dis­like, whereas a uni­ver­sal sys­tem makes this harder. Other things must be at work here.

This is the op­po­site of C, a case where peo­ple do op­pose the change, but the change would be ob­vi­ously good.

I call this the “too im­por­tant to know” prob­lem.

To me this is a clear case of The Copen­hagen In­ter­pre­ta­tion of Ethics and Asym­met­ric Jus­tice in­ter­act­ing with sa­cred val­ues.

An ex­per­i­ment in­ter­acts with the prob­lem and in par­tic­u­lar in­ter­acts with ev­ery sub­ject of the ex­per­i­ment, and with ev­ery po­ten­tial in­ter­ven­tion, in a way suffi­cient to ren­der you blame­wor­thy for not do­ing more, or not do­ing the op­ti­mal thing.

The con­trast be­tween the two cases is clear.

Without an ex­per­i­ment, we’re forced to make a choice be­tween op­tions A and B. Peo­ple mostly ac­cept that one must do their best, and po­ten­tially sa­cred val­ues are up against other po­ten­tially sa­cred val­ues, and one must guess and try their best.

In the cases in the study, it’s even more ex­treme. We’re choos­ing to im­ple­ment A or im­ple­ment B, in a place where nor­mally one would do noth­ing. So we’re com­par­ing do­ing some­thing about the situ­a­tion to do­ing noth­ing. It’s no sur­prise that ‘try to re­duce in­fec­tions’ comes out look­ing good.

With an ex­per­i­ment, the choice is be­tween ex­per­i­men­ta­tion and non-ex­per­i­men­ta­tion. You are choos­ing to pri­ori­tize in­for­ma­tion over all the sa­cred val­ues the non-ob­jec­tion­able choices are trad­ing off. Even if the two choices are fully non-ob­jec­tion­able, choos­ing be­tween them still means plac­ing the need to gather in­for­ma­tion over the needs of the peo­ple in the ex­per­i­ment.

The needs of spe­cific peo­ple are, ev­ery­where and always, a sa­cred value. Some­one, a par­tic­u­lar real per­son, is on the line. When put up against “in­for­ma­tion” what chance does this amor­phous in­for­ma­tion have?

Copen­hagen ex­plains why it is pointless to say that the ex­per­i­ment is bet­ter for these pa­tients than not run­ning one. Asym­met­ric Jus­tice ex­plains why the benefits to fu­ture pa­tients doesn’t make up for it.

There are other rea­sons, too.

Peo­ple don’t like their fates com­ing down to coin flips. They don’t like un­cer­tainty.

Peo­ple don’t like asym­me­try or in­equal­ity – if I get A and you get B, some­one got the bet­ter deal, and that’s not fair.

If you choose a par­tic­u­lar ac­tion, that pro­vides ev­i­dence that there was a rea­son to choose it. So peo­ple in­stinc­tively ad­just some for the fact that it was cho­sen. Whereas in an ex­per­i­ment, it’s clear you don’t know which choice is bet­ter (un­less you do know and are sim­ply out to prove it, in which case you are a mon­ster). That doesn’t in­spire con­fi­dence.

A fi­nal note is that if you look at the study in ques­tion, it sug­gests an­other im­por­tant el­e­ment. If you choose A, you’re blame­wor­thy for A and for ~B, but you’re cer­tainly not blame­wor­thy for ~A or for B! Whereas if you choose (50% A, 50% B) then you are blame­wor­thy for A, ~A, B and ~B, plus ex­per­i­men­ta­tion in gen­eral. That’s a lot of blame.

Re­mem­ber Asym­met­ric Jus­tice. If any el­e­ment of what you do is ob­jec­tion­able, ev­ery­thing you do, to­gether, is also ob­jec­tion­able. A sin­gle ‘prob­le­matic’ el­e­ment ru­ins all.

So if we look at Figure 1 in the study, we see in case C that the ob­jec­tion score for the A/​B test is ac­tu­ally be­low what we’d ex­pect if we thought the chances of ob­ject­ing to A and B were in­de­pen­dent, and peo­ple were ob­ject­ing to the ex­per­i­ment when­ever they dis­liked ei­ther A or B (or both). In cases B and D, we see only a small ad­di­tional rate of ob­jec­tion. It’s only in case A that we see sub­stan­tial ad­di­tional ob­jec­tion. Across the data given, it looks like this phe­nomenon ex­plains about half of the in­creased rate of ob­jec­tion to the ex­per­i­ments.

It also looks like a lot of peo­ple ex­plic­itly cited things like ‘play­ing with peo­ple’s lives’ via ex­per­i­ment, and ob­ject to ex­per­i­men­ta­tion as such at least when the stakes are high.

B] Po­lice Discretion

I do not think Robin’s story of ex­pec­ta­tion of per­sonal benefit is the cen­tral story here, ei­ther. The cor­re­la­tion isn’t even that high in his poll.

If po­lice have dis­cre­tion IN GENERAL re­gard­ing who they ar­rest, do you think they will on av­er­age use that dis­cre­tion to ar­rest those who ac­tu­ally do more net so­cial harm? Do you think that you will tend to be fa­vored by this dis­cre­tion?
49% Yes to both
13% No to both
10% Yes re net harm, No re me
28% No re net harm, Yes re me
— Robin Han­son (@robin­han­son) May 6, 2019

If you think net harm is re­duced (59%), you’re (49/​59) 84% to think you’ll benefit. If you think net harm is not re­duced, you are (28/​41) 68% to think you’ll benefit. Given that you’d ex­pect mod­els to give cor­re­lated re­turns to the two ques­tions – if dis­cre­tion is used wisely, it should tend to benefit both most peo­ple and a typ­i­cal civilian look­ing to avoid do­ing so­cial harm, and these are Robin’s Twit­ter fol­low­ers – I don’t think per­sonal mo­ti­va­tion is ex­plain­ing much var­i­ance here.

The ques­tion also ask­ing about only part of the pic­ture. Yes, we would hope that po­lice (and pros­e­cu­tors and oth­ers in the sys­tem) would use dis­cre­tion, at least in part, to ar­rest those who do more net so­cial harm over those who do less net so­cial harm.

But that’s far from the only goal of crim­i­nal jus­tice, or of pun­ish­ment. We would also hope that au­thor­i­ties would use dis­cre­tion to ac­com­plish other goals, as well.

Some of these goals are good. Others, not so much.

What are some other goals we might have? What are other rea­sons to use dis­cre­tion?

I think there are five broad rea­sons, be­yond us­ing it to judge so­cial harm.

  1. Discre­tion gives law en­force­ment au­thor­i­ties power. This power al­lows them to main­tain and ex­ert au­thor­ity. It keeps the sys­tem run­ning, en­sures co­op­er­a­tion and al­lows them to solve cases and en­force the law.

  2. Discre­tion gives law en­force­ment au­thor­i­ties power and sta­tus. This power and sta­tus is a large part of the com­pen­sa­tion we give to law en­force­ment.

  3. Discre­tion al­lows those with sta­tus and/​or power to gain ad­di­tional benefits from that sta­tus and/​or power.

  4. Discre­tion al­lows us to en­force rules with­out hav­ing to state those rules ex­plic­itly, be able to well-spec­ify those rules, or spec­ify them in ad­vance.

  5. Discre­tion guards against Good­hart’s Law and the gam­ing of the sys­tem.

To this we would add Robin’s ex­pla­na­tions, that one might want to benefit from this di­rectly, and/​or one might want to sig­nal sup­port for such au­thor­i­ties. And the more dis­cre­tion they have, the more one would want to sig­nal one’s sup­port – see rea­son 1.

A case worth grap­pling with can be made for or against each of these five jus­tifi­ca­tions be­ing net good. So one could ar­gue in fa­vor like this with ar­gu­ments like (I am not en­dors­ing these, nor do they nec­es­sar­ily ar­gue for to­day’s Amer­i­can level of dis­cre­tion):

  1. No one would co­op­er­ate with au­thor­i­ties. Every po­ten­tial wit­ness will stonewall, ev­ery defen­dant will fight to the end. In­timi­da­tion of wit­nesses would be im­pos­si­ble to stop. Good luck get­ting any­one even as­so­ci­ated with some­one do­ing any­thing shady to ever talk to the po­lice. Cases wouldn’t get solved, en­force­ment would break down. Crim­i­nals would use Black­mail to take con­trol of peo­ple and put them out­side the law. Author­i­ties’ hands would be tied and we’d be liv­ing in Bat­man’s Gotham.

  2. In many ways, be­ing a cop is a pretty ter­rible job. Who wants to con­stantly have hos­tile and po­ten­tially dan­ger­ous in­ter­ac­tions with crim­i­nals? Others hav­ing to ‘re­spect my au­thor­i­tah‘ im­proves those in­ter­ac­tions, and goes a long way in mak­ing up for things on many fronts.

  3. In­creas­ing re­turns for sta­tus and power in­crease the power of in­cen­tives. We want peo­ple to do things we ap­prove of, and strive to gain so­ciety’s ap­proval, so we need lev­ers to re­ward those who do, and pun­ish those who don’t. We want to re­ward those who walk the straight and nar­row track, and make good, and also those who strive and achieve. This has to bal­ance the lure of anti-so­cial or crim­i­nal ac­tivity, es­pe­cially in poorer com­mu­ni­ties. And it has to bal­ance the lure of in­ac­tivity, as we provide in­creas­ingly liv­able bas­kets of goods to those who opt out of sta­tus and pres­tige fights to play video games.

  4. If we state our rules ex­plic­itly, we are now blame­wor­thy for ev­ery pos­si­ble cor­ner case in which those rules re­sult in some­thing per­verse, and for ev­ery mo­ti­va­tion or fac­tor we state. Given that we can be con­demned for each of these via Asym­met­ric Jus­tice, this isn’t an op­tion. We also need to be able to im­ple­ment sys­tems with­out be­ing able to spec­ify them and their im­pli­ca­tions in an un­am­bigu­ous way. If we did go by the let­ter of the law, then we’re at the mercy of the mis­takes of the leg­is­la­ture, and the law would get even more com­plex than it is as we at­tempt to cod­ify ev­ery case, be­yond the reach of any per­son to know how it works. Hu­mans are much bet­ter at pro­cess­ing other types of com­plex sys­tems. And we need ways for lo­cal ar­eas to en­force norms and do lo­cally ap­pro­pri­ate things, with­out hav­ing their own leg­is­la­tures. If the sys­tem couldn’t cor­rect ob­vi­ous er­rors, both Type I and Type II, then it would lose the pub­lic trust. And so on.

