A mathematician wakes up to find his house on fire. He frantically looks around before seeing the fire extinguisher on the far wall of the room. "Aha!" he says, "a solution exists!" and goes back to sleep.

    -- Popular math students' joke

There has been much discussion of coulds, woulds, and shoulds recently. Agents imagine different counterfactual states of their own minds or actions, then select the most desirable. Something similar seems to happen during political discussions, but the multiplicity of agents involved muddles it a little.

I recently read a letter to the editor in my local paper. The city was launching a public education campaign against binge drinking, and this letter writer thought that all the billboards and lectures and what-not were a waste of time. She said that instead of a flashy and expensive public awareness campaign, the real solution was for binge drinkers to take responsibility for their own actions and learn that there were ways to have fun that didn't involve alcohol.

This struck me as a misguided line of thinking. Consider this analogy: pretend that the city government was, instead, increasing the number of police to prevent terrorist attacks. And that the writer was arguing that no, we shouldn't get the police involved: the real solution was for terrorists to stop being so violent and attacking people. This would be a weird and completely useless response.

Attempts to solve political problems are counterfactuals in the same way attempts to solve other problems are. In Newcomb's Problem, I modify the "my decision" node and watch what happens to the "money in the box" and "money I get" nodes. When I say "Increasing local police would prevent terrorist attacks," I am modifying the "local police" node, and positing that this would have a certain inhibitory effect on the connected "terrorist attacks" node.

The hypothetical second letter-writer's argument, then, is that if we counterfactually modified the "terrorists' attitude" node, then the "terrorist attacks" node would change. This is correct but useless.

But it's harder to see exactly why it's useless. Consider the original argument "We should raise the number of police and train them in counter-terrorism techniques." In this case, I would be counterfactually modifying the attitudes of (for example) the chief of police. But I'm not the chief of police, any more than I'm Osama bin Laden. If I'm going to let myself modify the chief of police's attitude just because it would be convenient, I might as well let myself turn Osama into a pretty decent guy. Yet "the chief of police should train more policemen" sounds like a potential solution, whereas "terrorists should be nicer" doesn't.

Here's one possible resolution to the problem: it's much more likely I could convince the chief of police to train more policemen than that I could convince terrorists to be nonviolent. Since the police chief shares my goal of stopping terrorists, all I need to do is tell them why training more police would accomplish this goal, and the problem is solved. So the reason why "train more police" counts as a solution is that, as soon as the statement and the evidence supporting the statement reaches the right person, the problem will be solved.

In this case, changing the "chief of police" node is really a proxy for changing the "my actions" node. The solution could be rephrased as "You or I could go to the chief of police and sugget they add more policemen, thus preventing further terrorist attacks." Counterfactually changing your own actions is entirely kosher. This also throws into greater relief the problems with "You or I could find and approach all terrorists and convince them to be nicer," or "You or I could go to every single binge drinker in the city and convince them to be more responsible."1

Actually, when phrased like that, the binge drinking example doesn't sound so bad. Add a comprehensive plan for doing it, enough funding to reach them all, and some idea of how you're going to phrase the "be more responsible" point, and it sounds like, well, a grassroots public awareness campaign. Which is kind of ironic, seeing as the letter started out as an argument against a public awareness campaign, and maybe a sign that I'm taking the Principle of Charity too far here.

...in more realistic situations

It's more complex when there are only small probabilities of your own actions having any effect, but the principle stays the same. For example, I recently heard a doctor say that a single-payer system would best solve the US' health care woes, but since that was politically infeasible he was backing Obama's plan. This one doctor's support will have minimal effect on the chances of Obama's plan passing, but it will have even less of an effect on the chances of single-payer passing. If the expected utilities multiply out in such a way as to make supporting Obama more likely to gain more utility than supporting single-payer, the doctor is justified in his strategy of support for Obama's plan.

One more example from real history I learned recently. Suppose you are a Communist, and your fellow Reds are proposing ways to create a socialist paradise. One says that you must incite the workers to violent revolution. Another says you must petition the current government to support labor reform laws. A third says you must petition the current government to oppose labor reform laws.

Before you expel the third communist from the Party, let them make their argument. They say that the Party doesn't have enough resources to incite violent revolution, and the workers don't want to revolt anyway. Counterfactually modifying the "workers' actions" node to a revolutionary state is a waste of time, because there's no link between any modification of your own actions and that node reaching the state you want. Likewise, modifying "government policy" is useless, because the Communists don't have any clout in the government, so even if you found a wonderful value for that node that would make all workers happy forever, you couldn't change it.

Instead, she says, oppose labor reform laws. These are already unpopular, and even a small party like the Communists would probably have enough power to get them shot down. When there's no labor reform, workers will get angrier and angrier, until they gradually revolt and overthrow the system, getting you what you wanted in the first place.

There were communists in the early 1900s who actually tried this third approach. It didn't work, but I admire their thought processes. They ignored solutions that would never happen, and found an action they thought they could enact, that they thought they would raise the chance of revolution significantly. Compare this kind of cunning to the vapidity of the letter-writer who says "Binge drinkers should become more responsible."

...as an unrealized ideal

I like this way of viewing the problem, because it explains why a certain class of argument feels wrong: arguments that go "The solution to binge drinking is more personal responsibility" or "The solution to poverty is for the poor to work harder," or so on. But do people actually think this way?

The most glaring reason to believe they don't is that most people who "solve" societal problems have no interest in actually enacting their solution. The attitude is something like "Hey, if the federal government passed a single-payer health plan, then all our health troubles would be over!" and then don't bother to write a letter to their representative about it or even convince their next door neighbor.

For reasons that have been discussed ad nauseum on Overcoming Bias, politics is very much a signalling game. In particular, it seems to be a game in which you counterfactually propose different states of the "government policy" node and explain why these would have the best effects, and whoever can give the best explanation gets rewarded with higher status. Sometimes you're also allowed to edit the policies of large private organizations, or of influential individuals. In this case, the problem with the original letter writer wasn't just that she had no plan to enact her solution, but that she was breaking the rule which said that you're only allowed to play with relatively unified, powerful organizations, and not things like "the set of all binge drinkers".

In a way, this isn't so bad. When enough people play this game, their opinions get out to the voters, consumers, politicians, and business leaders, and eventually do change government and private policy.

The point is that if your goal is to actually personally affect things in a direct, immediate way, you can't just apply the rules of this game without thinking beforehand. Communists, when discussing politics for "fun", would never say "I think the government should oppose labor reforms," but that might be the winning move for them when they're actually trying to increase utility. Likewise, libertarians spend a lot of time discussing different ways the government could implement libertarian policy, but when they actually have to take action, the best choice might be seasteads or charter cities or something else that doesn't involve policy at all.

And if you are content to just play the game, at least keep it interesting. No fair counterfactually editing things like "terrorists' behavior" or "poor people's work ethic" or "how responsible binge drinkers are".


1: Or, here's an alternate interpretation of "Binge drinkers should be more responsible": it's not worth trying to prevent binge drinking, because binge drinkers could prevent it themselves if they were more responsible, so it's their own fault. This is not illogical, but applying the argument to a case like "Drunk drivers should be more responsible" would be. There, even if we have no sympathy for drunk drivers, we still need to prevent drunk driving because many of the victims are innocents. The other issue is that people process the two statements "The solution to binge drinking is for binge drinkers to be more responsible" and "We don't need to solve binge drinking; it's the drinkers' own fault and we need not care" differently; the first sounds wise and reasonable, the second callous. For both these reasons, I don't think this interpretation is entirely what the original  letter writer, or other people who use this sort  of argument, are thinking of.

New Comment
36 comments, sorted by Click to highlight new comments since: Today at 7:46 PM

it seems to be a game in which you counterfactually propose different states of the "government policy" node and explain why these would have the best effects, and whoever can give the best explanation gets rewarded with higher status.

no, no. The game is to counterfactually propose different states of the "government policy" node that involve making the government conform more to some ideology X, and then confabulate reasons why this would result in great success. In doing this, you signal your allegiance to ideology X.

