Tuesday, December 28, 2010

Finding Holes In Psych Experiments Is So Fucking Easy

So in a previous post today I explained a psych experiment Eliezer Yudkowsky uses to try to push his narrative that human beings are hopelessly flawed creatures that must bow down and kowtow to him because he is a Nietzschean super-man.

In that post I point out how Yudkowsky's interpretation of the data is pure self-serving crap. But it's worse than that. You see, the experiment is complete and utter crap. I came to that realization when I considered using hyperreals as a possible model of human morality. Hyperreals are more complicated than transfinite numbers and it's not clear that they would work to model human morality. It's also not clear they wouldn't work either.

Now in the bullshit experiment Yudkowsky likes, they think they're comparing spending money on saving 20 birds against spending money on saving 2000 birds. And if you're a Utilitarian that makes sense because all other goods and services such as television, computers, and other equally morally relevant luxuries ... are all fungible. If you're a utilitarian, all these things are comparable.

Of course, utilitarianism is crap and those things aren't comparable and the existence of those other things in the back of people's minds makes the experiment completely invalid. In the experiment, saving 20 birds isn't compared to saving 2000 birds, rather BOTH are compared to other luxuries, and since they're not really comparable, the predictable answer is that they end up having roughly the same value.

Now, in order to make it a VALID psych experiment to measure people's moral values what you have to do is set it up so there are no other competing moral concerns. And also no other competing psychological concerns such as that other people will pick up the slack. I'm going to come back to the second point in my next post about meta-rationality. Dealing with the first point can be done trivially by switching from saving birds to saving human beings.

And just to keep the experiment valid and negate the limits of people's imaginations, the 2000 people the subject hypothetically saves should be an actual real village. Then we can measure how much effort people put into saving little Timmy from drowning versus saving Springfield from a hydroelectric dam break.

My intuition says people will put about 2000 times as much effort to save 2000 people as just one person. If that's true then hyperreals are a perfect model of human morality as it actually works. And the experiment Yudkowsky pontificates about is just fraudulent crap. And of course hyperreals don't rescue utilitarianism.

This is just one psych experiment, but poking holes in it was so fucking easy. Honestly, I've yet to run into a credible psych experiment, one where I can't poke holes in it within 20 minutes. Except for the ones that use monkeys, those are great and usually reliable. Unless the lazy-ass psych mofos take "shortcuts" like triaging their monkeys by intelligence. Lesson learned: monkey experiments are great! And human experiments suck because human minds are far too complex to be understood by some arrogant pea-brained psychologist.

The only human psych experiments I actually like are Asch's conformity experiments. I especially like the extra effort they made to find out what's required to shatter conformity (answer: minimal). Because of course, finding out that you all are mindless group-thinking morons isn't exactly a discovery. Which highlights the fact that I think the only reason I "can't" poke holes in Asch's experiment is because I don't want to. That's why I'm not counting it.

I will start respecting psychology when its practitioners are able to perform a single experiment with a surprising, unexpected or even undesirable result that I can't poke holes in. I'm not going to give myself a time limit because sometimes (eg, 6 degrees of seperation) all of the data in the experiment is fraudulent, and uncovering that fact takes vastly more time than poking holes in the interpretation and setup.

What makes this all so exasperating is that figuring out the human mind is so easy. For me anyways. I only spent two decades on it, off and on, in between my other competing interests. What the fuck are psychologists doing?

On a completely different note, it's worth observing that Eliezer Yudkowsky 1) claims to understand human minds, 2) bases himself on a lot of fraudulent experiments, 3) doesn't understand his own mind. Meet the Great and Wonderful OZ!

No comments: