8/09/2011

08-09-11 - Randomness and Fault

Recent comment ranting has made me think of something that frequently annoys me.

I get quite aggravated when people invite a certain negative outcome on themselves and then act like it's random or unpredictable or "shit happens" or "just roll with it" or whatever.

There are three separate but similar categories of this : 1. Risky Behavior, 2. Intentional Ignorance, and 3. Futility of Fighting the System.

1. Risky Behavior : these people act like because something is probabilistic, their behavior has no effect on the outcome.

A classic example is risky drivers; someone might be speeding, talking on the phone, not paying attention to the road. They have an accident, and act like "accidents happen, it's random". No, it's not. You chose your behavior, and your behavior increased the probability of an accident. You just (probabilistically) crashed your car on purpose. It was a willful intentional choice to be risky.

More benign cases happen all the time; maybe you have a friend over and they're clearing plates from the table and are carrying way too many at once. Of course they drop one and break it. You are supposed to act like "ha ha, no big deal, accidents happen". But it wasn't an accident. They just (probabilistically) threw your plate into the ground.

Now I don't actually mind if someone comes over and breaks my plate, no big deal it's a fucking plate (it's a whole 'nother rant about how stupid it is to buy expensive plates and get upset when they break), but don't act like it was random, sure there was an element of chance, but it was your actions that (probabilistically) caused it.

It's particularly annoying when the person who has the "accident" told me to "chill out, it'll be fine" or whatever when I warned them to be aware of the risk.

Of course this happens in coding all the time too. I tend to be very cautious in my coding; I'd rather spend time testing and asserting now then have problems later. Inevitably I get into situations where someone on the team is having a nasty hard to reproduce bug. They act like "bugs happen" and it's sort of a random act of god. Did you robustly assert your code? Do you have unit tests? Did you separate out classes that have strict invariants? No? Then you just (probabilistically) chose to have bugs in your code, don't act like they're random.

(there's a separate issue of whether the precautions are actually worth it or not; there's a spectrum of behavior from having to be super careful in advance so that you never have problems in the future (eg. NASA) vs. just being sloppy and fast and accepting a high probability of risk (eg. Game Jam)).

Just because something has a probabilistic element doesn't mean there's no correlation to your actions, or that you're not to blame when things go bad.

2. Intentional Ignorance : this is chosing not to do the research that you easily could have done and thus getting into a bad situation. Now, there's nothing wrong with that per se, that's a life choice and has different trade offs. The thing that annoys me is when people act like they "couldn't have known" or it's perfectly normal not to have known. Not true, you could have easily known.

Say you're visiting a strange town and you go out to eat somewhere and it sucks. It's not random that it sucked - it's because you didn't do any research (probabilistically). Okay, that's fine if that's the choice you want to make, but don't act like it's not your fault - it is a direct result of your choice to not do research that it sucked.

3. Futility of Fighting the System : this is perhaps the most naive and self-defeating variant, and mainly affects the young or the poor (except when it comes to voting, in which case it surpisingly runs across all demographics).

These people act like it doesn't matter what they do, that someone their bank or cell phone carrier or the cops or whatever will find a way to screw them. Basically they refuse to recognize the cause/effect connection between their own actions and the outcomes.

A lot of this is because of the same failure to connect cause/effect in probabilistic situations. Maybe this person tried to be really careful one month and do everything right, and they still got some absurd bank fee or roaming charge or whatever, they conclude that "you can't win" and "what I do doesn't matter". They don't see that their actions might reduce the probability of fuckage even if it doesn't eliminate it.

(of course to some extent this is just an excuse; they really know the truth, but they pretend not to because they don't want to be accountable for their own actions, they want to be able to fuck up and act like they're not to blame, that "it doesn't matter what I do, the system fucks me anyway).

Amazingly even smart people will talk this way about voting, that it "doesn't matter who I vote for the politicians always fuck us" ; well yes, there will be fuckage no matter what, but don't be retarded, of course you can affect the probability of fuckage through your actions. Just because it's not deterministic doesn't mean you are divorced from responsibility.

A lack of determinstic feedback is of course what makes poker so hard for many people. Almost everyone learns well when there is immediate determinstic feedback on whether their action is right or not. (this isn't saying much, dogs and monekys also learn well under those conditions). Many people struggle when the feedback is randomized or unclear or very delayed. For example when you try a new line in poker, like maybe you try three-betting from the blinds with medium range hands, if it goes badly a few times most people will conclude "that was a bad idea" and won't try it any more. It's very hard for these people to learn and get better because they're just looking at what they did in the instant and whether it paid off.

2 comments:

brian said...

I don't vote because I think the vanishingly small odds of my vote causing a change in the outcome of almost any election except very local ones isn't worth the time I'd have to put into learning about the candidates and positions to feel informed enough to actually be pushing the system in a direction I believed in.

Voting is interesting to me because it's kind of nonlinear because of the binary nature of any individual vote's effect (either it determines the election or is irrelevant - I guess if you include the possibility of ties it's trinary, but still.)

It means that, for example, if I really wanted to influence the outcome of elections (the only pragmatic reason to vote I could find compelling), I'd actually be better off spending my time doing the equivalent of PR and marketing campaigns to influence large groups of people who agree with me to vote. Bang for buck on that, assuming I'm a good salesman / marketer / PR guy, is probably pretty high, though the investment is high too. It's like buying half the tickets at a raffle - it's expensive, but if you actually want to win the thing you might as well.

In this case, if I'm swaying huge groups of voters to the polls and helping turn the tide of the election, it's STILL basically irrelevant whether I vote or not because the odds of it being the clincher are the same (vanishingly small.) At that point I guess I would go ahead and vote just because to convince thousands of other people to vote I'd probably have to be really informed so the investment required in then voting myself is small enough it's easier to justify, but in terms of expected value versus time investment I think it's still exceedingly low. Even for a mail-in ballot.

Aaron said...

Voting is a situation where 'magical thinking' is pretty useful. I've found it somewhat satisfying to 'believe' that whatever I do politically, millions of people who are just like me will do the same. So, if I'm inclined to donate a few bucks to Darcy Burner or something, I go ahead and do it (with the magical thinking back-filling it so that I believe that a lot of other people are doing the same thing). Same with voting.

Also, the idea that a large number of votes is important is not totally true. Many races, and indeed many of the most important are decided by razor-thin margins (thousands of votes). So don't think of your vote as one out of millions. Think of it as one of a thousand that decide a close race. That's an astonishing amount of power for such a tiny cost (esp in a mail-in ballot state, where it costs you essentially nothing time-wise to vote).

Being educated about candidates is a basic requirement for being an citizen (though of course most people don't do it because the system is not set up to make that easy for you). Typically your local weekly magazine (Stranger, etc) will give you enough information to be 90% correct without any effort at all (and the 10% you get wrong will be mostly hair-splitting). Or if you don't want to think at all, just go vote against every republican candidate you see and you'll be getting things 75% right.

The final bit is to imagine your vote as an anti-vote against someone particularly noxious.

You are right that the vote itself is a tiny fraction of the effect of convincing other people (though first you have to convince them to vote at all :)). So, do that. You don't have to be a PR person even. Donate a bit of money to the people you like. Brag about it to your friends. Your friends all vote like you anyway, so the key is making sure all your friends aren't doing the same thing as you and not voting, and bragging about making donations to try to get them to do the same.

old rants