= 27d567bc2d48d9f40e9ccc20ea370b15)
SaidAchmiz comments on Rationality Quotes June 2013 - Less Wrong
You are viewing a comment permalink. View the original post to see all comments and the full post content.
= 783df68a0f980790206b9ea87794c5b6)
Loading…= b715d02e85f7b3b4ae65ca3d3e450868)
Subscribe to RSS Feed
Powered by Reddit
= f037147d6e6c911a85753b9abdedda8d)
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (778)
Of course, "bad things", and even more so "iterated bad things", have to be viewed relative to expectations, and at the proper level of abstraction. Explanation:
Right level of abstraction
"I punched myself in the face six times in a row, and each time, it hurt. But this is not mere bad luck! I conclude that I am bad at self-face-punching! I must work on my technique, such that I may be able to punch myself in the face without ill effect." This is the wrong conclusion. The right conclusion is "abstain from self-face-punching".
Substitute any of the following for "punching self in face":
Right expectations
"I've tried five brands of water, and none of them tasted like chocolate candy! My water-brand-selection algorithm must be flawed. I will have to be even more careful about picking only the fanciest brands of water." Again this is the wrong conclusion. The right conclusion is "This water is just fine and there was nothing wrong with my choice of brand. I simply shouldn't have such ridiculous expectations."
Substitute any of the following for "brands of water" / "taste like chocolate candy":
Ah, I've been in that job. My favorite in the stupid-expectations department was a customer who expected us to lie about the cause of a failure on the work order, so that his insurance company would cover the repair. When we refused, he made his own edits to his copy of the work order....and a few days later brought the machine back (I forget why) and handed us the edited order.
We photocopied it (without telling him) and filed it with our own copy. That was entertaining when the insurance company called.
This can be easily generalized as an algorithm.
Scientific method applied to everiday life, if you want :)
The thing is, some of the steps are very vague. If you have a bad case of insufficient clue, what's the cure?
I'm not sure I understood what you mean, but I guess you're thinking about cases where you can't have a "perfect experimental setup" to collect information. Well, in this case one should do the best with the information one has (though information can also be collected from other external sources of course). Sometimes there's simply not enough information to identify with sufficient certainty the best course of action, so you have to go with your best guess (after a risk/reward evaluation, if you want).
Sorry, I wasn't very clear.
I meant that if you have a deep misunderstanding of what's going on, as here, what do you do about it?
Well, it's somewhat hidden in steps 2 and 3. You have to be able to correctly state your hypothesis and to indentify all the possible variables. Consider chocolate water: your hipothesis is "There exist some brands of water that tastes like chocolate candy". Let's say for whatever reson you start with a prior probability p for this hypothesis. You then try some brands, find that none tastes like chocolate candy, and should therefore apply bayes and emerge with a lower posterior. What's much more effective, though, is evaluating the evidence you already have that induced you to believe the original hypothesis. What made you think that water could taste like chocolate? A friend told you? Did it appear in the news? In the more concrete cases: