No Safe Defense, Not Even Science

Eliezer Yudkowsky18 May 2008 5:19 UTC
108 points
71 commentsLW linkArchive

I don’t ask my friends about their childhoods—I lack social curiosity—and so I don’t know how much of a trend this really is:

Of the people I know who are reaching upward as rationalists, who volunteer information about their childhoods, there is a surprising tendency to hear things like: “My family joined a cult and I had to break out,” or “One of my parents was clinically insane and I had to learn to filter out reality from their madness.”

My own experience with growing up in an Orthodox Jewish family seems tame by comparison… but it accomplished the same outcome: It broke my core emotional trust in the sanity of the people around me.

Until this core emotional trust is broken, you don’t start growing as a rationalist. I have trouble putting into words why this is so. Maybe any unusual skills you acquire—anything that makes you unusually rational—requires you to zig when other people zag. Maybe that’s just too scary, if the world still seems like a sane place unto you.

Or maybe you don’t bother putting in the hard work to be extra bonus sane, if normality doesn’t scare the hell out of you.

I know that many aspiring rationalists seem to run into roadblocks around things like cryonics or many-worlds. Not that they don’t see the logic; they see the logic and wonder, “Can this really be true, when it seems so obvious now, and yet none of the people around me believe it?”

Yes. Welcome to the Earth where ethanol is made from corn and environmentalists oppose nuclear power. I’m sorry.

(See also: Cultish Countercultishness. If you end up in the frame of mind of nervously seeking reassurance, this is never a good thing—even if it’s because you’re about to believe something that sounds logical but could cause other people to look at you funny.)

People who’ve had their trust broken in the sanity of the people around them, seem to be able to evaluate strange ideas on their merits, without feeling nervous about their strangeness. The glue that binds them to their current place has dissolved, and they can walk in some direction, hopefully forward.

Lonely dissent, I called it. True dissent doesn’t feel like going to school wearing black; it feels like going to school wearing a clown suit.

That’s what it takes to be the lone voice who says, “If you really think you know who’s going to win the election, why aren’t you picking up the free money on the Intrade prediction market?” while all the people around you are thinking, “It is good to be an individual and form your own opinions, the shoe commercials told me so.”

Maybe in some other world, some alternate Everett branch with a saner human population, things would be different… but in this world, I’ve never seen anyone begin to grow as a rationalist until they make a deep emotional break with the wisdom of their pack.

Maybe in another world, things would be different. And maybe not. I’m not sure that human beings realistically can trust and think at the same time.

Once upon a time, there was something I trusted.

Eliezer18 trusted Science.

Eliezer18 dutifully acknowledged that the social process of science was flawed. Eliezer18 dutifully acknowledged that academia was slow, and misallocated resources, and played favorites, and mistreated its precious heretics.

That’s the convenient thing about acknowledging flaws in people who failed to live up to your ideal; you don’t have to question the ideal itself.

But who could possibly be foolish enough to question, “The experimental method shall decide which hypothesis wins”?

Part of what fooled Eliezer18 was a general problem he had, with an aversion to ideas that resembled things idiots had said. Eliezer18 had seen plenty of people questioning the ideals of Science Itself, and without exception they were all on the Dark Side. People who questioned the ideal of Science were invariably trying to sell you snake oil, or trying to safeguard their favorite form of stupidity from criticism, or trying to disguise their personal resignation as a Deeply Wise acceptance of futility.

If there’d been any other ideal that was a few centuries old, the young Eliezer would have looked at it and said, “I wonder if this is really right, and whether there’s a way to do better.” But not the ideal of Science. Science was the master idea, the idea that let you change ideas. You could question it, but you were meant to question it and then accept it, not actually say, “Wait! This is wrong!”

Thus, when once upon a time I came up with a stupid idea, I thought I was behaving virtuously if I made sure there was a Novel Prediction, and professed that I wished to test my idea experimentally. I thought I had done everything I was obliged to do.

So I thought I was safe—not safe from any particular external threat, but safe on some deeper level, like a child who trusts their parent and has obeyed all the parent’s rules.

I’d long since been broken of trust in the sanity of my family or my teachers at school. And the other children weren’t intelligent enough to compete with the conversations I could have with books. But I trusted the books, you see. I trusted that if I did what Richard Feynman told me to do, I would be safe. I never thought those words aloud, but it was how I felt.

