Precommitment Considered Harmful
(Disclaimer: whenever I talk about some sort of personal practice or habit being potentially not useful for harmful, I mean that it may be not useful or harmful for some people, not that it can't be useful for anyone or that it will always be harmful to everyone. People are diverse enough that I expect most personal practices and habits to work for some people and not others. I would have phrased the title in a less provocative way were it not for ancient and hallowed tradition.)
So, one of the memes floating around in the "rationalist" community ("rationalist" is an unfortunate misnomer; it has nothing to do with philosophical rationalism; it's a community and memeplex spawned from the blog Overcoming Bias and the website Less Wrong) is the idea that you can do "precommitments". This is an idea that has always intrigued me particularly ever since I heard about it. Part of the reason for that is probably that it struck me as a very... exotic idea. Like, a lot of the ideas I've been exposed to via the rationalist community are basically just elaborated and expanded versions of ideas I'd already arrived at myself at an early age. But precommitment was genuinely new to me. Before I started associating with rationalists I had never done anything like a precommitment and I didn't even conceive of it as something that it was possible for me to do.
As I understand, precommitment is just committing to do something before you actually do it. That sounds simple enough, but the key thing is that there's a time delay between the committing and the actually doing. And commitment means commitment—so if at any time between the committing and the actually doing you decide not to do the thing, or you even are theoretically capable of deciding not to do the thing, then you aren't really committed and you didn't really precommit. Note: the ordinary word "commitment" can be used to describe the same thing as precommitment, but it's a bit more vague, so I guess "precommitment" is used just to make it clear about the time delay and the unshakeability of the commitment.
Now, since I was introduced to the concept I've made attempts a few times to precommit to things. However, I have not once succeeded at doing this when it counted (i.e. when I didn't actually want in the moment to do the thing that I had ostensibly precommitted to doing already).
The problem is, if I don't want to do something in the moment (after considering all the potential benefits and costs; of course I can do things that I wouldn't want to do if the benefits and costs were different) then I don't want to do it and I'm not prepared to let anything override that. My present self is a stubbornly autonomous person who's not about to let himself get bossed around by a past self with strictly less knowledge of the circumstances at hand. Indeed, if you take this conception of the past and present selves as distinct seriously enough, you might say that it's unethical for me to be forcing my future self to do things without its consent. The autonomy of my future self ought to be respected. And I wouldn't be surprised if it turned out that this sort of coercion towards one's future selves can end up having long-term negative effects on executive functioning, willpower, etc. But all I'm asserting here is that it's simply not in practice possible to force the future self to do things it doesn't want to do, for some people at least.
So what should I do instead of precommitting to things? Well, I guess I just have to trust that my future self will want what's best for me. I can try to set things up so that my future self's desires are likely to be consistent with my present self's desires. But ultimately I don't have the level of control that "precommitment" implies. It's still fundamentally down to what my future self wants to do in the moment. The "precommitment" framing has encouraged a sort of antagonistic relationship between my selves at different times, where the past self tries to keep the naughty future self from misbehaving, but really it's probably better for it to be co-operative. The past self can give the future self some friendly advice and it's up to the future self to act on it; of course if the future self acts on it then he can notice whether things work out well or they don't, and likewise if he doesn't act on the past self's advice he can notice whether things work out well or they don't, and that can influence its subsequent decisions.
I decided to have a look at some actual LessWrong content before making this post just to check I had the right impression about what "precommitment" means in the rationalist memeplex (I think I do), and I came across Anti-Akrasia Reprise, a post by user dreeves, which contains this interesting paragraph:
And although I'm using the multiple selves / sub-agents terminology, I think it's really just a rhetorical device. There are not multiple selves in any real sense. It's just the one true you whose decision-making is sometimes distorted in the presence of immediate consequences, which act like a drug.
This may go some way towards explaining why the "precommitment" meme is popular among rationalists, but seems weird to me. You see, I totally think people really are a string of separate selves at different times. I mean, time is continuous so there's gradation between them, and obviously these selves share some attributes that the selves of different people don't share, such as being in the same body and not being able to meet each other. But I do think that acts such as personal planning, stating intentions, etc. are fundamentally similar in how they work to negotiations between different people, and I also think that all the momentary selves are equal in their status and autonomy: I can't just dismiss a future self as being subject to distortion due to "immediate" consequences. Immediate consequences are consequences too!
This idea of past selves and future selves is another one I think of as associated with the rationalist community—I think I picked up that way of phrasing it from there—but for me it goes in the other box from precommitment: it's one of those that I kind of always had already. But maybe the... affect I associate with it is quite different from that of rationalists. I've sometimes got the impression that rationalists consider the existence of past selves and future selves a scary thing, a sort of flaw. Even if, unlike dreeves, they'll agree that different past selves and future selves can exist in a real sense... they have this idea that ideally they'd all have consistent desires, so that they'd be realizations of a single timeless self. Maybe this stems ultimately from the Gandhi murder pill thought experiment and the idea of Coherent Extrapolated Volition. Involuntary modification of the self by some external force especially is something I've noticed many rationalists regard as particularly horrifying, to an extent that I don't share. Anyway, I have a different perspective. I think people change as time passes, and this is natural, and you don't really get to decide how you will change.
This is a sort of fatalistic attitude, and rationalism as I understand it is very much opposed to fatalism. I've always been aware that rationalists are much less sympathetic to fatalism than I am (it's one of the starkest, most salient aspects of the community's collective psychology). So that's how I would put these differences I've identified between my thinking and those of rationalists in general in context.