  5. If peo­ple know ex­actly what is and isn’t ille­gal in ev­ery case, and what the pun­ish­ments are, then it will be open sea­son for any­thing you didn’t make ex­plic­itly ille­gal. Peo­ple will find tech­ni­cally per­mit­ted ways to do all sorts of bad things. They’ll op­ti­mize for all the wrong things. Even when they don’t, they’ll do the max­i­mum amount of ev­ery bad thing that your sys­tem doesn’t pun­ish enough to stop them from do­ing, and jus­tify it by say­ing ‘it was le­gal’. Your laws won’t ac­com­plish what they set out to ac­com­plish. Any com­plex sys­tem deal­ing with the messi­ness of the real world needs some way to en­force the ‘spirit of the law.’ One would be shocked if ex­ten­sive gam­ing of the sys­tem and epic Good­hart failures didn’t re­sult from a lack of dis­cre­tion.

Or, to take the op­po­site stances:

  1. Power cor­rupts. It is bad and should be min­i­mized. The sys­tem will use its power to pro­tect and gather more power for it­self, and rule for the benefit of its mem­bers and those pul­ling their strings, not you. If we didn’t have dis­cre­tion, we could min­i­mize the num­ber of laws and ev­ery­one wouldn’t need to go around afraid of au­thor­ity. And peo­ple would see that au­thor­ity was le­gi­t­i­mate, and would be in­clined to co­op­er­ate.

  2. Power not only cor­rupts, it also at­tracts the cor­rupt. If be­ing a cop lets you help peo­ple and stand for what’s right, and gives you a good and steady pay­check, you’ll get good peo­ple. If be­ing a cop is an oth­er­wise crappy job where you get to force peo­ple to ‘re­spect my au­thor­i­tah‘ and make bank with cor­rupt deals and spu­ri­ous over­time, you’ll get ex­actly the peo­ple you don’t want, with ex­actly the au­thor­ity you don’t want them to have.

  3. The last thing we want is to take the pow­er­ful (and hence cor­rupt) and give them even more dis­cre­tion and power, and less ac­countabil­ity. The in­evitable re­sult is in­creas­ing amounts of scape­goat­ing and ex­pro­pri­a­tion. This is how the lit­tle guy, and the one mind­ing their own busi­ness try­ing to pro­duce, get crushed by cor­po­ra­tions and petty tyrants of all sorts.

  4. If you can’t state what your rules are, how am I sup­posed to fol­low them? How can it be fair to pun­ish me for vi­o­lat­ing rules you won’t tell me? This isn’t rules at all, or rule of law. It’s rule of man, rule of power beget­ting power. The sys­tem is in­creas­ingly rigged, and you’re de­stroy­ing the records to pre­vent any­one from know­ing that you’re rig­ging it.

  5. The things you’re try­ing to pro­tect from be­ing gamed are power and the pow­er­ful, which are what are go­ing to de­ter­mine what the ‘spirit of the rules’ is go­ing to be. So the spirit be­comes what­ever is good for them. The point of rule of law, of par­tic­u­lar well-speci­fied rules, is to avoid this. Those rules are not sup­posed to be the only thing one cares about. Rather, the law is sup­posed to guard against spe­cific bad things, and other mechanisms al­lowed to take care of the rest. If the law is re­sult­ing in Good­hart is­sues, it’s a bad law and you should change it.

The best ar­gu­ments I know about against dis­cre­tion have noth­ing to do with the so­cial harm caused by pun­ished ac­tions. They are ar­gu­ments for rule of law, and to guard against what those with dis­cre­tion will do with that power. Th­ese effects are rather im­por­tant and prob­le­matic even when the sys­tem is work­ing as de­signed.

The best ar­gu­ments I know about in fa­vor of dis­cre­tion also have noth­ing to do with the so­cial harm caused by pun­ished ac­tions. They have to do with the sys­tem de­pend­ing on dis­cre­tion in or­der to be able to func­tion, and in or­der to en­sure co­op­er­a­tion. A sys­tem with­out dis­cre­tion by de­fault makes the spread of any lo­cal in­for­ma­tion ev­ery­one’s en­emy, and pro­vides no lev­er­age to over­come that. If we didn’t have dis­cre­tion, we would have to rad­i­cally re-ex­am­ine all of our laws and our en­tire sys­tem of en­force­ment, lest ev­ery­thing fall apart.

My model says that we cur­rently give au­thor­i­ties too much dis­cre­tion, and (partly) as a re­sult have pun­ish­ments that are too harsh. And also that the au­thor­i­ties have so much dis­cre­tion partly be­cause pun­ish­ments have been made too harsh. Since dis­cre­tion and large pun­ish­ments give those with power more power, it would be sur­pris­ing if this were not the case.

D] Na­tional Health Service

The Na­tional Health Ser­vice gets crit­i­cized con­stantly be­cause it is their job to deny peo­ple health care. There is not enough money to provide what we would think of as an ac­cept­able level of care un­der all cir­cum­stances, be­cause our con­cept of ac­cept­able level of care is all of the health care. In such a cir­cum­stance, there isn’t much they could do.

Us­ing de­ter­minis­tic rules based on num­bers is the ob­vi­ously cor­rect way to ra­tion care. Us­ing hu­man dis­cre­tion in each case will mean ei­ther always giv­ing out care, since the choice is be­tween care or no care – which is a lot of why health care costs are so high and go­ing higher – or not always giv­ing out care when able to do so, which will have peo­ple scream­ing un­usu­ally literal bloody mur­der.

Deter­minis­tic rules let in­di­vi­d­u­als avoid blame, and al­low health care bud­gets to be used at all. But that doesn’t mean peo­ple are go­ing to like it. If any­thing, they’re go­ing to be mad about both the rules and the fact that they don’t have a hu­man they can ei­ther blame or try to lev­er­age to get what they want. There’s also the is­sue of putting a value on hu­man life at all, which is bad enough but clearly un­avoid­able.

More than that, once you ex­plic­itly say what you value by putting num­bers on lives and im­prove­ments in qual­ity of life, you’re do­ing some­thing both com­pletely nec­es­sary and com­pletely un­ac­cept­able. The ex­am­ple of some­one in a wheelchair is pretty great. If you don’t provide some dis­count in value of qual­ity of life for phys­i­cal dis­abil­ity, then you are say­ing that phys­i­cal dis­abil­ities don’t de­crease qual­ity of life. Which has pretty ter­rible im­pli­ca­tions for a health care sys­tem try­ing to pre­vent phys­i­cal dis­abil­ities. If you do say they de­crease qual­ity of life, you’re say­ing peo­ple with dis­abil­ities have less value. There are tons of places like this.

Another way to view this is that the only way for one to make health care de­ci­sions to ra­tion care or oth­er­wise sac­ri­fice sa­cred val­ues to stay on bud­get, with­out blame, is to have all those de­ci­sions be seen as out of your con­trol and not your choice. The only known way to do that is to have a sys­tem in place, and point to that. That sys­tem then be­comes a way to not in­ter­act with the sys­tem, avoid­ing blame. Whereas propos­ing or con­sid­er­ing any other sys­tem in­volves in­ter­ac­tion, and thus blame.

E] Value of Life

If you are caught mak­ing a trade-off be­tween a sa­cred value (life) and a non-sa­cred value (money), it’s not go­ing to go well. Of course a com­pany do­ing an ex­plicit calcu­la­tion here is go­ing to get pun­ished, as is a gov­ern­ment policy mak­ing an ex­plicit com­par­i­son. Hu­mans don’t care about the tran­si­tive prop­erty.

Thus, firms and gov­ern­ments, who ob­vi­ously need to value risk to hu­man life at a high but finite nu­mer­i­cal cost, will need to do this with­out writ­ing the num­ber down ex­plic­itly in any way. This is one of the more silly things one can­not con­sider, that one ob­vi­ously must con­sider. In a world where we are blame­wor­thy (to the point of be­ing sued for mas­sive amounts) for do­ing ex­plicit calcu­la­tions that ac­knowl­edge trade-offs or im­por­tant facts about the world, firms and gov­ern­ments are forced to make their de­ci­sions in in­creas­ingly opaque ways. One of those opaque prefer­ences will be to fa­vor those who rely on opaque­ness and de­stroy records, and to get rid of any­one who is trans­par­ent about their think­ing or oth­er­wise, and keeps ac­cu­rate records.

F] Tenure Requirements

Tenure is about eval­u­at­ing what a po­ten­tial pro­fes­sor would bring to the uni­ver­sity. No mat­ter what ex­tent poli­tics gets in­volved, this is some­one you’ll have to work with for decades. After this, rule of law does at­tach. You won’t be able to fire them af­ter­wards un­less they vi­o­late one of a few well-defined rules – or at least, that’s how it’s sup­posed to work, to pro­tect aca­demic free­dom, whether or not it does work that way. You’ll be count­ing on them to choose and do re­search, pur­sue fund­ing, teach and ad­vise, and help run the school, and be play­ing poli­tics with them.

That’s a big com­mit­ment. There are lots more peo­ple who want it and are qual­ified on pa­per than there are slots we can fund. And there are a lot more things that mat­ter than how much re­search one can do. Some of them are things that are ille­gal to con­sider, or would look bad if you were found to be con­sid­er­ing them. Others sim­ply are not re­search done. You can’t use a for­mula, be­cause peo­ple bring unique strengths and weak­nesses, and you’re fac­ing other em­ploy­ers who con­sider these fac­tors. Even if a sim­ple sys­tem could af­ford to mostly ‘take its licks’ you would face mas­sive ad­verse se­lec­tion, as ev­ery­one with bad in­tan­gibles would knock at your door.

You need to hold power over the new em­ploy­ees, so they’ll do the work that tenured em­ploy­ees don’t want to do, and so they’ll care about all as­pects of their job, rather than do­ing the bare tech­ni­cal min­i­mum ev­ery­where but re­search.

Then there’s the Good­hart fac­tors on the pa­pers di­rectly. One must con­sider how the pub­li­ca­tions them­selves would be gamed. If there was a thresh­old re­quire­ment for jour­nal qual­ity, the eas­iest jour­nals that count would be the only place any­thing would be pub­lished. If you have a point sys­tem, they’d game that sys­tem, and spend con­sid­er­able time do­ing it. If you don’t eval­u­ate pa­per qual­ity or value, they won’t care at all about those fac­tors, fo­cus­ing purely on be­ing good enough to make it into a qual­ify­ing jour­nal. Plus, be­ing able to eval­u­ate these ques­tions your­self with­out an out­side guide or au­thor­ity will be part of the job you’re try­ing to get. We need to test that, too.

