Let’s play a game. Some of you may already know this game; if so, please don’t spoil it for everyone else.

Everybody, pick a number between 0 and 100. Whoever guesses closest to 2/3rds of the average of all the guesses, wins.

These kinds of puzzles are called Keynesian Beauty Contests and they are interesting, as they explore a unique area of ideaspace. Most kinds of contests and direct: Guess The Correct Answer. But the KBC adds a layer of indirection to the process. You’re not supposed to guess The Correct Answer. You’re supposed to guess What Everyone Thinks The Correct Answer Is. When experiments are run comparing the results of these different types of questions, researchers get radically different results.

It occurs to me that, in a sense, leadership means “winning the Keysean Beauty Contest game”. We like to imagine that great leaders do The Right Thing, but if they did that, they wouldn’t stay in charge for long. Like it or not, a large group of people will not all agree on what The Right Thing is. Perhaps they have different starting data, coming to different conclusions (which may or may not be legitimate). Perhaps they have conflicting interests, and associate The Right Thing (generally) with The Right Thing for them. Perhaps they are not the smartest people, and cannot properly grasp The Right Thing. Or perhaps it’s simply a noisy communciations channel, and something gets corrupted between their brains and yours.

Leaders find themselves playing this game. Their actions are constrained by popular opinion, and they must, in general, attempt to predict what everyone thinks the right thing is, and do that. Often times, this will already coincide with what they think the right thing is (this somewhat follows from the definition of “good leader”), but sometimes there will be a conflict. When such a conflict arises, what do they do?

There is a flaw in my formulation. Doing “What everyone thinks the right thing is” pre-supposes that everyone has coherent ideas of what the right thing is. It pre-supposes that people have ideas about the right thing, that are communicable to and understandable by a third party. It also pre-supposes that these ideas have a foundation. They will not change suddenly or arbitrarily. But are these facts true?

Here is a graph of Trump and Clinton approval ratings last year. It claims to be from Nate Silver, though I found it in a Google image search so maybe it is not. In any case, it highlights my point.

Imagine the polling question is changed from “who do you want to win the 2016 presidential race?” to the equivalent question “who do you want to have control over the button that ends life on planet Earth?” While somewhat manipulative, my question is still literally true. Now, while imagining that all along we were asking this manipulative question of survey respondents, ask yourself: do you really think that peoples’ answer to this question would swing so wildly? Do you really, truly believe that a rational person would on day 1 say “I have evaluated the data and decided that Trump is to be trusted with the nuclear football”, but, only a month later, change their minds and say “woops actually no he’s bad, I’d rather Hillary have it”?

I don’t believe this, and this implies a rather unpleasant conclusion: public opinion (“The Thing Everyone Thinks The Right Thing To Do Is”) doesn’t real. It does not represent the deep-seated beliefs and goals of the populace, at least not entirely. There is a component of it, a large component, that is highly context sensitive and subject to change with no fundamental reason. This admits a second, degenerate strategy, for rulers who aren’t really that great: The people don’t like your leadership? Let’s dissolve the people and elect a new one!

Framed another way: there are two different kinds of “what people think the right thing is” (hereby shortened to “the standard”). There is the descriptive standard, the platonic ideal of what I’m talking about. This is the honest expression of peoples’ collective goals, “the real answer” to the KBC. But there are also prescriptive standards. After all, if there is a lot of leeway in what people think The Standard is, why not pressure that leeway to shift in a way that makes your life easier?

This may be a mechanism by which preference falsification cascades happen. Because of the indirection of the question “what do you think other people think”, someone who influences sources of information can put a thumb on the scale. Depending one’s influence within society, as long as the gap between “what people think” and “what people think that people think” doesn’t get too big or noticeable, a bad leader can coast along on prescriptive standards for quite some time.

Even worse than this, though, is the power of prescription in the hands of a would-be social reformer. Someone who sincerely believes in the moral superiority of their position. Convinced of the fundamental goodness of their position, they will not exert as much self-control over the prescription of standards.

Ideally, if done slowly enough, this can cause lasting changes in society and, possibly, for the better. By slowly convincing everyone that everyone else is becoming more pro-social, you can incentivize everyone to become more pro-social! But it’s a knife edge. Human nature is only so malleable. And each individual case of success empowers more and larger future prescriptions.

This is a viable model to understand a lot of the culture war that we’ve seen recently. There have been moral reformers at the top. Undoubtably some have been cynical and self-serving, but we’ll exercise charity today and say that they sincerely believe in the goodness of what they’re dong. But they long ago stopped being good leaders. As I said earlier, good leaders honestly and fairly reflect the standards that are already there. They do not attempt to change those standards through manipulative fudging of the numbers. Ours, on the other hand, do.

Maybe this was ok. Maybe it was even good. But it can only go so far. We appear now to have become caught up in a feedback loop. Leaders twist, manipulate, and falsify information at the margins in order to push the prescriptive standard towards an ideal. They they observe the twisted reports as evidence that their prescription changed the descriptive standard. They follow through with another round of prescription, unaware that the first one had not yet completed.

Eventually, that gets us to where we are now: bubbled off elites, attempting to do good, but provided with such absurdly incorrect information that everything they try falls somewhere between ineffective and dangerous. Meanwhile, frank and honest discussion of the descriptive standard becomes impossible, as enough people outwardly (and possibly inwardly?) accept the prescription as the description, and write off dissidents as evil or stupid. It’s a powder keg, waiting to go off.

It’s as if someone came along to the 2/3rds game and started telling everyone: “2/3rds of the average will be X. You should pick X”. It might work once or twice, but it will not work forever.

The solution? Radical honesty and realism. A recognition that people aren’t perfect, and we shouldn’t pretend they are. A deep-seated skepticism towards utopians. A healthy, rational, self-informed and self-aware electorate that is not snowed by manipulated data. Open and honest communication between people, so that people have a well-grounded idea of what others think. A recognition that the power of leaders to change culture is dangerous and should be avoided.