But really, the game can work with pretty much anything in the place of the "government policy" node; it can be pretty much any decisionmaking entity, including the companies or diffuse classes of individuals. E.g.

"If binge drinkers went to church more, then they would find the inner strength to overcome the addiction!" (signalling religious allegiance)

"Binge drinkers have the right to run their own lives, the government should keep its hands off them!" (signalling libertarian allegiance)

"Binge drinkers usually come from deprived families and had poor childhoods, it isn't their fault, it's the government's fault for not having enough social welfare!" (signalling liberal allegiance)

"Binge drinking is caused by the breakdown of traditional family values, we need a return to the good-old-fashioned traditional family values!" (Signalling conservative allegiance)

"Binge drinking could be prevented by human neuroenhancements that prevented alcohol from being addictive, we should push for faster research into such technology!" (Signalling h+ allegiance)

Oh, very clever. This is a much better explanation for why proposing changes to unchangeable nodes is so well accepted.

I do sometimes wonder what proportion of people who think about political matters are asking questions with genuine curiosity, versus engaging in praise for the idea that they and their group have gone into a happy death spiral about.

I suspect that those who ask with genuine curiosity are overwhelmingly chlidren.

EDIT: Others disagree that children are more genuinely curious. Perhaps it's just the nerds who ask genuine questions then?

And what proportion of that genuine curiosity is an adaptation for gaining information and what proportion is an adaptation that encourages signalling a willingness to absorb the happy death.

What makes you believe that? It's as good a theory that they're just trying to find out what Big Idea group they belong to so they can give the right answers / political suggestions when they grow up.

A priori we should expect children to be genuine knowledge seekers, because in our EEA there would have been facts of life (such as which plants we poisonous) that were important to know early on. Our EEA was probably sufficiently simple and unchanging that once you were an adult there were few new abstract facts to know.

This "story" explains why children ask adults awkward questions about politics, often displaying a wisdom apparently beyond their age. In reality, they just haven't traded in their curiosity for signalling yet.

At least, that is one possible hypothesis.

I do expect children to be knowledge seekers in a sense. When they see their parents avoid a plant, they learn to avoid it also. When they hear them say that binge drinkers should go to church more, they learn to say this also. In both cases it is the same behavior.

The difference between our descriptions is that calling them "knowledge seekers" implies some kind of deliberate rationality, whereas they are really just executing the adaptation of copying their parents. Most children who repeat their parents' political views won't try to understand what the words actually mean, or check different sayings for consistency.

Of course this is a generally good adaptation to have. Even if children had better innate rational skills and even if they could fact-check their parents' words, there's little benefit to a dependant child from ever disagreeing with its parent on politicized issues.

But the greater vulnerability of children means that we should also expect them to be more clannish. They should be all the more eager to demonstrate their loyalty to a group, because they rely more on support from others to remain alive.

I've observed far more clannishness among children than political perspicuity. I don't see that there's much displaying of "wisdom apparently beyond their age" in need of explanation.

I've observed far more clannishness among children than political perspicuity

but what about the relative amounts in children vs adults?

Of course children are more clannish than adults. But the "clan" of a child is that of its parents, not of its friends and peers.

Adults can move to a new clan, band together to start a clan or sub-clan, replace or influence a clan's leadership. Children are pretty much powerless and are tied to their parents' clan. If anything ever really threatens that bond, I expect "clannishness" to completely override other priorities.

The tech might already exist, but prejudices, inertia, and the lack of a strong financial incentive have kept it from being adequately tested.

Maybe if people would eat more fish oil they'd be more mentally flexible.

That website just had claims of "It works!" which by itself isn't all that credible.

How do we know it actually works? Whats the theory behind it?

