Generalized Efficient Markets in Political Power

post by johnswentworth · 2020-08-01T04:49:32.240Z · LW · GW · 6 comments

Contents

  Schelling Points
  Governance as Schelling Point
  Political Power
  Competition and Generalized Market Efficiency
  Democracy’s Seedy Underbelly
None
6 comments

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:

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 [LW · GW] 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.

6 comments

Comments sorted by top scores.

comment by Zack_M_Davis · 2020-08-01T19:07:10.765Z · LW(p) · GW(p)

It gets worse. We also face coordination problems on the concepts we use to think with [LW · GW]. In order for language to work, we need shared word definitions, so that the probabilistic model in my head when I say a word matches up with the model in your head when you heard the word. A leader isn't just in a position to coordinate what the group does, but also which aspects of reality the group is able to think about [LW · GW].

comment by Raemon · 2020-08-04T02:00:42.708Z · LW(p) · GW(p)

I feel obligated to link CGP Grey's The Rules for Rulers, which makes some similar points through a somewhat different lens with a bunch of cute graphics.

comment by betulaster · 2020-08-04T01:21:33.847Z · LW(p) · GW(p)

I'm probably missing something obvious, but I don't trivially see how this

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!

follows from this

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.

Could you please elaborate?


Also, I have an outsider's view of American (or, indeed, Western in general) politics, so I can be wrong, but I think an argument from empirics could be made against this:

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.

At least in the two most recent American elections (2016 and then the 2018 midterms) it seems like it was very much not the case of people racing for the most focused benefits and most diffuse cost, but rather for the most efficient way to galvanize their voters, cost be damned. Think of the wall on the Mexican border - it would probably be exorbitantly expensive, including to those that voted for it, but it was a very powerful symbol that people who felt strongly about the issue could rally behind.

538 here do a kind of literature review - and find, amongst other things, that

racial attitudes mattered more in 2016 than in any recent election — even 2008, when the presence of an African-American candidate shaped the political conversation.

Unless I misunderstand the idea, I don't think issues of race have a narrow focused scope of benefits and costs diffuse enough not to be noticed by other voters.


I also think this point

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.

makes an assumption of voters being more-or-less perfectly informed about what the Schelling point (policies and laws) actually is. What if a leader could get elected by pre-commiting to certain policies, but then actually not act on them, while managing to convince the voters that they, in fact, are doing their best to implement these policies, but are failing to for a certain (probably not a very falsifiable) reason? Or does the model already support this in a way that I don't notice?

Replies from: johnswentworth
comment by johnswentworth · 2020-08-04T02:53:43.543Z · LW(p) · GW(p)

On "group with preferences orthogonal to your own": the idea is you can give the members exactly what they want, and then independently get whatever you want as well. Since they're indifferent to the things you care about, you can choose those things however you please.

At least in the two most recent American elections (2016 and then the 2018 midterms) it seems like it was very much not the case of people racing for the most focused benefits and most diffuse cost, but rather for the most efficient way to galvanize their voters, cost be damned.

I expect that politics in most places, and US Congressional politics especially, is usually much more heavily focused on special interests than the overall media narrative would suggest. For instance, voters in Kansas care a lot about farm subsidies, but the news will mostly not talk about that because most of us find the subject rather boring. The media wants to talk about the things everyone is interested in, which is exactly the opposite of special interests.

Also I am extremely skeptical that racial issues played more than a minor role in the election, even assuming that they played a larger role in 2016 than in other elections. Every media outlet in the country (including 538) wanted to run stories about how race was super-important to the election, because those stories got tons of clicks, but that's very different from actually playing a role.

Or does the model already support this in a way that I don't notice?

Nope, you are completely right on that front, poor information/straight-up lying were issues I basically ignored for purposes of this post. That said, most of the post still applies once we add in lying/bullshit; the main change is that, whenever they can get away with it, leaders will lie/bullshit in order to simultaneously satisfy two groups with conflicting goals. As long as at least some people in each constituency see through the lies/bullshit, there will still be pressure to actually do what those people want. On the other hand, people who can be fooled by lies/bullshit are essentially "neutral" for purposes of influencing the political equilibrium; there's no particular reason to worry about their preferences at all. So we just ignore the gullible people, and apply the discussion from the post to everybody else.

