Buttons, Blenders, and Coordination

Framing effects make a big difference

Every now and then, a Red vs Blue Button poll goes viral on Twitter:

Everyone in the world has to take a private vote by pressing a red or blue button. If more than 50% of people press the blue button, everyone survives. If less than 50% of people press the blue button, only people who pressed the red button survive. Which button would you press?

This framing makes Blue sound like the obvious and ethical choice. But the same payoff matrix is found in Roko’s Blender:

Everyone in the world has to decide whether to step into a giant blender. If at least 50% of the people do step into the blender, it will be unable to overcome their inertia to get started, and everyone survives. If less than 50% of the people step into the blender, then they all get blended up into paste and die. People who do not step into the blender suffer no adverse effects. (Blue = step into the blender, Red = don’t do that)

Other analogies I’ve seen include Blue = take a poison pill (everyone gets an antidote iff¹ a majority take the pill), or Blue = pick up a loaded gun (and shoot yourself in the head iff only a minority of the population made the same choice).

As far as the objective payoffs go, the relevant conditions to note are:

• Everyone who picks red is guaranteed to survive, but if a majority pick red then every blue-picker dies. If a majority pick blue then everyone survives.

• The only conditions under which your choice makes a difference to the outcome are:

  • If an independent majority pick red, then your picking blue increases the death count by 1 (yourself).

  • In the unlikely event that your vote determines the majority, picking blue saves half the population.

  • (If an independent majority pick blue, then your choice doesn’t matter.)

Since picking blue can either make things better or worse, depending on what others choose, the “correct” answer depends on your empirical predictions about human psychology. (One sees online partisans arguing either that there’s “no reason” to pick blue, or that red is simply “selfish”, and I don’t think either claim is correct. It’s not selfish to recognize that your own life has value, and there are clearly altruistic reasons to pick blue if you expect enough others to do so. What I think the red-partisans get right is that there’s a sense in which the “first”² people to pick blue are being deeply irrational and creating a problem that a society of logically perfect beings would avoid. But one has to deal with humanity as it is, not as we’d wish it to be, and saving others from lethal “mistakes” is a good thing to do if you can manage it.)

In Blender-style framings that make the blue option intuitively extremely unappealing, you should obviously choose red. (Blue is gratuitous suicide when you can easily predict that only a tiny minority will choose it.) In the original version, where the framing makes Red sound “selfish” and Blue “cooperative”, you might predict that the result will be close, in which case it would be impartially worthwhile (in expectation) to pick blue.³ So there’s no purely logical answer to the problem. It depends on your expectations about others’ behavior, which should probably be extremely sensitive to framing effects.

Note that both options are universalizable: Red has the feature of being safe (for you) no matter what others choose, which seems likely to appeal to people thinking about this from within a game-theory frame. Blue has the feature of being more forgiving of imperfect coordination: you still get the optimal outcome if most-but-not-all successfully coordinate on it. Framing effects can change which feature is more salient. Insofar as these roughly pattern-match to rational vs empathetic vibes, you can see how it’s a great case for polarizing people into rival mutually-contemptuous camps. “Everyone should think like me and those who don’t are stupid or evil” is a perennially popular form of self-expression :-)

1. “iff” = “if and only if” in philosophy jargon.

2. In a non-temporal sense of “first”. Like, those who pick it with the least amount of reasoning or modelling of other minds.

3. In Twitter polls, Blue tends to win by a moderate amount, but if people’s lives were really on the line I suspect more would choose Red.