« Friendly Teams | Main | Boston-area Meetup: 11/18/08 9pm MIT/Cambridge »

November 15, 2008


I've been thinking a little lately about the difference between doing something useful, and doing the most useful thing.

Who was the economist who won a Nobel price proving that any company going for the optimum would go bankrupt? The problem is: the time and resources spent trying to find the optimum makes this a no go. Conclusion: be a satisficer, not an optimizer. The other points in your post are valid though.

Could you go so far as to set theoretic upper bounds? Or even just practical ones. What is the most good that $700B has ever done? This would at least give a scale.

It amounts to saying 'look for global maxima' rather than 'local maxima' which is essentially like saying 'be smarter!'

The question to throw more resources at another question is itself a thorny question. Are 18 month long presidential elections better than 12?

Outofculture - one of the motivations for the post is the new book about what else could have been done with the $1T spent on the Iraq War. Perhaps developing a set of standard items would help people be more opportunity-cost numerate?

between doing something useful, and doing the most useful thing
1) It's often VERY hard to distinguish between utility of different actions. You're usually faced with something more like three categories of option:
a) immediately and obviously useful
b) more useful, but more risky
c) wait until #b is clearer or closer.

I believe there are different situations in which each of the three have merit.

In summary: consider opportunity costs

@Roland: You're thinking of Herbert Simon http://en.wikipedia.org/wiki/Herbert_Simon

I've been thinking a little lately about the difference between doing something useful, and doing the most useful thing. The latter is a lot harder, yet a lot more productive.

Is it? Can you demonstrate that "the most useful thing" is much more productive than "doing something useful"? Can you even defend the claim?

One tidbit of ancient wisdom seems to be genuinely useful: concentrate on the problems you have here and now, keep your mind on what you're doing and not what you're going to do or could have done.

Never thinking ahead or about alternate possibilities is lethal. But we spend far more time and effort on such than we need or is good for us.

@Caledonian - I agree that, when working, being present is valuable. But what people choose to work on matters enormously. And I think the default mode for many people (certainly for myself), is to work on things that "seem" urgent, those that grab attention the best (answering unread emails, for example), rather than what is truly important.

I'm surprised the claim needs defending. What I mean by "the most useful thing" is, by definition, the most productive - that which will advance your goals the most for the least effort. Sure, costs in determining that matter. But it's almost tautological that that's what is best to work on.

@Dagon - I am trying to make a simpler point. I often find myself doing things which "feel urgent" but are not long-term critical, and interfere with long-term critical tasks. Of course it is hard to determine the best strategy. But most of the time, I think most people don't even try. They do what seizes their attention.

Great post. Have you found the productivity techniques you list helpful in overcoming the urgency-vs.-importance bias?

Oh, this reminds me of Eisenhower priorities, named after the president and general. He, I believe, suggested you categorize things to do in terms of important and urgent. You'll have four categories then, the important and urgent, the important but not urgent, the urgent but not important, and finally the not important and not urgent. The urgent and important are first priority, followed by the important but not urgent, then urgent but not important, and the not important and not urgent. I think Eisenhower said something like 'the important is seldom urgent and the urgent seldom important.'

Would optimum be more practical if we understand it not as 'best' but as 'best that occurs to you'.

And does our action reveal what we really think is more important?

If someone had 1 day's advance notice on trading markets, so that they knew for each 24 hours what the most lucrative trade would be, they could go from being very poor to being the world's wealthiest person in just a few weeks, if that long. But, realistically, you can be one of the most successful investors or traders out there just by taking bold positions & being right just slightly more often than you are wrong. But, even doing that is very, very difficult. So, at least in that case, the difference between failure and doing well enough (even very well) is much less than the difference between doing very well & being optimal.
For many good stock investments, the entire gains from years of holding the position come from just the best few days of trading. So, even in holding one good position, one frequently must put up with long periods of settling for mediocre results even when the total return for the entire period would be considered excellent.

There very rarely is a "best" to go for. Even when there is, the best is rarely the same for different groups. A taxpayer would probably prefer lower taxes and think that best; the politician who spends the money on pork to buy votes thinks that is best. All you can do is look for what is **better** for whoever is making the decision (or what he thinks is better at the time he makes the decision).

I like Rubin's advice "The Big Fact" in his Overcoming Indecisiveness; "In very few instances is one decision actually better than another." His claim is that your commitment to making a decision work is usually more important than the particulars of the decision. Out and out bad choices are usually ruled out early in the decision process. I think this is one reason many people have trouble visualizing the situation with Eliezer's paperclip maximizer, that is so obviously bad that it wouldn't make it far enough in a human's decision making process as to even be consciouslly considered.

This is similar to the cost of capital: http://en.wikipedia.org/wiki/Cost_of_capital

A company will not make an investment/purchase if the expected return (%age) is less than the cost of capital. The cost of capital is determined by the rate of return that the money could get elsewhere WRT the risk.

In terms of a person doing something, the expected return is the easier part to calculate.

The risk is the hardest part to calculate by far.

It seems to me that when people start to aspire to do the most useful thing rather than just a useful thing with their limited resources without having serious hang-ups that prevent them from doing things that are too weird, they fairly reliably end up on my team sooner rather than later. I'm sure that organizational design techniques and experience motivating radical autonomy and independence from existing institutions will prove valuable in the sorts of IA efforts that FAI work and especially safe MNT development will need.

Welcome Aboard Patri!

What I mean by "the most useful thing" is, by definition, the most productive - that which will advance your goals the most for the least effort. Sure, costs in determining that matter. But it's almost tautological that that's what is best to work on.

But it's not generally possible to determine what will actually advance your goals.

Imagine for a moment Queen Victoria had established an overarching goal, "produce a means of communication that will permit Us to send Our Voice to all corners of the empire", and given it to her scientists and engineers.

How much progress would they have made?

Now, consider the equations produced by J.C. Maxwell. There were no indications ahead of time that his interest in electromagneticism would produce such a powerful result - a result that was instrumental in the eventual production of the radio, a device which fits Victoria's requirements to the tee.

You can't determine what the benefits of funding basic research will be. Yet doing so was the best way to reach the hypothetical goal, not funding research into the goal directly, which almost certainly would have produced a greatly inferior method within the bounds of then-known principles.

Aiming at the target is often the worst way to hit it.

I'm surprised the claim needs defending.
You're not much of a rationalist, then. All claims need defending. Some can be defended with trivial effort, some require a great deal more - but ALL require it.

The comments to this entry are closed.

Less Wrong (sister site)

May 2009

Sun Mon Tue Wed Thu Fri Sat
          1 2
3 4 5 6 7 8 9
10 11 12 13 14 15 16
17 18 19 20 21 22 23
24 25 26 27 28 29 30