What you’re re­ally test­ing for when you con­sider tenure, ideally, is not only skill but also virtue. You want some­one who is nat­u­rally driven to schol­ar­ship and the academy, to drive for­ward to­wards im­por­tant things. While also car­ing enough to do a pass­able job with other fac­tors. Other­wise, once they can’t be fired, you won’t be able to get them to do any­thing. Test­ing for virtue isn’t some­thing you can quan­tify. You want some­one who will aim for the spirit rather than the let­ter, and who knows what the spirit is and cares about it in­trin­si­cally. If you judge by the let­ter, you’ll se­lect for the op­po­site, and if you spec­ify that ex­plic­itly, you’ll lose your sig­nal that way as well.

I’d write this one up to power and ex­ploita­tion of those lower on the totem pole, the need to test for fac­tors that you can’t say out loud, the need to test for virtue, and the need to test for know­ing what is valuable.

G] A Good Lawyer

Peo­ple right­fully don’t think this num­ber will tell us much, even now when it is not be­ing gamed and vuln­er­a­ble to Good­hart. Robin seems to be as­sum­ing that one should think that a pre­vi­ous win per­centage should be pre­dic­tive of a lawyer’s abil­ity to win a par­tic­u­lar case, rather than be­ing pri­mar­ily a se­lec­tion effect, or a func­tion of when they set­tle cases.

I doubt this is the case, even with a rel­a­tively low level of ad­ver­sar­ial Good­hart effects.

Most lawyers or at least their firms have great flex­i­bil­ity in what cases they pur­sue and ac­cept. They also have broad flex­i­bil­ity in how and when they set­tle those cases, as clients largely rely on lawyers to tell them when to set­tle. Some of them will mostly want cases that are easy wins, and set­tle cases that likely lose. Others, prob­a­bly bet­ter lawyers for win­ning difficult cases, will take on more difficult cases and be will­ing to roll the dice rather than set­tle them.

I don’t even know what counts as a ‘win’ in a le­gal pro­ceed­ing. In a civil case you strate­gi­cally choose what to ask for, which might have lit­tle re­la­tion to re­al­is­tic ex­pec­ta­tions for a ver­dict, so get­ting a lesser amount might or might not be a ‘win’ and any set­tle­ment might be a win or loss even if you know the terms, and of­ten the terms are con­fi­den­tial.

Thus, if I was look­ing for a lawyer, I would con­tinue to rely on per­sonal recom­men­da­tions, es­pe­cially from lawyers I trust, rather than look at over­all track records, even if those track records were eas­ily available. I don’t think those track records are pre­dic­tive. Ask­ing ques­tions like some­one’s suc­cess in similar style cases, with richer de­tail in each case, seems bet­ter, but one has to pay care­ful at­ten­tion.

If peo­ple started us­ing win-loss records to choose lawyers, and lawyers started op­ti­miz­ing their win-loss records, what lit­tle in­for­ma­tion those records might have gets even less use­ful. You would mostly be mea­sur­ing which lawyers pri­ori­tize win-loss records, by se­lect­ing win­ners and forc­ing them to ver­dict, while avoid­ing, set­tling or pawn­ing off losers, and by get­ting onto win­ning teams, and so on. By ma­nipu­lat­ing the client and get­ting them to do what was nec­es­sary. It’s not like lawyers don’t mostly know which cases are win­ners. By choos­ing a lawyer with too good a win-loss record, you’d be get­ting some­one who cares more about how they look in a statis­tic than do­ing what’s right for their clients, and also who has the flex­i­bil­ity to choose which cases they have to take.

The ad­verse se­lec­tion here, it burns.

That’s what I’d ac­tu­ally ex­pect now. Some lawyers do care a lot about their track records, they’ll have bet­ter track records, and they’re ex­actly who you want to avoid. I’d take any­one brag­ging about their win rate as a very nega­tive sign, not a pos­i­tive one.

So I don’t think this is about sim­ple rules, or about peo­ple’s cog­ni­tive er­rors, or any­thing like that. I think Robin is just propos­ing a ter­rible mea­sure that is not ac­cu­rate, not well-defined and eas­ily gamed, and ask­ing why we aren’t mak­ing it available and us­ing it.

Con­trast this with eval­u­a­tions of doc­tors or hos­pi­tals for suc­cess rates or death rates from par­tic­u­lar surg­eries. That strikes me as a much bet­ter place to im­ple­ment such strate­gies, al­though they still have big prob­lems with ad­ver­sar­ial Good­hart if you started look­ing. But you can get a much bet­ter idea of what challenges are be­ing tack­led and about how hard they are, and a much bet­ter mea­sure of the rate of suc­cess. I’d still worry a lot about doc­tors se­lect­ing easy cases and avoid­ing hard ones, both for ma­nipu­la­tion and be­cause of what it would do to pa­tient care.

A gen­eral theme of sim­ple rules is that when you re­ward and pun­ish based on sim­ple rules, one of the things you are re­ward­ing is a will­ing­ness to pri­ori­tize max­i­miz­ing for the sim­ple rule over any other goal, in­clud­ing the thing you’re try­ing to mea­sure. Just like any other rule you might use to re­ward and pun­ish. The prob­lem with sim­ple rules is that they ex­plic­itly shut out one’s abil­ity to no­tice such op­ti­miza­tion and pun­ish it, which is the nat­u­ral way to keep such ac­tions in check. Without it, you risk driv­ing out any­one who cares about any­thing but them­selves and gam­ing the sys­tem, and cre­at­ing a cul­ture where gam­ing the sys­tem and car­ing about your­self are the only virtues.

H] Land Allocations

If all you care about is the ‘pro­duc­tivity’ of the as­set and/​or the rev­enue raised, then of course you use an auc­tion. Easy enough, and I think peo­ple rec­og­nize this. They don’t want that. They want a pre­vi­ously pub­lic as­set to be used in ways the pub­lic prefers, and think that we should pre­fer some uses to other uses be­cause of the ex­ter­nal­ities they cre­ate.

It seems rea­son­able to use the op­por­tu­nity of sel­l­ing pre­vi­ously pub­lic goods to ad­vance pub­lic policy goals that would oth­er­wise re­quire con­fis­cat­ing pri­vate prop­erty. Pri­vate sel­l­ers will also of­ten at­tach re­quire­ments to sales, or choose one buyer over an­other, sac­ri­fic­ing pro­duc­tivity and rev­enue for other fac­tors they care about.

We can point out all we like how mar­kets cre­ate more pro­duc­tion and more rev­enue, but we can’t tell peo­ple that they should care mostly about the quan­tity of pro­duc­tion and rev­enue in­stead of other things. When there are as­sets with large pub­lic policy im­pli­ca­tions and ex­ter­nal­ities to con­sider, like the spec­trum, it makes sense to think about monopoly and oli­gopoly is­sues, about what use the as­sets will be put to by var­i­ous buy­ers, and what we want the world to look like.

That doesn’t mean that these good fac­tors are the pri­mary jus­tifi­ca­tions. If they were, you’d see con­di­tional con­tracts and the like more of­ten, rather than pri­vate deals. The real rea­son is usu­ally that other mechanisms al­low in­sid­ers to ex­tract pub­lic re­sources for pri­vate gains. This is largely a story of brazen cor­rup­tion and theft. But if we’re go­ing to ar­gue for sim­ple rules be­cause they max­i­mize sim­ple pri­ori­ties, we need to also ar­gue for why those pri­ori­ties cover what we care about, or we’ll be seen as tone deaf at best, al­low­ing the cor­rupt to win the ar­gu­ment and steal our money.

I] In­vest­ment Funds

Low fee in­dex funds are grow­ing in­creas­ingly pop­u­lar each year, tak­ing in more money and a greater share of as­sets. Their mar­ket share is so large that be­ing in­cluded in a rele­vant in­dex has a mean­ingful im­pact on share prices.

Man­aged funds are on the de­cline. Most of these funds are not es­pe­cially pres­ti­gious and most peo­ple in­vested in them don’t brag about them, nor do they have much spe­cial faith in those run­ning the funds. They’re just not enough on the ball to re­al­ize they’re be­ing taken for a ride by pro­fes­sional thieves.

Nor do I think most peo­ple care about as­so­ci­at­ing with high sta­tus hedge funds or any­thing like that. I don’t see it, at all.

Also, those sim­ple rules? You can find them in ac­tive funds, too. A lot of them are pretty pop­u­lar. Sim­ple tech­ni­cal anal­y­sis, sim­ple mo­men­tum, sim­ple value rules, and so on. What counts as sim­ple? That’s a mat­ter of per­spec­tive. In­dex providers are of­ten do­ing stag­ger­ingly com­plex things un­der the hood. And in­dex­ing off some­one else’s work is a ma­gi­cian’s trick, free rid­ing off the work of oth­ers in a way that gets dan­ger­ous if too many start rely­ing on it.

Most reg­u­lar in­vestors who think about what they’re do­ing at all, know they should likely be in in­dex-style funds, and in­creas­ingly that’s where they are. If there’s a mys­tery at all it’s at least con­tained at the high end, in hedge funds with large min­i­mums.

One can split the re­main­ing ‘mys­tery’ into two halves. One is, why do some peo­ple think there ex­ist funds that have suffi­cient alpha to jus­tify their fees? Two is, why do some peo­ple think they’ve found one of those funds?

The first mys­tery is sim­ple. They’re right. There ex­ist funds that have alpha, and pre­dictably beat the mar­ket. The trick is find­ing them and get­ting your money in (or the even bet­ter trick is figur­ing out how to do it your­self). I don’t want to get into an ar­gu­ment over effi­cient mar­kets here and won’t dis­cuss it in the com­ments, but the world in which no one can beat the mar­ket doesn’t ac­tu­ally make any sense.

The sec­ond mys­tery is also sim­ple. Mar­ket­ing, win­ners curse, fooled by ran­dom­ness and ad­verse se­lec­tion, and the laws of mar­kets. Of course a lot more peo­ple think they’ve found the win­ner than have ac­tu­ally found one.

This is a weird case in many ways, but my core take here is that the part of this that does be­long on this list, is an ex­am­ple of com­plex­ity as jus­tifi­ca­tion for theft.

J] Sel­ling the Company

Google is the auc­tion com­pany. They were uniquely qual­ified to run an auc­tion and by­pass the banks, and did it (as I un­der­stand it) largely be­cause it was on brand and they’d have felt ter­rible do­ing oth­er­wise. A more in­ter­est­ing case is Spo­tify, who re­cently sim­ply let peo­ple start trad­ing its stock with­out an IPO at all. Although they still paid the banks fees, which I find su­per weird and don’t un­der­stand. There never was a re­bel­lion.

How do banks ex­tract the money?

My model is some­thing like this, com­ing mostly from read­ing Matt Lev­ine. The banks claim that they provide es­sen­tial ser­vices. They find and line up cus­tomers to buy the stock, they vouch for the stock, they price the stock prop­erly to en­sure a nice bump so ev­ery­one feels happy, they back­stop things in case some­thing goes wrong, they han­dle a ton of de­tails.

What they re­ally do are two things. Both are cen­tered around the gen­eral spread­ing by banks of FUD: Fear, Uncer­tainty and Doubt.

First, they pre­vent firms from sud­denly hav­ing to nav­i­gate a legally tricky and po­ten­tially risky, and po­ten­tially quite com­plex, world they know noth­ing about, where mess­ing up could be a dis­aster. One does not sim­ply sell the com­pany or take it pub­lic, as much as it might look sim­ple from the out­side. And while the bank’s marginal costs are way, way lower than what they charge, try­ing to get that ex­per­tise in house in a con­fi­dent way is hard.

Se­cond, they are what peo­ple are com­fortable with. You’re not blame­wor­thy for pay­ing the bank. It’s the null ac­tion. If you do it, no one says ‘hey they’ve robbed us all of a huge amount of money.’ In­stead, they say ‘good on you for not be­ing too greedy and try­ing to max­i­mize the price while risk­ing the com­pany’s fu­ture.’

They’re do­ing this at the cru­cial mo­ment when how you look is of cru­cial im­por­tance, when you’re about to get a huge wind­fall for years or a life­time of work and give the same to ev­ery­one who helped you. When you’re spend­ing all your en­ergy ne­go­ti­at­ing lots of other stuff. A dis­rup­tion threat­ens to un­ravel all of that. What’s a few per­cent in that situ­a­tion? So what if you don’t price your IPO as high as you could have so that bankers can en­joy their bounce?

Banks are con­spiring with the buy­ers to cheat the sel­l­ers out of the value of what they bring to the table. Buy­ers who ob­ject are threat­ened with os­tracism and be­ing some­one no one is com­fortable with, with the other side walk­ing away from the table af­ter buy­ers put in the work to get here.

Is this all guillo­tine-wor­thy high­way rob­bery? Hell yes. Com­pletely.

Banks (and the buy­ers who are their best cus­tomers and al­lies) are col­lud­ing with this pric­ing, and that’s the nicest way to put this. Again, this is theft. Com­plex­ity is in­tro­duced to al­low rent seek­ing and theft, ex­ploit­ing a mo­ment of vuln­er­a­bil­ity.

K] Col­lege Admissions

In­ter­est­ing that Robin says the sys­tem ‘ap­pears sta­ble.’ To me it does not seem sta­ble. We just had a huge col­lege ad­mis­sions scan­dal that dam­aged faith in the sys­tem and a quite-well jus­tified law­suit against Har­vard. We have the SAT promis­ing to in­tro­duce ‘ad­ver­sity scores.’ We have in­creas­ingly se­lec­tive ad­mis­sions eat­ing more and more of child­hood, and the rule that what can’t go on for­ever, won’t. This calls for some pop­corn.

What’s caus­ing the sys­tem to be com­plex? We see sev­eral of the an­swers in play here.

We see the ‘fac­tors you can’t cite ex­plic­itly’ prob­lem and the ‘we don’t want some­thing we can be sued or blamed for’ here. Ad­mis­sions officers are try­ing to pick kids who will be suc­cess­ful at school and in life, as well as satisfy other goals. A lot of the things that pre­dict good out­comes in life are not things you would want to be caught dead us­ing as a de­ter­mi­nant in ad­mis­sions even if they weren’t ille­gal to use in ad­mis­sions. The only solu­tion is to make the sys­tem com­plex and opaque, so no one can prove what you were think­ing.

We also see com­plex­ity as a way for the rich and pow­er­ful to ex­pro­pri­ate re­sources, in the sense that the rich and pow­er­ful and their chil­dren are likely to be more suc­cess­ful, and more likely to give money to the school. And of course, if the school has dis­cre­tion, that gives the school power. It can ex­tract re­sources and pres­tige from oth­ers who want to get their kids in. Em­ploy­ees, es­pe­cially high-up ones, can ex­tract things even with­out ille­gal bribes. Why pass that up?

We see the Good­hart’s Law and ad­verse se­lec­tion prob­lems. If you ad­mit purely on the ba­sis of a test, and the other schools ad­mit on the ba­sis of a range of fac­tors, you don’t get the best test scor­ers un­less you’re Har­vard. You get the kids who are an epic fail at those other fac­tors.

If you give kids an ex­plicit tar­get, they and their par­ents will struc­ture their en­tire lives around it. They’ll do that even with a vague im­plicit tar­get, as they do now. If it’s ex­plicit, you get things like you see in China, where (ac­cord­ing to an eye­wit­ness who once came to din­ner) many kids are pul­led from school and do noth­ing but cram facts into their heads for the col­lege ad­mis­sions exam for years. And why shouldn’t they?

So you get kids whose real ed­u­ca­tions are crip­pled, who have no life ex­pe­rience and no joy of child­hood. The only al­ter­na­tive is to al­low a gen­eral sense of who the kid is and what they’ve done to mat­ter. To be able to holis­ti­cally judge kids and prop­erly ad­just.

As always, the more com­plex and hard to un­der­stand the game, the greater the ex­pert’s ad­van­tage. The rich and pow­er­ful who un­der­stand the sys­tem and can make them­selves look good will have a large edge from that. And the more we ex­plic­itly pe­nal­ize them for those ad­van­tages, but not for their gam­ing of the sys­tem, the more we force them to game the sys­tem even harder. If you use an ad­ver­sity score to set im­pos­si­bly high stan­dards for rich kids, they’re go­ing to use ev­ery ad­van­tage they have to make up for that even more than they already do.

And of course, part of the test is see­ing how you learn to game the test and what ap­proach you take. Can you do it with grace? Do you do too much of it, not enough or the right amount?

This is all an anti-in­duc­tive arms race. The art of gam­ing the sys­tem is in large part the art of mak­ing it look like you’re not gam­ing the sys­tem, which is an ar­gu­ment for sim­pler rules. At this point, what por­tion of suc­cess­ful ad­mis­sions is twist­ing the truth? How much is flat out ly­ing? How much is pre­sent­ing your­self in a mis­lead­ing light? To what ex­tent are we train­ing kids from an early age to have high simu­lacrum lev­els and sac­ri­fice their in­tegrity? A lot. In­tegrity be­ing ex­plic­itly on the test just makes it one more thing you need to learn to fake.

I hate the cur­rent situ­a­tion, and the ed­u­ca­tional sys­tem in gen­eral, but I think the al­ter­na­tive of a sim­ple, sin­gle writ­ten test, with the sys­tem oth­er­wise un­changed, would be worse. But of course, we’d never let it be that sim­ple. That’s all be­fore the fights over how to ad­just those scores for ‘ad­ver­sity’ and ‘di­ver­sity,’ and how to quan­tify that, and the other things we’d want to fac­tor in. Can you imag­ine what hap­pens to high schools if grades don’t mat­ter? What if grades did mat­ter in a for­mu­laic mat­ter and stu­dents and teach­ers were forced to con­front the in­cen­tives? The end­less bat­tles over what other life ac­tivi­ties should score points, the death of any that don’t, and the twist­ing into point ma­chines of those that do?

So here we have all the Good­hart prob­lems, and the theft prob­lems, and the power prob­lems, and the blame­wor­thy con­sid­er­a­tions and jus­tifi­ca­tions prob­lems and law­suit prob­lems with their in­cen­tive to de­stroy all in­for­ma­tion. The gang’s all here.

L] Fire that CEO

I love me a pre­dic­tion mar­ket, but you have to do it right. Would enough peo­ple and money par­ti­ci­pate? If they did, would they have the right in­cen­tives? If both of those, would you want that to be how you make de­ci­sions?

I think the an­swer to the first ques­tion is yes, if you struc­ture it right. If there are only two pos­si­bil­ities and one of them will hap­pen, you can make it work.

The an­swer to the sec­ond ques­tion is, no.

We can con­sider two pos­si­bil­ities.

In sce­nario one, this acts as an ad­vi­sory for the board, to help them de­cide what to do.

In sce­nario two, this is the sole thing looked at, and CEOs are fired if and only if they are judged to be bad for the stock price, or can oth­er­wise only be fired for spe­cific causes (e.g. if found shoot­ing a man on Fifth Av­enue or steal­ing mil­lions in com­pany funds and spend­ing them on free to play games, you need to pull the plug with­out stop­ping to look at the mar­ket).

The prob­lem with sce­nario one is that you’re trad­ing on how much the com­pany is worth given that the CEO was fired. That’s very differ­ent from what you think the com­pany would be worth if we de­cided to fire the CEO. The sce­nar­ios where the CEO is fired are where the board is un­happy with them, which is usu­ally be­cause of bad things that would make us think the stock is likely to be less valuable, like the stock price hav­ing gone down or in­sid­ers know­ing the CEO has done or will do hard to mea­sure long term dam­age. That doesn’t mean it won’t also take into ac­count other things like whether the CEO is pay­ing off the board, but the cor­re­la­tion we’re wor­ried about is still su­per high. Giv­ing the board dis­cre­tion, that mar­ket par­ti­ci­pants would ex­pect the board to use, hope­lessly mud­dles things.

You could try to solve that prob­lem by hav­ing the mar­ket trade only very close in time to the board de­ci­sion. You kind of have to do that any­way, to avoid hav­ing a lame duck CEO. But it still de­pends on a lot of pri­vate in­for­ma­tion, and the de­ci­sion will still re­veal a lot about the firm. So I think that re­al­is­ti­cally this won’t work.

The prob­lem with sce­nario two is that you’ve taken away any abil­ity to pun­ish or re­ward the CEO for any­thing other than fu­ture stock prices. This effec­tively gives the CEO ab­solute power, and al­lows them to get away with any and all bad be­hav­ior of all kinds. Even if past be­hav­ior low­ers the stock price, it only mat­ters to the ex­tent that it pre­dicts fu­ture ac­tions which would fur­ther lower the stock price. So CEOs don’t even have to care about the stock price. They only need to care about the stock price pre­dic­tions in re­la­tion to each other. So the best thing the CEO can do is make get­ting rid of them as painful as pos­si­ble. Even more than now, they want to make sure that los­ing them de­stroys the com­pany as much as pos­si­ble. Their pri­mary jobs now are to hype them­selves as much as pos­si­ble to out­siders, and to spend cap­i­tal ma­nipu­lat­ing these pre­dic­tion mar­kets.

Again, we’re see­ing Good­hart prob­lems, we’re see­ing re­in­force­ment of power (in this case, of the board over the CEO, so it’s a bal­ance of power we likely wel­come), and the abil­ity to take things into con­sid­er­a­tion with­out need­ing to make them ex­plicit or mea­surable, as com­pa­nies both care about things they’re not legally al­lowed to care about and which we wouldn’t like hear­ing they cared about, es­pe­cially ex­plic­itly, and they need to main­tain con­fi­den­tial­ity.