Generalized Efficient Markets in Political Power

Schel­ling Points

In Thomas Schel­ling’s clas­sic ex­per­i­ment, we imag­ine try­ing to meet up with some­one in New York City, but we haven’t speci­fied a time or place in ad­vance and have no way to com­mu­ni­cate. Where do we go, and when, to max­i­mize the chance of meet­ing? There are some “nat­u­ral” choices—places and times which stand out, like the top of the Em­pire State Build­ing at noon. Th­ese are called Schel­ling points.

More gen­er­ally, Schel­ling points are rele­vant when­ever two or more peo­ple need to make “match­ing” choices with limited abil­ity to com­mu­ni­cate in ad­vance. For in­stance, cer­tain mar­kets, like Ebay or Uber, serve as “meet­ing points” for buy­ers and sel­l­ers. Schel­ling him­self wrote a fair bit about ne­go­ti­a­tions, where peo­ple need to agree on how to di­vide some spoils, or where to draw a bound­ary, or … They can talk to each other, but ac­tu­ally com­mu­ni­cat­ing is hard be­cause both par­ties have no in­cen­tive to be hon­est—and there­fore no rea­son to trust each other when e.g. when one per­son says “I just can’t af­ford to sell it be­low $10”. Schel­ling points be­come nat­u­ral out­comes for the ne­go­ti­a­tions—e.g. split the spoils evenly, draw the bound­ary at the river, etc.

In prac­tice, it’s of­ten use­ful to cre­ate Schel­ling points. In the New York City ex­per­i­ment, one could put up a gi­ant billboard that says “meet­ing point”, and place signs all over the city point­ing to­ward the meet­ing point, mak­ing that point a nat­u­ral place for peo­ple to meet. Some air­ports ac­tu­ally do this:

Ebay and Uber are of course also ex­am­ples of pur­pose-built Schel­ling points.

One in­ter­est­ing fea­ture of cre­at­ing a Schel­ling point is that we may have some de­grees of free­dom available, and we can use those de­grees of free­dom to ex­tract value.

In our meetup ex­am­ple, we could imag­ine putting the meetup point in­side a build­ing, and charg­ing peo­ple to get in—much like Ebay or Uber charge fees for their ser­vices. Or, we could imag­ine lo­cal busi­nesses want­ing to put the meetup point nearby, in hopes of at­tract­ing busi­ness from meeters—one could imag­ine a gimicky air­port restau­rant with a bunch of “MEET HERE!” signs hop­ing to sell peo­ple over­priced na­chos and drinks while they wait to meet up with friends or fam­ily. Alter­na­tively, we could imag­ine users of the meetup point want­ing it in lo­ca­tions con­ve­nient to them—e.g. in the New York City ex­am­ple, peo­ple in a par­tic­u­lar neigh­bor­hood might cam­paign to es­tab­lish the meetup point there for their own con­ve­nience.

How­ever, the Schel­ling point cre­ator/​con­trol­ler only has so many de­grees of free­dom. Charge too high a fee, and peo­ple will go to some other Schel­ling point. Move the meetup point to a neigh­bor­hood in the out­skirts of town, and it will be too in­con­ve­nient for peo­ple from other neigh­bor­hoods. By de­fault, peo­ple will usu­ally stick to Schel­ling points which ev­ery­one is already us­ing—if ev­ery­body has always met up un­der this par­tic­u­lar sign, then that’s the ob­vi­ous place to keep meet­ing up—so the con­trol­ler of the origi­nal Schel­ling point can ex­tract more value than “new” points. But there are always limits.

We can think of a Schel­ling-point-con­trol­ler’s “power” as their range of free­dom in mov­ing the point around, or as the amount of value they can ex­tract with­out los­ing out to some other Schel­ling point. Just be­cause some­one nom­i­nally “con­trols” the Schel­ling point does not mean they can ac­tu­ally do any­thing with­out los­ing it! It may be that even a small fee will drive ev­ery­one to switch to a differ­ent Schel­ling point. It may be that the Schel­ling point is in the dead cen­ter of the city and peo­ple will keep meet­ing in the dead cen­ter even if the signs move (e.g. maybe some­one will just put up new signs for the “city cen­ter” and peo­ple will meet there). It may be that main­tain­ing all the signs costs roughly as much as one can earn from the Schel­ling point (oth­er­wise a com­peti­tor would come along and put up more signs of their own). There are many ways to ex­tract value from a Schel­ling point, but if there’s some mechanism for open com­pe­ti­tion over con­trol of the point, then the net value one can ex­tract may be driven to near-zero.

That’s roughly how I think poli­tics works.

Gover­nance as Schel­ling Point

When peo­ple are op­er­at­ing in a group, it’s use­ful to have stan­dard­ized Schel­ling points for a wide va­ri­ety of in­ter­per­sonal con­flicts, so that we don’t need a bunch of ex­pen­sive ne­go­ti­a­tion/​con­flict to re­solve each one. Th­ese Schel­ling points are things like “rules” and “lead­ers”.

An ex­am­ple: Alice likes to rock out to loud mu­sic af­ter sun­down, while her neigh­bor Bob likes to go to bed early. They have con­flict­ing prefer­ences for when quiet hours should be. But nei­ther of them wants to get in a fight about it, or spend a bunch of time and effort ne­go­ti­at­ing. So, the build­ing/​neigh­bor­hood has a rule: “quiet hours run from 10pm to 6am”. The main pur­pose of the rule is to act as a Schel­ling point: by de­fault, those are the quiet hours which ev­ery­one re­spects and ex­pects ev­ery­one else to re­spect. They might be en­forced if needed, but usu­ally that doesn’t ac­tu­ally hap­pen. Of course, Alice and Bob could still work out a sep­a­rate deal—e.g. maybe Alice talks to all her neigh­bors and gets their ok to play loud mu­sic on Fri­day night—but that would re­quire a bunch of ex­tra ne­go­ti­a­tion. The offi­cial rule is the Schel­ling point ev­ery­body co­or­di­nates on, by de­fault.

More gen­er­ally, laws and courts serve as Schel­ling points in ne­go­ti­a­tions. Where does my prop­erty end and yours be­gin? The gov­ern­ment land records provide a Schel­ling point an­swer, so we don’t need to fight/​ne­go­ti­ate over it our­selves. In a dis­agree­ment over land, the po­lice and mil­i­tary all want to co­or­di­nate and back the same per­son, so the gov­ern­ment’s land records tell them all who the “right­ful” owner is. Since the po­lice and mil­i­tary co­or­di­nate around that Schel­ling point, it be­comes the nat­u­ral Schel­ling point for oth­ers as well.

In prin­ci­ple, the legally-rec­og­nized Schel­ling point could sim­ply be ig­nored—e.g. if a chunk of land is legally rec­og­nized as your prop­erty, and I build some­thing on it with­out per­mis­sion, the two of us could just agree that this is fine, effec­tively ne­go­ti­at­ing a differ­ent Schel­ling point. Some non-gov­ern­ment group could even have mechanisms to en­force the al­ter­na­tive Schel­ling point. But the le­gal Schel­ling point is the one which po­lice and the mil­i­tary are will­ing to en­force.

Poli­ti­cal Power

The Death Eaters don’t always agree on when, where or how to launch an at­tack, but they know that any at­tack will go bet­ter if they’re all in it to­gether. So, they band to­gether be­hind a leader, and at­tack when, where and how the leader di­rects. The leader’s or­ders be­come the Schel­ling point ac­tion for the group.

The Death Eaters ex­am­ple illus­trates the no­tion of “poli­ti­cal power” par­tic­u­larly well: the leader’s “power” is roughly the set of or­ders he could give with­out his or­ders ceas­ing to be a Schel­ling point for group ac­tivity. If the Death Eaters are mostly in agree­ment on some course of ac­tion, and the leader di­rects against it, then his or­ders be­come less of a Schel­ling point, and mul­ti­ple such or­ders will likely see him re­moved from nom­i­nal power. He has some de­gree of free­dom in which or­ders to give, but only to the ex­tent that the Death eaters are, on av­er­age, mostly in agree­ment with his choices.