From http://www.olivierameisen.com/faq

A Baclofen is one of only two substances known to affect the GABA-B receptor in the brain, and the only one that is itself non-addictive. Through the GABA-B receptor, baclofen has a beneficial effect on three neurotransmitters — dopamine, glutamate, and GABA — that are part of the brain’s reward system and are involved in all addictive and compulsive behaviors, as well as in disorders such as anxiety and depression. More research is needed to discover exactly how baclofen does this.

Wow, that really didn't take much legwork. I guess I should have at least looked that far.

Speaking of promising solutions that won't catch on due to predjudices, inertia, and lack of financial incenives, "Psychedelics" (loosely defined) are also pretty effectiveat this if used right.

Great! now that we've both signalled our allegiance to the h+ ideology, would you like to mate with me!?

for an explanation of why I call this "Hansonian", see, for example, this. Hanson has lots of posts on how charity, ideology, etc is all about affiliating with a tribe and finding mates.

Hansionain, twice? Really?

As an aside, I love what you get when you google Hansonian. Most of the top results are in reference to Robin Hanson, and among my favorites are "Hansonian Normality", the "Hansonian world", and "Hansonian robot growth". (Un?)Fortunately, "Hansonian abduction" is attributed to a different Hanson.

I wish my name was an adjective.

A few points:

-Maybe I'm applying the principle of charity way too hard here, but I think you totally misread the first woman's letter. I think her point was more about the "take responsibility for their own actions" bit. In other words, binge drinkers are dodging responsibility for the aspects of binge drinking that spill over onto others (medical bills they don't have to pay, disruptions at bars, drunk driving, etc.), and the law should force them to pay restitution and/or endure full prison sentences and license suspension ("take responsibility"), thus shoving in their collective face the consequences don't they personally bear. Which is a plausible change.

Now, with that said, I do frequently see the kind of fallacy you were describing about terrorism. It's most common when talking about rush hour congestion, where some genius suggests, "the solution is for people to carpool!" (Um, yeah, but the question is how to get those people to carpool.) Even more frustratingly, when you propose a method that would clear congestion -- like very high tolls -- they respond as if buses would be competing against the pre-toll level of traffic and therefore still more time-consuming for any individual to do.

-In discussing political change, I have seen libertarians suggest the strategy in the Communist scenario. When discussing the FairTax, the exchange went like this:

me: "Okay, even if the FairTax would take the same money for me, isn't the greater simplicity and related labor savings a good thing?"

them: "I don't want taxes to be simple! I want them to hurt! So people see why they suck and demand real change!"

Which is a consistent position, I guess, except that everyone who took that position on the FairTax also voiced quite a lot of opposition to anti-libertarian proposals that would conceivably also "make things worse so they can get better". Go fig.

ETA: And of course, those same libertarians' other arguments against the FairTax were reasons why it would be harmful in general, which was supposed to be a good attribute for a tax...

Maybe I'm applying the principle of charity way too hard here, but I think you totally misread the first woman's letter.

In charity to Yvain I don't think 'totally misread' is quite credible in this case. The writer threw her 'shoulding' at the binge drinkers in this letter. She sounds like she would also throw her 'should' at the government to make them enforce her will too but she doesn't go there in this letter.

Asking that someone "take responsibility" for a wrong, can mean either a) an individual decision, OR b) that they be obligated to fix said wrong. "Bob should take responsibility for the mess he made when he sprayed graffiti on the wall."

Given that ambiguity, the principle of charity obligates us to assume she meant they be forced to bear the consequences, and the following remark that "they should learn other ways to have fun" must be understood in this light: i.e. having to pay for what you've done would make you consider alternatives to binge drinking.

That word 'responsibility' can be used ambiguously. It was not in this case.

The assertion that Yvain has 'totally misread' something based on this creative spin makes a mockery of the concept of 'charity'.

Well, all we have is Yvain's paraphrase of what she said. And considering how much he piled on, and how he compared her to a much stupider suggestion, I think more caution was appropriate.

I don't see what's so creative about my "spin"; I frequently hear people use the term "take responsibility" as a euphemism for "take punishment". E.g., "They should take responsibility for what they've done" = "they should receive the appropriate punishment", not "they should privately admit to themselves that they could do better".

