Are Sunk Costs Fallacies? (2012) | Hacker News Reader

Are Sunk Costs Fallacies? (2012)

gwern.net

18 pointsyamrzou8 comments

> much of what we call “sunk cost” looks like simple carelessness & thoughtlessness

Politics make the sunk cost fallacy pervasive in business and government.

If you’re in management in most companies, you get fired if a signature initiative of consequence fails.

So the actual calculus is to either declare defeat and essentially fire yourself today, or risk a bit more of the company’s capital (not your own!) to see out the x% chance that the venture succeeds.

Yeah, I think that’s almost kinda the thrust. That sunk cost is a fallacy only at an organizational scale and at the individual scale it’s merely strategy or thoughtlessness.

From the article:

> People commit sunk cost much more easily if someone else is paying, possibly in part because they are trying to still prove themselves right—an understandable and rational choice!

This article seems to argue that sunk costs are not so much a fallacy beccause they often come with real ongoing costs that make them hard to ignore. This just begs the question. The reason we harp on sunk costs is for exactly this fact. If they were easy to ignore, no one would harp on them. You never hear about the "don't play dice with strangers in an alley" fallacy because that is a fairly easy trap to avoid.

Sunk cost itself is not a fallacy; it's usually coupled with something else that is a fallacy.

For example, in the case of gambling, the fallacy is that future random events are influenced by past random events. If you pulled 99 times on a slot machine with 1% chance of winning, your chance of winning on the next pull isn't 100% or even 50%; it's 1%.

There are many cases where past effort does contribute to a payout threshold, which is likely why we evolved the "sunk cost" behavior to begin with. For example, if you spent 5 minutes trying to loosen a tight bolt, it's possible that the bolt has been loosened somewhat and you could fully loosen it in a few more minutes, so giving up is suboptimal.

The problem is trying to determine which cases are which.

The problem is trying to determine which cases are which.

If you don't know there is nothing you can do. But the situations where teaching sunk cost can help is is when you DO know, but emotionally don't want to let go.

Like when you your business is lost, and shouldn't put any more money in, but emotionally are clinging to the "I've spent so much time and effort and savings on this, I can't just walk away".

No, sunk cost is just a fallacy. If you've already put $90 into something with a 50% chance of returning $100 if you invest another $10, with all other things being equal, it's a bad idea to spend that $10 if there's a cheaper $9 option that also has a 50% chance or returning $100.

> For example, if you spent 5 minutes trying to loosen a tight bolt, it's possible that the bolt has been loosened somewhat and you could fully loosen it in a few more minutes, so giving up is suboptimal.

This is exactly the sunk cost fallacy, though. Wishful thinking: "it's possible that the bolt has been loosened somewhat"

If "it's possible," then it's not a sunk cost fallacy, it's the best option. The sunk cost fallacy is when you prefer a tight bolt that you've put a lot of effort into to a looser bolt that you haven't put any effort into.

The best way I know how to express it-

We live short lives, and only have enough time to make a few real investments of our time. We can't throw these investments off too casually, or we end with nothing.

But returns are not known a priori.

Reply on news.ycombinator.com