There’s a gen­eral prin­ci­ple of “power” here: a leader’s power is the set of or­ders they could give with­out their or­ders ceas­ing to be Schel­ling points for the group’s ac­tivi­ties. A leader’s power is high when group mem­bers all want to co­or­di­nate their choices, but care much less about which choice is made, so long as ev­ery­one “matches”. Then the leader can just choose any­thing they please, and ev­ery­one will go along with it. (In­ter­est­ingly, this sug­gests that a leader can get high value from a group whose prefer­ences are or­thog­o­nal to their own; pur­sue power in groups which care about differ­ent things than you!) Con­versely, a leader’s power can be low in two ways:

  • Group mem­bers care a lot about which choice is made. In this case, the leader has lit­tle free­dom to choose, and is mostly just a figure­head.

  • Group mem­bers only weakly care about co­or­di­nat­ing. In this case, the group is in­her­ently un­sta­ble; lots of deals and con­ces­sions are needed just to keep it to­gether.

Key thing to keep in mind: both of these con­di­tions are rel­a­tive-to-the-next-best-op­tion. Group mem­bers may care a lot about co­or­di­nat­ing and only have weak prefer­ences about which choice is made, but if there’s a com­peti­tor who could co­or­di­nate just as well and satisfy the weak prefer­ences bet­ter, then that com­peti­tor’s or­ders may be­come the new Schel­ling point.

That’s poli­tics, in a nut­shell: peo­ple try to turn their own or­ders/​poli­cies/​sug­ges­tions into Schel­ling points for group ac­tivity. They do this mainly by offer­ing con­ces­sions and fa­vors to group mem­bers/​sub­groups, in ex­change for those mem­bers’ sup­port for the new Schel­ling point.

Com­pe­ti­tion and Gen­er­al­ized Mar­ket Efficiency

In demo­cratic coun­tries/​groups, we have a built-in mechanism for com­pe­ti­tion be­tween would-be lead­ers. In other words, there’s a Schel­ling point for when and how to switch Schel­ling points. That im­me­di­ately sug­gests a gen­er­al­ized effi­cient mar­kets-style hy­poth­e­sis: lead­ers’ power in such groups is driven by com­pe­ti­tion to near-zero.

What does that look like?

Well, most peo­ple want to co­or­di­nate; re­gard­less of what the rules are, we want to agree on what the rules are, oth­er­wise we end up in ex­pen­sive fights. But most peo­ple also have some prefer­ences about the rules—i.e. poli­ti­cal poli­cies. Would-be lead­ers make promises: they pre­com­mit to cer­tain poli­cies, thereby cut­ting off cer­tain op­tions if they win (i.e. sac­ri­fic­ing po­ten­tial power), but gain­ing more sup­port for their Schel­ling point in the pro­cess. To max­i­mize power, a would-be leader wants to just barely “out­bid” all the other would-be lead­ers—i.e. promise just a bit more to just a few more par­ties, keep­ing as much power as pos­si­ble while still win­ning the po­si­tion.

Of course, the other com­peti­tors are try­ing to do the same thing. Solve for the equil­ibrium: the com­peti­tors ei­ther bid away any de­gree of free­dom which any con­stituency cares about, or lose to some­one who bids more. Gen­er­al­ized effi­cient mar­kets kick in; the leader ends up with near-zero power. They’re mostly just a figure­head im­ple­ment­ing all the poli­cies they had to pre­com­mit to in or­der to win the elec­tion.

Now, con­sider the re­verse—a dic­ta­tor or sin­gle-party state or the like. How do they max­i­mize power?

To max­i­mize power, they want to avoid gen­er­al­ized effi­cient mar­kets—i.e. they want to min­i­mize com­pe­ti­tion over the Schel­ling point. Elec­tions en­courage com­pe­ti­tion by pro­vid­ing a Schel­ling point for when and how to switch Schel­ling points; the power-hun­gry leader wants ex­actly the op­po­site of that. They want to make sure that there is no Schel­ling point for when and how to switch Schel­ling points.

If a new Schel­ling point does show up (e.g. an op­po­si­tion group), there won’t be any agree­ment on when and how to switch, so there will prob­a­bly be some ex­pen­sive con­flict (i.e. civil war). That ex­pen­sive con­flict it­self cre­ates a big po­ten­tial en­ergy bar­rier for any po­ten­tial com­peti­tor: for the peo­ple sup­port­ing a switch to the new Schel­ling point, the ex­pected gains from the switch must ex­ceed costs of the con­flict. So from the dic­ta­tor’s stand­point, the worse a civil war would be, the fewer con­ces­sions and hand­outs they need to make and the broader their power.

(Of course, a dic­ta­tor can use other strate­gies to max­i­mize power as well—e.g. threat­en­ing to kill peo­ple/​de­stroy things if a new Schel­ling point comes along. But that’s a sym­met­ric strat­egy: the dic­ta­tor’s en­e­mies can just as eas­ily threaten to kill peo­ple/​de­stroy things if no new Schel­ling point is adopted. The dic­ta­tor may have an ad­van­tage in re­sources, but that gap can in prin­ci­ple be closed by other means. It’s mainly the lack of a Schel­ling point for switch­ing Schel­ling points which con­fers an asym­met­ric ad­van­tage to the in­cum­bent.)

Democ­racy’s Seedy Underbelly

Based on the pre­vi­ous sec­tion, some­one ac­cus­tomed to a “democ­racy=good, dic­ta­tor=bad” wor­ld­view might think that lead­ers be­ing forced to bar­gain away all their po­ten­tial power is good news. “Lead­ers are just figure­heads” and “lead­ers are just im­ple­ment­ing the poli­cies which won the elec­tion” both say the same thing. This is “good”, yes?

The failure modes of democ­racy are baked-in here too.

Con­sider a would-be leader figur­ing out the perfect mix of promises and con­ces­sions to make, in or­der to win an elec­tion. From their point of view, differ­ent peo­ple want­ing op­po­site things is a prob­lem. Mov­ing the meetup point closer to one neigh­bor­hood means mov­ing it fur­ther from an­other. But di­men­sion­al­ity is a ma­jor boon for the leader: there’s thou­sands of di­men­sions along which policy can change. If Alice cares strongly about one par­tic­u­lar di­men­sion—like, say, gov­ern­ment sup­port for her pro­fes­sion—which no­body else cares about very much, then that promise can be made to Alice with­out los­ing the sup­port of some­body else. It’s spe­cial-in­ter­est poli­tics: look for poli­cies with fo­cused benefits and diffuse costs. Pile many such poli­cies to­gether, and you have a win­ning coal­i­tion.

That out­come may be “effi­cient” in the sense that no other bun­dle of poli­cies can beat it in an elec­tion, but that’s very differ­ent from “effi­cient” in the sense of “not wast­ing ridicu­lous amounts of re­sources on pork-bar­rel pro­jects and reg­u­la­tory bar­ri­ers to en­try”.

Now, sup­pose we’re un­happy with this out­come. We want to build a bet­ter world. What can we do?

Ob­vi­ously “run for office” is not a work­able an­swer here. Gen­er­al­ized effi­cient mar­kets mean we can’t win an elec­tion with­out trad­ing away any abil­ity to en­act our preferred poli­cies.

If we have some ex­ter­nal re­sources—e.g. a gi­ant pile of money—then we could po­ten­tially use that to “force” the poli­ti­cal equil­ibrium in a differ­ent di­rec­tion. This could look like old-fash­ioned bribery, where we just pay some stake­hold­ers to back our preferred Schel­ling point. It could look like a pay­ment to the poli­ti­cal sys­tem as a whole, e.g. offer­ing a pri­vate sub­sidy for road re­pair. It could in­volve re­sources other than money, as in a celebrity or me­dia out­let offer­ing an en­dorse­ment, or a na­tion offer­ing some con­ces­sion in ex­change for lower tar­iffs. We could change the op­tions available to the group via tech­nol­ogy, e.g. bit­coin. We could sim­ply try to con­vince peo­ple to sup­port our preferred poli­cies—though this means com­pet­ing in memes­pace, which has gen­er­al­ized effi­cient mar­kets of its own. The gen­eral pat­tern: we use our re­sources to change the set of op­tions available or to di­rectly in­fluence the prefer­ences of group mem­bers.

Point is: there are no hun­dred-dol­lar bills ly­ing on the ground. If we want to change the poli­ti­cal equil­ibrium in a highly-poli­ti­cally-com­pet­i­tive en­vi­ron­ment, we need to change the un­der­ly­ing op­tions available to the group and/​or the prefer­ences of in­di­vi­d­ual group mem­bers.