I suspect that the communists were just trying to show intelligence by proposing something counterintuitive.

This post is entirely sensible, and is in line with the approach of policy analysts to addressing policy problems.

When you're devising a solution to some policy problem, part of the process is to identify "policy levers", where a policy lever is essentially some variable that is causally entangled with the policy problem and causally entangled with the actions of your agency (or some other agency you hope to persuade).

So the reason why "train more police" counts as a solution is that, as soon as the statement and the evidence supporting the statement reaches the right person, the problem will be solved.

I think we need to get more meta that this- due to public choice theory, we can't even have 'Government Policy' as a freely-adjustable node. If Bryan's right, public opinion is the most important node, so 'Binge drinkers should change their minds' may infact be no harder to achieve.

Also, I think the women's letter could be interpreted as "Binge drinkers could behave responsibly if they wanted to. As they do not, this is not a bad enough problem for the Government to intervene.

Also, I think the women's letter could be interpreted as "Binge drinkers could behave responsibly if they wanted to. As they do not, this is not a bad enough problem for the Government to intervene.

That is a position that a letter writer could take but I don't think that this one can be interpreted that way reasonably.

Maybe a slightly more moralistic version- they should sort themselves out, and it's entirely their own fault if they don't.

Alternatively, her letter is helping to (slightly) raise the social stigma associated with Binge Drinkin.

Alternatively, her letter is helping to (slightly) raise the social stigma associated with Binge Drinkin.

This one seems close.

"You can only (directly) intervene in your own actions; when advocating change to others, give a plan following from their and your action." Agreed.

In a democracy, the most likely outcome of widespread proletariat outrage at anti-labor laws is simply pro-labor reform of those laws. So, the Machiavellian communists must have had some plan to further escalate working class anger before it could take its natural course. "We must destroy X to save it" also reminds me of the motivation of every other bad-movie villain (nothing annoys me more than an illogical villain, except a bad screenwriter).

"You can only (directly) intervene in your own actions; when advocating change to others, give a plan following from their and your action." Agreed.

Even that is perhaps overstating how much influence our abstract thought has over our actions. Admittedly, this is probably (for better or worse) greater in this audience than in most.

This is a quite thorough exploration of the way solutions to political problems invoke counterfactuals. Sufficiently so that apart from our diversion to the signalling implications there is not too mutch to add!

When discussing such problems a counterfactual assumption of influence over one agent must be assumed. Some such assumptions are more useful than others. While at one end of the scale it is futile to assume we have direct control over politically distant agents, assuming for the sake of the discussion that I only have control of my own actions makes the reasoning somewhat complicated. It can be simpler (and more fun) to first analyse the problem with counterfactual control over the government and use this as an intermediate step when analysing what more local actions may be useful.

Attempts to solve political problems are counterfactuals in the same way attempts to solve other problems are.

At least, they would be if they weren't attempts to signal one's own status while also influencing public morality in a way that benefits oneself and one's allies. This mostly explains why the absurd futility of the letter writer's solution doesn't occur to her.

A mathematician wakes up to find his house on fire. He frantically looks around before seeing the fire extinguisher on the far wall of the room. "Aha!" he says, "a solution exists!" and goes back to sleep.

That's a classical mathematician, perpetuator of Hilbert's lunacy. He deserves to burn.

Are you trying to start a flame war?

I did not understand this joke, assuming it was supposed to be funny.

Non-constructive proofs of existence are one of the undesirable features of mainstream or classical mathematics. Some non-mainstream mathematicians have investigated how to modify logic to eliminate non-constructive proofs. For example:


Intuitionism is a broad term for this kind of tendency within mathematics. However, the term seems to mean a lot of different things to different people.


The (original) joke (as opposed to Nesov's flame) isn't necessarily ideological. It simply lampoons the general attitude of pure mathematicians, who by nature are typically more interested in the theoretical existence of solutions than in the practicalities of finding them, leaving the latter for scientists, engineers, etc. (Mathematicians, in other words, are very "meta" kind of people.)