Comment by roenxi | Hacker News Reader

Comment by roenxi | Hacker News Reader

Parent Full thread

> A rational anaylsis would therefore conclude I'm wasting my time and energy...

No it wouldn't. You haven't established the link between your premise and conclusions and the rational view is the opposite of what you came to. You've established that in the best case the candidate you prefer will win by more than one vote - which is correct but not the end of the line of thought.

Think of it like building an embankment. If you design the embankment to exactly hold back the maximum amount of water you expect it to hold it'll probably fail in an emergency when it shouldn't have because something slightly unexpected happens. The rational thing to do is engineer in a margin of safety. You're trying to optimise the wrong metric which is ~30-60 minutes of a voters time vs. probability of the legislature behaving in a way that is favourable to them. In both voting and embankment building it'd be silly (dare I say, irrational) to aim for a narrow success. The median election (analogy: median storm) should make the margins for your candidate (analogy: embankment tolerance) look excessive (analogy: over-engineered).

Otherwise, you're basically arguing that rational people should optimise their way to being on the losing side of elections - which is a big tip-off you're making a logic error in your argument somewhere. Rational behaviour can't, almost by definition, predictably lead to bad outcomes.

I find the embankment metaphor confusing, but to play along: I don't have the option of engineering a margin of safety because I only have one brick. Is it worth me lugging it to the riverside?

Maybe, it depends on the risk and the distance I've got to carry it.

In all likelihood many won't bother without being directed, encouraged or otherwise socially motivated. And they must ignore the surprisingly persuasive pro-flooding lobby, of course.

> Rational behaviour can't, almost by definition, predictably lead to bad outcomes.

Given you previously mentioned Game Theory, this is a surprising claim. Prisoner's dilemma? Tragedy of the commons?

Note that I'm not sure how much any of this applies to the real world. My main argument is against a model of voters as purely "rational". They are not, and in some ways at least, that's probably a good thing. I certainly think rationality is a very poor argument against extending the franchise to 16 year-olds.

> I find the embankment metaphor confusing, but to play along: I don't have the option of engineering a margin of safety because I only have one brick. Is it worth me lugging it to the riverside?

Yes. You should put the brick where the engineer tells you too, and at the end of it you will observe there are more bricks than were strictly needed. Any other approach would be reckless.

Again, you're arguing that the rational thing to do is to not build the thing properly. You've misunderstood rationality and you're getting nonsense conclusions where people choose to get to get poor results by their own standards because they aren't very good at risk assessment. That isn't how rational actors behave. Rational people take a holistic view of all the foreseeable costs and benefits of an action (like, for example, having a government that aligns with their policy preferences) and are capable of probabilistic thinking.

> In all likelihood many won't bother without being directed, encouraged or otherwise socially motivated.

Well sure, but in practice people aren't rational. If they were rational, they would just quietly assess whether their coalition had the potential to win at some point and - if so - identify that it is in their best interests to vote. Then do it without prompting.

> Given you previously mentioned Game Theory, this is a surprising claim. Prisoner's dilemma? Tragedy of the commons?

The game theoretic optimum in both those cases is to achieve the best possible result through communication and coalition building.

I think this disagreement stems from the fact that, by instinct, you think socially. When I hear the word "rational", I think of the economist's model of a individual acting to optimise their own utility.

Of course in many (most?) situations, we get the best outcome overall by acting socially. But, to me, free-riding on the socially-motivated activities of others is a "rational" strategy (with a insignificant chance of one brick/vote making a difference), if not a laudable one. Which is why I don't advocate the "rational" strategy.

I appreciate your point that if everyone took that strategy we'd all suffer. I can never control what everyone does, but maybe I can build or support a big enough coalition to improve outcomes for the majority, or at least people like me. But free-riding will still be an option many choose, unless punished severely (which most countries don't when it comes to voting).

I don't argue that people "choose to get poor results", I argue that some proportion may recognise their possible effect on the democratic outcome is so small that on some level they have no effective choice at all. That's just the reality of being one among an elaborate of many millions. Democracy is still the best option we have, I hasten to add.

In Prisoner's Dilemma, as originally formulated, I believe communication is expressly forbidden. And even where it's allowed, in a one-off game the optimal strategy for an individual uninterested in the welfare of their comrade is to collaborate with the authorities. If both participants take this strategy, it's to the detriment of both. This apparent paradox is what makes it so interesting. Now you can solve the paradox with communication through repeated games, which I think is what you allude to.

> I think this disagreement stems from the fact that, by instinct, you think socially.

No, the disagreement stems from different beliefs over whether you've taken a rational position. Especially odd since at some level you understand it is an irrational conclusion since you're voting anyway.

What you are calling "rational" probably would make sense in a world where people were unable to communicate and playing a one-off game. The issue that idea runs in to is people can communicate and elections are a repeated game. If voting is modelled as a game where people don't communicate then a lot of nonsense results turn up. The rational strategy for most people when communication is possible is to join up with a coalition that can win, then vote. There is a minority of people with unusual enough political opinions that they can't realistically join a coalition and they rationally wouldn't vote, but by definition they are fringe groups. In the main, most people would vote if they are rational. And indeed, in a move that makes one hopeful for humanity, most people do indeed make the rational decision on that one.

Any study of coalition games like voting quickly discover that voting is entirely rational and a theoretical optimum in practice for most participants in a society. If you do a course on game theory there should be entire lectures on the subject. I suspect you might have done a course on game theory so I'm not sure why that wasn't drilled in. The mechanics of coalition building among rational actors is a fundamental topic.

> In Prisoner's Dilemma, as originally formulated, I believe communication is expressly forbidden.

You can look it up [0]; as originality formulated it was a 100-round game where cooperation is an entirely rational behaviour (for most rounds, anyway). It is a very powerful example of how cooperation followed by tit-for-tat is a near-optimal strategy under a lot of realistic assumptions and requires only the tiniest of communication channels to pull off and improve from an inefficient Nash equilibrium to a Pareto efficient one.

In fact, I suspect the actual mistake you're making is thinking that a Nash equilibrium is equivalent to a rational one, when in fact it is not. When communication and coalitions are possible the Nash equilibrium is usually just a starting point for negotiations before the rational agents decide to get a better result for themselves.

[0] https://en.wikipedia.org/wiki/Prisoner's_dilemma

I was uneducated on the subject of coalition games, so looked it up. But I leaned these games are defined by a mechanic where players can form binding contracts with external enforcement/punishment.

I don't see how this applies to voting at all in countries without compulsory voting.

Communication and iterations also don't seem that relevant to my argument about free-riding, since one's voting record is not usually visible.

Reply on news.ycombinator.com