In Excerpts from a larger discussion about simulacra, I worked through a well-known schema for distinguishing different relationships towards semantic reference, that are a natural result of interactions between shared-production games and expropriation games. Here, I analyze the coalition politics of such games.

The Survivor game

In zero-sum games, majoritarian decision rules (such as democracy) create an asymmetry - it's much easier to expropriate from a minority than from a majority - or, easier to transfer wealth to a majority than to a minority. Why would the majority vote for something they don't all benefit from?

A simple variant of this is the Survivor game, in which a single player is voted off the island at a time (see also the ancient Greek custom of ostracism). Since there's comparatively little advantage to being singled out for good, players will tend to want to avoid revealing information about themselves or their allies. Loudly voicing consensus opinion in ways that don't specify the implications for any person is fine because it's not informative. Anything that lets people distinguish you from the others is dangerous.

The idea of a Schelling point is that if players in a game need to converge on one location in a map, then in the absence of a strong incentive to favor one location, they will tend to converge on some obviously identifiable feature. For instance, in surveys, Thomas Schelling found that a surprisingly large number of people, if tasked with meeting someone on a specified day, in New York, with no further information, would converge on the information booth in Grand Central - and if no time was specified, they favored noon.

In a pure Survivor game, the first player to reveal their "location" loses. They become the feature everyone else converges on as an expropriation target. One natural side effect of this is coordination against any players who are narratively constrained by something other than the zero-sum game. For instance, if a widget-making group isn't under sharp performance pressure, anyone who's focused on actually making the widgets is going to have a hard time staying in lockstep with the group story, and is therefore the easiest target for expropriation and exclusion.

(Compare with Sarah Constantin's claim that group-coordination activities like dancing serve as a way to identify and exclude people who are out of sync with the whole. This stands in some tension with her more recent claim that people should exaggerate differences in order to have some social standing within a group.)

The Scapegoating game

What if you try to play the Survivor game in the real world, where there are other games going on? Now your environment is not exclusively populated with zero-sum players and strategies, which means that revealing info isn't always an unforced error.

Level 1: Fault analysis

I already mentioned that people trying to coordinate in objective reality will be narratively constrained in ways that make them easier targets for expropriation. But there's another feature of group coordination that's very exploitable in the Survivor game: fault analysis. We try to improve maps to improve productive capacity and mitigate risks external to the social game. An important part of this is revealing flaws in the current arrangement.

If you reveal a flaw, you might try to repair the defect (e.g. getting someone to change their behavior - "the squeaky wheel gets the grease") or you might just discard the flawed part (e.g. punishments for bad behavior, "the squeaky wheel gets replaced"). This is what fault analysis looks like in simulacrum level 1 - the meaning of "flaw" correspond to the anticipation that if you remove the flaw, some objective problem is eliminated or ameliorated.

Level 2: Framing

If there's any amount of zero-sum conflict going on inside the group, the fault-analysis machinery - if coupled to punishment at all - becomes a weapon in the hands of anyone willing to lie. If I want to target someone for expropriation in a zero-sum conflict, I can recruit naive level-1 players by accusing them of some objective flaw - framing them.

Consider the story of the vineyard of Naboth, from 1 Kings 21:

And it came to pass after these things, that Naboth the Jezreelite had a vineyard, which was in Jezreel, hard by the palace of Ahab king of Samaria. And Ahab spake unto Naboth, saying, Give me thy vineyard, that I may have it for a garden of herbs, because it is near unto my house: and I will give thee for it a better vineyard than it; or, if it seem good to thee, I will give thee the worth of it in money.
And Naboth said to Ahab, The Lord forbid it me, that I should give the inheritance of my fathers unto thee.
And Ahab came into his house heavy and displeased because of the word which Naboth the Jezreelite had spoken to him: for he had said, I will not give thee the inheritance of my fathers. And he laid him down upon his bed, and turned away his face, and would eat no bread.
But Jezebel his wife came to him, and said unto him, Why is thy spirit so sad, that thou eatest no bread?
And he said unto her, Because I spake unto Naboth the Jezreelite, and said unto him, Give me thy vineyard for money; or else, if it please thee, I will give thee another vineyard for it: and he answered, I will not give thee my vineyard.
And Jezebel his wife said unto him, Dost thou now govern the kingdom of Israel? arise, and eat bread, and let thine heart be merry: I will give thee the vineyard of Naboth the Jezreelite.
So she wrote letters in Ahab's name, and sealed them with his seal, and sent the letters unto the elders and to the nobles that were in his city, dwelling with Naboth. And she wrote in the letters, saying, Proclaim a fast, and set Naboth on high among the people: And set two men, sons of Belial, before him, to bear witness against him, saying, Thou didst blaspheme God and the king. And then carry him out, and stone him, that he may die.
And the men of his city, even the elders and the nobles who were the inhabitants in his city, did as Jezebel had sent unto them, and as it was written in the letters which she had sent unto them. They proclaimed a fast, and set Naboth on high among the people. And there came in two men, children of Belial, and sat before him: and the men of Belial witnessed against him, even against Naboth, in the presence of the people, saying, Naboth did blaspheme God and the king. Then they carried him forth out of the city, and stoned him with stones, that he died. Then they sent to Jezebel, saying, Naboth is stoned, and is dead.
And it came to pass, when Jezebel heard that Naboth was stoned, and was dead, that Jezebel said to Ahab, Arise, take possession of the vineyard of Naboth the Jezreelite, which he refused to give thee for money: for Naboth is not alive, but dead.
And it came to pass, when Ahab heard that Naboth was dead, that Ahab rose up to go down to the vineyard of Naboth the Jezreelite, to take possession of it.
And the word of the Lord came to Elijah the Tishbite, saying, Arise, go down to meet Ahab king of Israel, which is in Samaria: behold, he is in the vineyard of Naboth, whither he is gone down to possess it. And thou shalt speak unto him, saying, Thus saith the Lord, Hast thou killed, and also taken possession? And thou shalt speak unto him, saying, Thus saith the Lord, In the place where dogs licked the blood of Naboth shall dogs lick thy blood, even thine.
And Ahab said to Elijah, Hast thou found me, O mine enemy?
And he answered, I have found thee: because thou hast sold thyself to work evil in the sight of the Lord. Behold, I will bring evil upon thee, and will take away thy posterity, and will cut off from Ahab him that pisseth against the wall, and him that is shut up and left in Israel, And will make thine house like the house of Jeroboam the son of Nebat, and like the house of Baasha the son of Ahijah, for the provocation wherewith thou hast provoked me to anger, and made Israel to sin. [...]
And it came to pass, when Ahab heard those words, that he rent his clothes, and put sackcloth upon his flesh, and fasted, and lay in sackcloth, and went softly.

King Ahab, a level-1 player, sees no way to acquire his neighbor's vineyard lawfully. But his foreign queen Jezebel, used to higher simulacrum level royal politics, sees no impediment to simply framing Naboth, a simulacrum level 2 tactic.

Elijah sees this as an existential threat, flips out, and yells at the king that he deserves the death penalty for this, since by going along with this he's raised the simulacrum level of his kingdom, making object-level coordination harder in a way that can, if it goes too far, become irreversible. Ahab, still a level 1 player, accepts the validity of Elijah's critique and tries to learn his lesson.

When the Survivor game is coupled to fault-analysis in this way, it becomes the Scapegoat game. If the simulacrum level 1 players are naive about this, a minority of zero-sum players can quickly acquire an advantage, since they're working harder to avoid becoming expropriation targets.

Level 3: Prosecutorial discretion

When enough players are mainly using fault-analysis to play the Scapegoat game instead of to fix things, the penal code can be redefined so that nearly everyone is technically guilty of some serious crime, and prosecutorial discretion is required. Then, you don't even need to lie to target someone (thus opening yourself up to expropriation for the crime of lying) - since everyone's guilty, actually being guilty of a crime doesn't single you out anymore. The crimes that get punished are the ones where the governing majority sees a shared interest in expropriating from someone. This is simulacrum level 3, where there's no underlying consistent mapping of crimes to punishments that would be good if enforced, just a standardized list of approved attacks.

Consider the case of Martin Shkreli, who everyone hated because of some perfectly legal price gouging (not morally innocent or sympathetic like Naboth, but not actually criminal), and was consequently prosecuted for the common and totally unrelated crime of securities fraud. There's not really a norm against securities fraud in the sense of effectual coordination to prevent it from happening, there's just a norm that it's a valid accusation. It's increasingly expensive to be innocent.

(But Martin Shkreli was a bad guy and deserved prison? Whatever. Once we're arguing about that instead of trying to criminalize the behavior we actually object to, we've abandoned the pretense that the penal code is a serious attempt to represent which behavior we intend to punish.)

Completely fictitious crimes like witchcraft are a natural outgrowth of this, provided there's a mechanism for confirming that some such claims are true and therefore that the target should be punished. At the limit, we start to see fully general fault-assignment stories, such as such as the Original Sin of Adam and Eve, for which humanity was punished with babies, crops, and the ability to kill snakes, or St. Andreas's Fault, for which Californians are punished with earthquakes.

