Unintended Consequences

Sunday, May 22, 2011
By dreeves

A picture of a cane toad, epitomizing 'unintended consequences'

David Reiley is an economist and a Beeminder beta user, albeit one who has yet to partake of a commitment contract. He asks the following:

For those of you who have given yourselves big incentives to do something, do you ever find that you are shortchanging other important areas of your life as a result? When you push yourself on an exercise goal or a work goal, do you miss out on sleep? Do you depreciate your “relationship capital”? Or do you have good reasons to believe that the multi-task incentive distortions are not a problem?

David is pointing out the danger of unintended consequences. There’s even a law of economics by that name: Any intervention in any complex system will result in unanticipated and often negative consequences. Joel Spolsky puts it even more starkly in an entertaining article, Sins of Commissions:

Incentive plans based on measuring performance always backfire. Not sometimes. Always. What you measure is inevitably a proxy for the outcome you want.

I brazenly claim that this is not a problem with Beeminder contracts. That’s because the person imposing the incentives is also the one being incentivized. The reason you need to self-bind is that immediate consequences (like the deliciousness of the pie in front of you) act like a drug that distorts your decision-making. Despite my use of the “multiple selves” rhetoric, there’s really just one true you: the one deciding when all the consequences are distant (or when they’re all near). So self-binding with a Beeminder contract means deciding something from a distance and then using a commitment device to constrain your future options so you actually stick to your decision.

That risk pales in comparison with the near certainty of your future self running roughshod over your long-term desires and intentions.

It’s true you can be wrong and the commitment can turn out to be a mistake but that’s a different sense of “unintended consequences” — a much more benign sense. For the kinds of goals suitable for beeminding, that risk pales in comparison with the near certainty of your future self running roughshod over your long-term desires and intentions. (See the definition of akrasia in the sidebar.) And you can always be very conservative with your goals. Forget your fantastical notions about losing 20 pounds by summer and just commit to at least not gaining weight. Something where you know that if you fail it’s overwhelmingly likely that it was due to akrasia.

To answer David’s original question more directly: No, I’m beeminding exactly the areas of my life that I’m shortchanging to begin with. I pre-commit to, for example, working more because I don’t like where my time goes when I decide moment to moment how to spend it. If I overdo it and find I’m working too much then I’ll just dial it back down next time.

UPDATE: This was originally published when Beeminder was still in private beta. We’ve come a long way since then! In particular, we now have the Akrasia Horizon, which means you can change your mind about what you’ve committed to much more easily. You’re only on the hook for what your past self thought was a good idea for the next week.

UPDATE: Also related is Goodhart’s Law which this post argues that Beeminder seems pretty robust to. Thanks to Will Eden.

Tags: , , , , , ,

  • Anonymous

     This is a good point – you should not be beeminding the items that aren’t important.  Also, if something is being shortchanged as a result, then you can take two ways I think.  Either you decide that the thing you’re beeminding is more important than the thing your shortchanging in which case you might need to shortchange those other things – time is a zero sum game, you can’t spend time on one thing without other things.   Alternately, it’s just as important and you should quantify that as well, or it’s more important and you should stop beeminding the first thing (maybe switch it?)