Replies from: betulaster
comment by betulaster · 2020-08-14T21:13:52.369Z · LW(p) · GW(p)

Thanks for the reply and sorry I couldn't get to this for some time! Hope you're still interested in the discussion.

I expect that politics in most places, and US Congressional politics especially, is usually much more heavily focused on special interests than the overall media narrative would suggest

This is really interesting and you probably have a good point. Do you think there's a more reliable way (for an outsider like myself, who's not able to, I dunno, go and ask people in a dive bar what they think) to get the lay of the political land in a particular point in space? (And time?) Maybe some centralized kind of poll repository?


Every media outlet in the country (including 538) wanted to run stories about how race was super-important to the election, because those stories got tons of clicks, but that's very different from actually playing a role.

On a side note, I can imagine this kind of perspective, when taken to an unmitigated extreme, leading to a very cartesian-demon view of the world. Most people that publish their thoughts are incentivized to make you, the reader, like it or be interested in it. Mass-media obviously so, bloggers or analysts or think tanks less obviously so, but still. If, when faced with a choice of writing about (a) things that are real but dull vs (b) things that are not real but get clicks, no one has an incentive to do (a), how do you form a view of the world?

(Had I not known about publish-or-perish and read Gelman/Falkovich on p-hacking, I myself would give the traditionally cartesian answer of "by reading scientific papers in peer-reviewed journals", but... yeah).


So we just ignore the gullible people, and apply the discussion from the post to everybody else.

I think we've made an important move in argumentation here - we've started to introduce the possibility of the voters differing by whether they believe the lies/bullshit or not. But if we do - that is, we introduce the possibility of the voter considering some of the politician's commitment to a future policy Schelling point not genuine - we also open the possibility for the voter to speculate on what the politician's true policies are.

Say, Alice runs for president on a conservative/jobs-centric platform, commits to outlaw work visas and expunge all foreign workers, and wins the race, and says that she's working hard to achieve that goal. Bob is a total supporter and he's sure that she does exactly that and thinks that deportations are only a couple days away. Charlie may be skeptical about the promise, because sounded very radical and campaign-y, but thinks she's probably going to cut work visas, but not be able to expunge foreigners already in the country. Dave agrees with Charlie that the promise was radical and will not be followed through on fully, but not on what will actually happen - he thinks Alice will be able to expunge already present foreigners, but never risk the political turmoil of removing work visas. Erin is a total skeptic and thinks Alice is merely exploiting the voters, and is actually not doing anything about the foreigners, and finally Frank is a conspiracy theorist and thinks that Alice is secretly working with a cabal of globalists to bring even more foreigners (maybe even illegally!) in while bullshitting him.

All of these people have different Schelling points! If Zack, a foreigner, asks Bob to loan him money, Bob is going to refuse, because he thinks Zack will be kicked out of the country tomorrow and he's not getting his money back. If he asks Erin, she's likely to agree, because she doesn't believe Zack is going anywhere.

Now, sure, there's only so many ways to interpret a single campaign promise, and there are bound to be groups within the voter base that will agree on what Alice will actually do, the Schelling point will work for them - but since Alice is incentivized to make a lot of focused-benefit-disperse-cost promises, voters who agree on what her actions on a certain policy are, may disagree on what her actions regarding a different policy are. So... when nobody agrees on what the Schelling point is, does it, for all intents and purposes, exist?

Replies from: johnswentworth
comment by johnswentworth · 2020-08-14T21:55:15.155Z · LW(p) · GW(p)

Great points!

Do you think there's a more reliable way (for an outsider like myself, who's not able to, I dunno, go and ask people in a dive bar what they think) to get the lay of the political land in a particular point in space?

This is wayyyy outside my zone of expertise, but I would look for specialist-oriented publications - e.g. newsletters specifically targeted at lobbyists/policymakers, or political information in the industry publications of special-interest industries.

If, when faced with a choice of writing about (a) things that are real but dull vs (b) things that are not real but get clicks, no one has an incentive to do (a), how do you form a view of the world?

I'd say the key is to generate your own questions, then proactively look for the answers rather than waiting around for whatever information comes to you. There's plenty of good information out there, it just isn't super-viral, so you have to go looking for it.

All of these people have different Schelling points!

Important point here: these people don't actually have different Schelling points. They presumably all agree that if Alice wins the election, then whatever Alice signs into law will be the new Schelling point. What these people disagree on is their expectations for what the future Schelling point will be.