Level 4: Shoot the messenger

Finally, at simulacrum level 4, people stop tracking the objective meaning of the law even locally, and it collapses to the pure Survivor game again. Prosocial behavior like revealing information about other people's crimes (e.g. Edward Snowden and Reality Winner, but also Frank Serpico) can be enough.

Good and Evil in the Færie courts

There are also natural coordination strategies between groups within a mixed simulacrum level blame game. One natural coordination mechanism for a majority (which has some control over which accusations are followed up on) is try to avoid being blamable for anything by only expropriating in "legitimate" ways that have narrative cover. This allows them to expropriate from others without being punished, and to recruit level-1 players who still take fault analysis literally into their coalition. The price of this coordination strategy is that they can't coordinate overtly. This kind of coalition tends to fly the "good" flag - in Lexical Doll's Seelie and Unseelie Courts paradigm, this is the Seelie Court.

The complement to this coalition is the Unseelie Court, or "Evil," which is willing to be maximally blameworthy. While the Seelie court coordinates to avoid any of its members being blamed, the Unseelie court aestheticizes blameworthiness. Both courts are fundamentally defined by the blame-allocation game.

The "Evil" strategy allows the Unseelie to more overtly coordinate to expropriate from others via mechanisms other than the blame game. Overt coordination - especially on otherwise-unobjectionable things that are simulacrum level 3 crime - makes the Unseelie Court sympathetic to a different class of level 1 players, who see and like that "Evil" is making concrete improvements to the world. The downside of this strategy is that "Evil" is structurally incapable of excluding bad actors, unless it gets big enough that it wants to convert from "Evil" to "Good."

Until recently, Google was "Good" and Uber was "Evil."

"Good" is winning. "Evil" is winning. Who's losing? The level-1 players who just want to fix the things that are wrong and don't want to expropriate from anyone.


Related: Talents, Model building and scapegoating

New Comment
13 comments, sorted by Click to highlight new comments since:
(Compare with Sarah Constantin's claim that group-coordination activities like dancing serve as a way to identify and exclude people who are out of sync with the whole. This stands in some tension with her more recent claim that people should exaggerate differences in order to have some social standing within a group.)

Hmm. So I was about respond that this seemed relatively straightforward to resolve. Then I read Sarah's twitter thread and it seems even more straightforward – AFAICT the two claims are literally the same claim. Group coordination activities are about seeing who is out of sync. Performative opinions are about seeing who is out of sync. You're incentivized to exaggerate your opinions such that they are a costly signal of group affiliation with whichever group you're trying to cohere with.

There's a different thing I thought you meant initially, which is that you also are incentivized (in a more complex fashion) to have contrarian opinions within a group. But this is a more dangerous game – it's the the game that distinguishes high and lower status people within the group.

Low-to-medium status people perform the standard default opinions. Weird opinions usually result in you losings status, but if you successfully gamble that people will end up liking your opinion, or fitting it into the overall paradigm of the group, distinguishing yourself allows you to become a schelling choice for gaining power as well as for getting scapegoated.

I think the emphasis on "distinctive" is important here. Trying to dance in sync means trying not to be distinctive. You only want to be clearly located within the social space insofar as there's more than one political coalition, and even then you don't really want to be locatable personally like a named cartoon character such as Waldo, you want to be clearly tagged as a member of a comparatively undifferentiated class. Unless you're making a bid to be king or something.

I fear you're assuming a consistency of expectations that leads toward a vastly oversimple model of incentives. Groups value some kinds of diversity, and some instances of deviant competence (unusual things that work out), while punishing some that on many dimensions seem very similar to the things they reward.

Details matter, and path-dependencies abound (where you can't get from here to there without a viable intermediate which may not be optimal for anything).

I want to be uniquely desirable to my mate and to my employer, to avoid competing on a level field with others, and to avoid the admission that I'm just one of 7.5 billion living humans, and so can't actually be all that special. I simultaneously want to signal that I'm predictably productive on standard dimensions, and a positive outlier on at least one. Counter-signaling is an important part of this. Showing that I deviate from norms on some (harmless) dimensions is an indicator that I'm competent enough in more important dimensions that I don't have to care about conformity.

You have to dance in sync for some parts of some songs, and show your unique strengths (which includes showing weaknesses and neutral distinctions as part of counter-signaling) during other parts of the dance. even line dances and highly-prescriptive victorian dances have segments where some amount of freestyle performance is beneficial.

I agree that mating complicates this substantially.

You only want to be clearly located within the social space insofar as there's more than one political coalition

I think I'm disagreeing about how much information you need to convey to locate yourself in location space. You don't want to stand out from the others in the dance. That doesn't mean the dance itself isn't extremely distinctive. There's a huge difference between a rave, square dancing, line dancing, being a Las Vegas showgirl, being one of the ensemble cast in a Broadway show.

Your entire model here is about zero-sum games, and being able to survive coalition politics. Insofar as there isn't more than one political coalition, the entire model seems less relevant. (But, there are almost always multiple coalitions so that doesn't seem to be a problem)

Second, even if you're just a small group in the wilderness and there are no other humans, if you want to preserve tight social bonds I suspect it's still important to have a distinctive culture.

Insofar as there isn't more than one political coalition, the entire model seems less relevant. (But, there are almost always multiple coalitions so that doesn't seem to be a problem)

This varies a lot depending on structural factors. The survivor game can show up within a largely homogenous group, and within that group one doesn't want to be distinctive.

In intergroup conflict, from one group's perspective, you want outsiders to all seem distinctive, and that has to involve making yourselves distinguishable from them. But from the inside that feels like your group just looks kinda normal, and everyone else is "ethnic" or "exotic" or "has an accent" or "uses jargon" or "has weird rules" or some other stereotype.

I think you want the ingroup to seem "normal", but I think there is selection pressure on groups for their "normal" seeming beliefs to require costly signals. (Or, in general, group cohesion is stronger when group membership requires costly signals. One such type of signal is What Group Members Believe. If your social games depend on that particular input, your beliefs are going to be weird enough to distinguish the ingroup from the outgroup)

(A lot of my thinking is downstream of this post by Scott from awhile back)

I really enjoyed your description of prosecutorial discretion, and wanted to explicitly state that these conditions occur on a socio-political level constantly. There's no question that if the standard of prosecution for appropriate commentary on Twitter outlawed tweets from Sarah Jeong or James Gunn, then just about everyone is guilty, but simply haven't raised their profile enough against some adversarial interest to be worthy of prosecution.

It also seems to me that given the fast and evolving understanding of what is appropriate to say on a public platform, and the archival nature of the internet, it is inevitable that comments once deemed appropriate will slip into inappropriate and eventually outrageous over time, solidifying the ability for adversaries to find skeletons in our digital closets.

All equilibria are a balance of at least two forces. You're laying out a case for the forces pushing toward privacy/lying/hypocrisy side of social interactions. What are the forces opposing them? Why is anyone trying to tell the truth in the first place?

Are there any advantages to playing the level 1 game, or does this all boil down to "human implies political, get good at it"?

Level 1 has a massive advantage in any sort of conflict where control over objective reality matters, such as shooting wars or ability to produce verifiable miracles. It also has a large enough survival advantage inside sufficiently harsh environments, since they produce "objective" feedback (e.g. marginal agricultural communities in cold areas, but also businesses under severe competitive pressure with tight margins, and armies fighting wars - incompetent people get weeded out fast when top level people see an overwhelming interest in achieving some difficult objective task).

The Elijah story is followed by the Elisha story where one dude outmaneuvers multiple heads of state, replacing them with puppets with varying levels of allegiance to him, by paying close attention to what's going on. But you can only do this if you know about and can detect the other levels

Hmm, maybe I don't understand the levels. I'd assumed that higher levels include lower ones, rather than denying their existence. It's absolutely possible, in my world, to recognize that there is a reality, and to still weigh the social appearance against it.

It's not that higher simulacrum level players can't do level 1 internally, it's that if people play higher levels inside an organization, that organization's information processing is corrupted, and it gets worse at the sorts of things level 1 is good at. There are huge advantages to groups being able to coordinate on level 1, and there are advantages to individuals knowing about all four levels.

To some extent levels 3 and 4, if practiced commonly enough, erode the ability to talk in level 1 language.

The clearest issue with OP's scenarios is that all the "accusations" portrayed involve cheap talk - thus, they are of no use other than as a pure "sunspot" or coordination mechanism. This is why you want privacy in such a world; there is no real information anyway, so not having "privacy" just makes you more vulnerable! Back in the real world, even the very act of accusing someone may be endowed with enough information that some truthful evidence is actually available to third parties. And this makes it feasible to coordinate around "telling the truth" - though truth-tellers still have to work hard at finding the best feasible signals and screening mechanisms! Yes, "human implies political" - but even useful truth-telling involves playing politics, of a sort. (This is something that the local OB/LW subculture is not always ready to acknowledge, of course. It's why we have a deeper problem with politics here.)