When Eliezer23 realized exactly how stupid the stupid theory had been—and that Traditional Rationality had not saved him from it—and that Science would have been perfectly okay with his wasting ten years testing the stupid idea, so long as afterward he admitted it was wrong...

...well, I’m not going to say it was a huge emotional convulsion. I don’t really go in for that kind of drama. It simply became obvious that I’d been stupid.

That’s the trust I’m trying to break in you. You are not safe. Ever.

Not even Science can save you. The ideals of Science were born centuries ago, in a time when no one knew anything about probability theory or cognitive biases. Science demands too little of you, it blesses your good intentions too easily, it is not strict enough, it only makes those injunctions that an average scientist can follow, it accepts slowness as a fact of life.

So don’t think that if you only follow the rules of Science, that makes your reasoning defensible.

There is no known procedure you can follow that makes your reasoning defensible.

There is no known set of injunctions which you can satisfy, and know that you will not have been a fool.

There is no known morality-of-reasoning that you can do your best to obey, and know that you are thereby shielded from criticism.

No, not even if you turn to Bayescraft. It’s much harder to use and you’ll never be sure that you’re doing it right.

The discipline of Bayescraft is younger by far than the discipline of Science. You will find no textbooks, no elderly mentors, no histories written of success and failure, no hard-and-fast rules laid down. You will have to study cognitive biases, and probability theory, and evolutionary psychology, and social psychology, and other cognitive sciences, and Artificial Intelligence—and think through for yourself how to apply all this knowledge to the case of correcting yourself, since that isn’t yet in the textbooks.

You don’t know what your own mind is really doing. They find a new cognitive bias every week and you’re never sure if you’ve corrected for it, or overcorrected.

The formal math is impossible to apply. It doesn’t break down as easily as John Q. Unbeliever thinks, but you’re never really sure where the foundations come from. You don’t know why the universe is simple enough to understand, or why any prior works for it. You don’t know what your own priors are, let alone if they’re any good.

One of the problems with Science is that it’s too vague to really scare you. “Ideas should be tested by experiment.” How can you go wrong with that?

On the other hand, if you have some math of probability theory laid out in front of you, and worse, you know you can’t actually use it, then it becomes clear that you are trying to do something difficult, and that you might well be doing it wrong.

So you cannot trust.

And all this that I have said, will not be sufficient to break your trust. That won’t happen until you get into your first real disaster from following The Rules, not from breaking them.

Eliezer18 already had the notion that you were allowed to question Science. Why, of course the scientific method was not itself immune to questioning! For are we not all good rationalists? Are we not allowed to question everything?

It was the notion that you could actually in real life follow Science and fail miserably, that Eliezer18 didn’t really, emotionally believe was possible.

Oh, of course he said it was possible. Eliezer18 dutifully acknowledged the possibility of error, saying, “I could be wrong, but...”

But he didn’t think failure could happen in, you know, real life. You were supposed to look for flaws, not actually find them.

And this emotional difference is a terribly difficult thing to accomplish in words, and I fear there’s no way I can really warn you.

Your trust will not break, until you apply all that you have learned here and from other books, and take it as far as you can go, and find that this too fails you—that you have still been a fool, and no one warned you against it—that all the most important parts were left out of the guidance you received—that some of the most precious ideals you followed, steered you in the wrong direction—

—and if you still have something to protect, so that you must keep going, and cannot resign and wisely acknowledge the limitations of rationality—

—then you will be ready to start your journey as a rationalist. To take sole responsibility, to live without any trustworthy defenses, and to forge a higher Art than the one you were once taught.

No one begins to truly search for the Way until their parents have failed them, their gods are dead, and their tools have shattered in their hand.

Post Scriptum: On reviewing a draft of this essay, I discovered a fairly inexcusable flaw in reasoning, which actually affects one of the conclusions drawn. I am leaving it in. Just in case you thought that taking my advice made you safe; or that you were supposed to look for flaws, but not find any.

And of course, if you look too hard for a flaw, and find a flaw that is not a real flaw, and cling to it to reassure yourself of how critical you are, you will only be worse off than before...

It is living with uncertainty—knowing on a gut level that there are flaws, they are serious and you have not found them—that is the difficult thing.

Eliezer Yudkowsky18 May 2008 5:19 UTC
108 points
71 commentsLW linkArchive