Doing good in a very noisy world

Suppose you live in a world where every time you try to do something good, it gives rise to such a giant waterfall of side effects that half the time the net effect of your actions is bad, and half the time it is good but largely from sources you didn’t anticipate. Also suppose that the analogous thing would happen if  you tried, hypothetically, to do bad things.

It sometimes seems plausible that we do live in such a world, and this sometimes makes it seem that doing good is a hopeless affair.

However I propose that in the most plausible worlds like this, when you try to do good things, in expectation you do a bit of good, and the good is merely overwhelmed by a vast random term, with expected value zero. In which case, even though your actions cause net bad half the time, they have positive expected value, and are about as good in expectation as you thought before considering the side effects. Is that so hopelessness inducing?

If so, consider a related scenario. Every time you do anything, it has exactly the desired consequences, and no others of importance. Except that it also causes a random number generator to run, and add or subtract a random amount of utility from the world, with expected value zero. Does this seem hopeless, or do you just ignore the random number generator, and do good things?

If our world is very noisy like this, is the aforementioned model a good description?

Inspired by a conversation with Paul Christiano, in which he said something like this.

7 responses to “Doing good in a very noisy world

  1. Your two scenarios have very different epistemological implications.

    Suppose we are not certain in advance of all the consequences of our choices or how to rank them (as of course we are not; even if we know for certain that cash transfers raise household incomes in Africa, we still don’t know the further consequences of that); then in world B, where choices cause precisely the planned consequences along with a random unobservable intervention elsewhere, we have been given a very easy causal inference setup: all we have to do is experiment with our various options and monitor the consequences, and we can home in on their value, and easily maximize. The randomness happens elsewhere and doesn’t need to be included in our analysis.

    In world A, on the other hand, we face an almost impossible inference question because any data we collect will be extremely noisy as the desired effects are lumped in with the random effects in the same data – all the side-effects mask the real effect. Without being able to tell what works, world A will in the long run be worse off than world B despite apparently starting off in a similar net situation.

    Given the choice, I’ll pick world B every time, because learning is easier.

    • B is preferable–but not in terms of the argument. It is much easier, it is true, to assess whether efforts to do good actually do even the intended good in A, but the argument assumes that you already know that you’re doing good.

      • > the argument assumes that you already know that you’re doing good.

        To expand on what I said: OP only says you know the immediate consequences of your actions, it doesn’t specify you are logically omniscient and know every ramification of your actions down to the end of time. If you know only the immediate consequences, you still need to investigate beyond that to figure out the gains from particular choices, and then B is much better than A for the statistical reasons mentioned. (And if you are logically omniscient, then that raises a lot of additional issues, like why you aren’t manipulating your choices to beat the RNG.)

        • If you know the immediate effects and are completely ignorant of the long-term effects, it’s reasonable to assume that the long-term effects are random. Your uncertainty about the long-term effects would increase the error variance, but it wouldn’t affect the expectation value of your do-gooding.

          • > The basic weakness of the argument is in the assumption that the unanticipated consequences will be random. There seems to be in societies a not-well-understand tendency (to put it crudely) for actions to create equal, opposite reactions–a kind of social homeostasis.

  2. The basic weakness of the argument is in the assumption that the unanticipated consequences will be random. There seems to be in societies a not-well-understand tendency (to put it crudely) for actions to create equal, opposite reactions–a kind of social homeostasis. (I don’t think this is absolute; that would imply a sort of cosmic direction.)

    To really “do good” (as opposed to advertising oneself as a do-gooder) you need to understand the concrete social forces. You need a sociological theory, not a formulaic utilitarianism.

    One big thing the professional altruists ignore is opportunity costs. What if you could do more good by focusing on professional achievement in the right occupation? Or (per Robin Hanson) saving like a miser? Or, at the opposite extreme, organizing a revolution? What if the do-gooders, by relying on a purely near-mode analysis, are actually diverting thousands of well-meaning youth from what really needs to be done?

    I don’t think it have been better in the U.S. in the late 60s early 70s if youth had demonstrated their idealism by contributing to charity rather than marching against war. But they’re (to some extent) mutually exclusive.

  3. … are about as good in expectation as you thought before considering the side effects.

    Even given the assumptions, this isn’t right. Even given random effects, the expected value will regress toward the mean. It will still be positive, but the noise will cause regression to the mean. If the random factors are overwhelming, the gain will still be positive, but it will be negligible.

Comment!

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s