DRMacIver's Notebook

How to make decisions

How to make decisions

The base model for decision making is a coin flip.

I'm not saying that you should flip a coin for all your decisions, I'm saying that any decision making strategy you adopt in preference to flipping a coin needs to be demonstrably better.

Or, when is more normal, you are making decisions between a number of options, your decision making strategy needs to be better than choosing among them uniformly at random (drawing lots, etc).

People think that the best decision making approach given n items is to choose the best item. This is foolish and wrong, which is why the entirety of economics is based on this theory.

(This is a mean dig and is not true, but it's also not unfair. Economists don't @ me)

The reason it is foolish and wrong is that the effort of determining which is the best may exceed the benefit of choosing the best over one of many other options. Consider Buridan's ass (but don't ogle it, that would be uncouth):

Buridan's ass is an illustration of a paradox in philosophy in the conception of free will. It refers to a hypothetical situation wherein a donkey that is equally hungry and thirsty is placed precisely midway between a stack of hay and a pail of water. Since the paradox assumes the ass will always go to whichever is closer, it dies of both hunger and thirst since it cannot make any rational decision between the hay and water. A common variant of the paradox substitutes two identical piles of hay for the hay and water; the ass, unable to choose between the two, dies of hunger.

Dying of hunger is an obviously worse strategy than flipping a coin. I do not recommend it.

But consider... what happens if one of the piles of hay has an extra mouthful on it. That's obviously the better pile of hay, right? You should pick that one and then eat from it, right? But how long will it take you to find out which one has more hay on it? You'll get quite hungry and waste a lot of time doing it, you ass.

So this is the first thing we should learn from the coin flip: If the decision doesn't matter much, just pick something, more or less at random.

What do we do if the decision does matter that much?

The following is a pretty good heuristic solution:

  1. Eliminate any options that are obviously worse than another. Like, if you're being offered £200 or £300 you don't need to think hard to pick the £300.
  2. Identify pairs of options that are obviously more or less equivalent to each other and randomly discard one of them.
  3. If you have more than one option left, maybe gather some more information to improve your knowledge of the situation if there's still a wide range of potential outcomes. Only do this if the cost of gathering information isn't too high or the benefit of gaining it is much higher. Repeat (1) and (2) if this changes anything there.
  4. If you still have more than one option do something like probability matching on what's left, picking an item with probability equal to the (subjective) probability of it being the best outcome. If this seems hard, just pick uniformly at random. Or pick a pair uniformly at random and invest in a bit more effort in deciding between the two.

You almost never need to do anything this formal, but the general idea is:

  1. Don't spend too much effort on eliminating options.
  2. Spend only as much effort on improving your knowledge of the situation as the situation warrants.
  3. If after you've done that it's still not obvious what to do from the remaining options, deploy whimsy and gut feel.