[books] Gary Drescher - Good And Real

[dot_fennel]

A book of science-inspired philosophy that suddenly starts to sound insane two-thirds of the way in. I retraced the argument and found the whole last section depended on a point covered VERY briefly, to wit:

Suppose the universe is deterministic. Now define the property P as being true at a given moment if and only if, were that state of the world run forward 100 years, some person on Earth would then be wearing two neckties at once.

Next, suppose that I decide it would be nice for property P to have been true 99 years and 364 days ago. All I have to do is wait a day and then put on two neckties. Presto! But my actions can't have caused property P to be true in 1909; causation as we understand it only works forward in time. Therefore, says Drescher, there is such a thing as an "acausal means-end link". And from there it's off to the races with a translucent-box version of Newcomb's Problem, which aaaaaagh never mind, the point is that I don't buy it.

I've run into "property P" before, and at first I thought this usage was obvious garbage-- just playing with words to make a future turn of events be "in the past". But of course, in a deterministic universe, property P really is a property of May 22, 1909, even if we have no way to evaluate it until tomorrow.

Anyway, there's a lot of relevant philosophical literature I haven't read, so I might be barking up the wrong tree. But I was not moved to make Drescher my guide on the topic.

Comments

[cuthalion.livejournal.com]

Suppose the universe is deterministic

I mostly do! Though I don't believe this has any impact on the argument. Is property P any different in a non-deterministic (but non-bifurcating) universe?


But as for property P. I disagree that by putting on two neckties you are affecting 5/22/1909. Instead you are affecting property p(5/22/1909). That's an abstract concept that doesn't really exist on 5/22/1909, any more than it exists 650 feet over the geographical center of Newark. And making it true doesn't really do anything different than every other action or non action you do. There's a really big infinity full of tuples {time, interval, condition} and while evaluating them may or may not be interesting (mostly not, I suspect), their occasional "truth" is not really more significant just because you assigned a letter to it. For instance, today, I imbued Alpha Centauri with property Q ("Is a mass of incandescent gas, at exactly the same time that Jesse is eating a burrito") -- at superrelativistic speeds!

[dot_fennel]

I think the significance of determinism is supposed to be that it shores up property P being a property of 5/22/1909, in the same way that "leads to a stable state" is a property that a pattern in Conway's Life can have even if nobody runs a simulation of it.

But the argument is slippery. I've actually had more fun shadow-boxing Drescher in my head than I did reading the last book of philosophy I found convincing. For example, I don't think the ridiculousness of property Q is a strike against Drescher, since our pre-theoretic idea of "doing X in order that Y" can handle people having idiotic goals just fine. I think the sleight of hand lies in claiming that eating a burrito bears the same relationship to it-being-the-case-that-you-did-eat-a-burrito as me jumping in the lake does to me getting wet.