What Does Make a Difference? It’s Really Simple

post by diegocaleiro · 2010-06-15T02:33:06.249Z · LW · GW · Legacy · 14 comments

Contents

14 comments

This is really simple:
Suppose you want to check if some action of yours makes a difference.
How to do it?
The wrong thing to do:
Think of the consequences of your action and evaluate them to see if they fit your purposes. If they do, go on and do it.

The reason for this being wrong: If someone else does something with the same consequences, and if your doing or not your action makes no difference to the fact that THAT person will do it, then you are not necessary for those consequences, they would happen anyway.
This is also true if something, not a person, would do an action with the same consequences.

The right thing to do: Consider what would happen if you DIDN’T do your action. Subtract that from what would happen if you DID do your action.
This is the difference it would make if you did it.

There is a reason it is called a ‘difference’, it is the difference between you doing it and you not doing it.

Example: Suppose you think you will make a difference by carefully considering your vote, and voting.

Wrong: Well, I’m partially causally responsible for the election of X so my action would make a difference.

Right: If I do vote or if I don’t vote, the same candidates will be elected. Therefore my vote makes no difference.
(In more than 16000 elections in the USA it was NEVER the case that one vote would have made a difference)

The AWFUL argument people usually say: But what if everyone did it?

The reason it does not work: Everyone will NOT do it. Yes. That simple.

The reason it is awful: Compare “I don’t think I should go to the movies today, what if everione did it?”

So, when you are willing to make a difference, not feel good, not do what everyone does, not clean your consciousness. When you want to REALLY, REALLY make a difference, you should consider the difference between doing and not doing it.

It is that simple.

 

(Note to the Less Wrong entry: I know most people here know that politics is a mind killer, but delving deeper into the simple argument, notice it works for any case of overdetermination, such as posting a comment or entry that someone else would. If you realize now how important it is to figure out what is important, don't forget more than one person have already done it, use them.)

14 comments

Comments sorted by top scores.

comment by Randaly · 2010-06-15T03:53:43.688Z · LW(p) · GW(p)

Suppose that you and somebody else shoot somebody at the exact same time. Because they would have died anyways, your killing them hasn't caused any harm, and so is morally OK by your theory. However, the other killer is also morally blameless under your theory, because if they hadn't killed the other person, you would have.

I believe that this requires a modification.

comment by Psychohistorian · 2010-06-15T04:13:48.116Z · LW(p) · GW(p)

This entire argument is definition driven. If you define "making a difference" in the way stated, the author is correct. If you don't, he's not. As cleverly pointed out elsewhere, the two-person shooting indicates that this method of ethical evaluation is completely at odds with almost any actual person's ethical beliefs. For a more real world example, consider the murder of Julius Caesar, or really any conspiracy plot - it would have happened without you, but your involvement was still improper. It may not have "made a difference," but virtually every person's concept of morality will register it as immoral.

Indeed, that is exactly why this is a useless definitional dispute: knowing that something "made a difference" doesn't tell us anything about that thing other than it would not have happened had you not done it. You seem to be endorsing this as some kind of moral framework, which I think it overwhelmingly fails as. This is particularly true if you hold the ideal that any moral framework should be universally practicable, which this one is obviously not.

And a point of forum etiquette: use italics rather than CAPS. CAPS tend to read like I'M SO RIGHT I'M YELLING! and, at least in my experience, correlate inversely with having a good point.

comment by JoshuaZ · 2010-06-15T04:04:57.077Z · LW(p) · GW(p)

Downvoting for poor explanation, unnecessary use of allcaps, and strong assertions made with little evidence or reasoning to back them up.

Replies from: Rune
comment by Rune · 2010-06-15T04:26:44.323Z · LW(p) · GW(p)

Seconded. Terrible exposition; it trivializes something that is non-trivial. Also, it would be nice if the writer used paragraphs and did not use CAPS (unless really shouting).

comment by SilasBarta · 2010-06-15T02:54:07.925Z · LW(p) · GW(p)

I'm not sure if I should respond, but ...

You also have to take into account the fact that other people are largely similar to you and therefore, when you set the output of the algorithm that defines you, you are also setting the output of every other algorithm if and to the extent that it resembles the one that is you, so the effect is actually larger than your action-difference taken in isolation.

Well, it made much more sense when Eliezer Yudkowsky and Wei Dai said all that stuff about similar computational processes and setting their logical output...

Replies from: diegocaleiro
comment by diegocaleiro · 2010-06-15T03:28:57.524Z · LW(p) · GW(p)

For reasons similar to those I gave here: http://lesswrong.com/lw/nc/newcombs_problem_and_regret_of_rationality/1ss1 Or using Bayescraft as described here: http://philsci-archive.pitt.edu/archive/00003169/01/noregrets.pdf I would say this should be part of what you consider to be a consequence of your actions.

comment by GuySrinivasan · 2010-06-15T02:59:34.563Z · LW(p) · GW(p)

What is the probability your vote will make a difference?

About 1 in 60 million for the average American voter on the last presidential election day. Was the total difference between your favorite candidate and the other one, divided by 60 million, worth your time and effort? Quite possibly. At the very least it seems absurd to say it definitely wasn't worth your time.

Replies from: Oscar_Cunningham, diegocaleiro
comment by Oscar_Cunningham · 2010-06-15T16:04:57.848Z · LW(p) · GW(p)

Tim Gowers has a blog post about the expected value of voting:

http://gowers.wordpress.com/2010/05/03/is-the-british-voting-system-fair/

comment by diegocaleiro · 2010-06-15T03:19:09.472Z · LW(p) · GW(p)

If 1/6000000 seems like worth it to you, I would suggest spending much less time and effort buying lottery tickets and after winning (here in Brazil the odds are 1 to 52 million) invest your money in financing SIAI or whatever you think makes a difference....

Replies from: Randaly
comment by Randaly · 2010-06-15T03:35:35.085Z · LW(p) · GW(p)

You've also got to consider the size of the outcome. Winning a lottery gives you, what, a few hundred thousand to a few million dollars? The expected value of buying a lottery ticket is then around 2 cents- less than the price of a lottery ticket, and so buying is a terrible decision.

Elections have much larger impacts. The 2000 election is a classic example: it's been argued that Gore would not have invaded Iraq, and that the Iraq invasion has cost around a trillion dollars; in that case, your vote has an expected value to the US of around 16,000 USD. (More if you live in a swing state, more if you believe Gore would have been better in other ways. Less otherwise.)

Not that I'm saying that this is a closed question- in fact, my point is that it isn't.

Replies from: Unknowns, JoshuaZ, diegocaleiro
comment by Unknowns · 2010-06-15T05:39:43.627Z · LW(p) · GW(p)

Note that you also have to consider how certain it is that your preferred candidate really is better. And since the division of opinion is usually 50/50 or so, unless you are extremely overconfident, it is quite uncertain that your candidate is better. So you need do multiply the benefit not only by the probability that you will cause your candidate to win, but also by the probability that your candidate is better, and then you have to subtract the disutility in the case that your candidate turns out to be worse.

In other words, don't vote.

comment by JoshuaZ · 2010-06-15T03:42:41.487Z · LW(p) · GW(p)

Note that the probability in general that an electoral system using something like the electoral college increases the chance for everyone that they will be the deciding vote assuming that no voting area is extremely in the direction of one candidate. To see this with a small example, consider the toy example of 9 people with three states each of three people with an election for two candidates.. If the election is by popular vote then the only way your vote matters is if exactly four of the people vote for one candidate and exactly four vote for the other. But, if you use the states then your vote breaks a tie whenever the other two states disagree and your state has a 1-1 for the other two voters. That's a much larger set of circumstances. This applies in general although the math to show it becomes a bit uglier when one has many different states of different sizes.

comment by diegocaleiro · 2010-06-15T03:44:37.422Z · LW(p) · GW(p)

Agree with your point. But notice that if we consider the case in which you actually win, which would be GuySrinivasan's proposal, then those 4 millions you win in the lottery could be invested in things that will have much greater proportional impact, for you'd be investing in the curve's tail....

Just a reminder. The post is NOT about politics and voting. It is about overdetermination and decision theory.

Replies from: GuySrinivasan
comment by GuySrinivasan · 2010-06-15T03:52:20.362Z · LW(p) · GW(p)

A quote from the linked article:

A probability of 1 in 10 million is tiny but, as discussed by Edlin, Gelman, and Kaplan (2007), can provide a rational reason for voting; in this perspective, a vote is like a lottery ticket with a 1 in 10 million chance of winning, but the payoff is the chance to change national policy and improve (one hopes) the lives of hundreds of millions, compared to the alternative if the other candidate were to win.

(it was 1 in 10 million in New Mexico, Virginia, New Hampshire, and Colorado)

If you want to make a post about overdetermination, I'd say don't use the voting example, since here's one person at the very least who thinks the example's far from clear-cut. The movies thing is fine - the probability enough people go to ruin everyone's experience times the experience ruined is still tiny, not plausibly large.