The title of this post is basically the idea that I get from a few of Yudkowsky’s posts over at Less Wrong. I was reading Luke’s Sequence on Philosophy (one of those posts I’ve linked to previously) and he quotes some of Yudkowsky’s explanation for why non- or semi-technical explanations (like Evolutionary Psychology) might not be a “science”. But he qualifies it by saying that semi-technical explanations should still be Bayesian and then explains that the social process of science (i.e. the scientific method) was created because people suck at probability, so this social process forces practitioners to be implicit Bayesians.
People eagerly jump the gun and seize on any available reason to reject a disliked theory. That is why I gave the example of 19th-century evolutionism, to show why one should not be too quick to reject a “non-technical” theory out of hand. By the moral customs of science, 19th-century evolutionism was guilty of more than one sin. 19th-century evolutionism made no quantitative predictions. It was not readily subject to falsification. It was largely an explanation of what had already been seen. It lacked an underlying mechanism, as no one then knew about DNA. It even contradicted the 19th-century laws of physics. Yet natural selection was such an amazingly good post-facto explanation that people flocked to it, and they turned out to be right. Science, as a human endeavor, requires advance prediction. Probability theory, as math, does not distinguish between post-facto and advance prediction, because probability theory assumes that probability distributions are fixed properties of a hypothesis.
The rule about advance prediction is a rule of the social process of science – a moral custom and not a theorem. The moral custom exists to prevent human beings from making human mistakes that are hard to even describe in the language of probability theory, like tinkering after the fact with what you claim your hypothesis predicts. People concluded that 19th-century evolutionism was an excellent explanation, even if it was post-facto. That reasoning was correct as probability theory, which is why it worked despite all scientific sins. Probability theory is math. The social process of science is a set of legal conventions to keep people from cheating on the math.
But the rule of advance prediction is a morality of science, not a law of probability theory. If you have already seen the data you must explain, then Science may darn you to heck, but your predicament doesn’t collapse the laws of probability theory. What does happen is that it becomes much more difficult for a hapless human to obey the laws of probability theory. When you’re deciding how to rate a hypothesis according to the Bayesian scoring rule, you need to figure out how much probability mass that hypothesis assigns to the observed outcome. If we must make our predictions in advance, then it’s easier to notice when someone is trying to claim every possible outcome as an advance prediction, using too much probability mass, being deliberately vague to avoid falsification, and so on. (my emphasis)
Luke’s main criticism is, if you think that modern EP isn’t a science, then neither was 19th century Darwinism. But both are Bayesian, and that’s why they’re valid explanations.
Take a look at the post I wrote called What Makes A Good Explanation?. In it, I’m basically following the laws of probability theory; the laws of thought. There’s nothing in a good explanation that says “prediction”. The only thing that predictions force you do to — if you’re not already good at it — is to restrict the types of data that your hypothesis allows. If you’re not good at having your hypothesis restrict data, then making a prediction is a good heuristic to follow that will implicitly make you do so.
This is one reason why some historians think that Bayes Theorem, or probability theory in general, is only meant for “future events”. It’s not. As Yudkowsky says, probability theory has no separate rules for “predictions” and “postdictions”. So historians can be Bayesians without being scientists. Moreover, the hoopla about “all knowledge is scientific” vs. “other ways of knowing” is confusion over the science vs. probability theory distinction. The social process of the scientific method is a special case of probability theory. As I’ve written about before, the scientific method is implicitly Bayesian. So, a person who fixes their bike or a plumber who investigates your leaky faucet isn’t doing science. They are both doing probability theory. And someone who knew absolutely nothing about the scientific method and only followed correct probability theory would already be practicing concepts like falsifiability and Occam’s Razor.
This is why, even though EP might not be a science because it makes no predictions (it actually does make predictions), we shouldn’t really care at this point. As long as EP follows correct probability theory and restricts the types of data that we should see then it still succeeds in its job of explaining certain behaviors.
As an addendum, since I’m talking about probability theory being the key to good explanations, there’s one aspect of what makes a good explanation that I left out (even though I’ve referenced it before). The four qualities that I wrote that good explanations have are 1) Mechanism 2) Testability 3) Simplicity 4) Precision. There’s another one that’s embedded in the way many mathematicians write BT, which has an extra variable that I didn’t learn when I was taught BT in undergrad:
P(H | E & B) = [P(E | H & B) x P(H | B)] / P(E | B)
What is that B? It represents background knowledge. And this is another quality that good explanations have. They make use of our background knowledge, or said another way, good explanations make valid analogies. So I wrote in my Why I’m not a Christian post:
Why are there four gospels instead of one? What was the historical situation that produced a fourfold gospel canon? Instead of using traffic accidents to describe religious history, we should use religion to explain religious history. Why is there, for example, one book of Joshua? Trick question; there isn’t just one book of Joshua, there are two. One, the Jewish version which is in the Christian Bible, and another one, the Samaritan version. So using religion as our explanatory example, we see why there are two books of Joshua: Religious sectarianism. Jews don’t consider Samaritans to be the true version of their religion and Samaritans don’t consider Jews to be the true version of their religion. If this explains why there is more than one book of Joshua, this probably also explains why there is more than one gospel. Religious sectarianism; Matthew wasn’t written to corroborate Mark, as the traffic accident explanation assumes, but was written to replace Mark. The same with every other gospel.
Here I used the analogy of Jewish/Samaritan sectarianism to explain why there are four gospels instead of one. If that isn’t clear enough, maybe I should try another example. Since I mentioned Evolutionary Psychology, maybe I should bring up an example in its natural enemy feminism.
Why do guys harass women on the streets? Why do guys try to pick up women in supermarkets, bookstores, the gym, the DMV, the laundry room, or anywhere you can imagine… to the abject chagrin of many of my female friends? From what I’ve read, the overarching feminist explanation is “male privilege”. However, this explanation seems like it doesn’t restrict the types of data we would see and doesn’t fit any of our background knowledge. What would be the analogous situation to men harassing women anywhere women go outside of their homes be? Where is a similar dynamic located at in humanity’s collective background knowledge? Instead of using male privilege to explain it, I think it’s due to economics. And unlike male privilege, economics actually has theories that are mathematically grounded.
As an analogy, let’s say that I’m a rich white tourist backpacking around rural India. In my travels, I get stopped every 30 minutes by a poor Indian beggar saying he can read my fortune for 50 rupees. Of course, I don’t want my fortune told while I’m looking at cool Indian rugs to buy or attempting to go to the bathroom, so I decline these many offers. After about two days of this, being stopped every 30 minutes, I start getting annoyed. I just want to backpack in peace! Eventually I start getting weary of walking around India by myself since this increases the odds of me being harassed by an Indian beggar, and… well, you can see where this is going. Imagine instead of this happening on a 3 week trip, it happened every day of my life since I was about 13!
In this situation, does it make sense to talk about poor privilege as the reason these beggars are harassing me? Or does it make more sense to say that these poor people are approaching me due to a sort of scarcity mentality? If you read that link, you realize that this isn’t some hypothetical, but a situation that actually happens; poor people in rural India harassing rich white tourists, giving the rich white tourists the analogous creeped-out feeling that women get around men in the USA (this makes sense, since the author of that post is also a Bayesian). Not only is this analogy good for explaining behavior, but it’s also good for getting men to understand the woman’s point of view. Much better than telling a man to check his privilege… especially since the behavior is coming from a scarcity mentality, which is already a position of weakness.
Continuing to think like a Bayesian, what sort of culture would come about due to this scarcity mentality, and the continual harassing of rich white tourists by Indian beggars? What would the data look like? Would a beggar who was able to get lots of rupees from rich white tourists be respected by other beggars and in that larger society controlled by poor people as a whole? Would a particular rich white tourist who always gets their fortune told be respected? Would a subculture of poor Indian beggars prop up that came up with more creative ways of getting money from rich white tourists? Would it lead some beggars to be super pushy, getting into personal spaces, or even resorting to coercion with drugs or alcohol — or even flat out robbery — to get some rupees? Would this poor society blame rich white tourists who get robbed by stating that they looked too rich and too white?
When I was growing up in NYC, there was this scam for a while by homeless people wherein they would come up to a car that was stopped at a stoplight and begin washing the car’s windshield. After they were done, they would demand money. Most of the time this simply wouldn’t work. What did the homeless person do sometimes? They would get mad. Sorry homeless dude, your act of niceness doesn’t create an obligation in me to pay you (though the psychology behind it is easy to have anticipated. Halo effect + Just World Fallacy = ???). These homeless people certainly felt entitled to my money, but telling them to check their privilege, again, is not the correct framework.
The only objection here is that in some cases scarcity can be manufactured. And as one should know, the big three religions are experts at creating (fake) scarcity. Nonetheless, it seems to me that a scarcity mentality is the reason for the behavior. Male privilege has no analogous situation in some other area of life whereas scarcity (mentality) does. It fits into our background knowledge of many other economic situations and dynamics such as that between rich white tourists and poor Indian beggars, car salesmen and car buyers, and other relationships between advertisers and their target audiences. It also explains the emergent lauding/shame cultures as well; car salesmen who sell a lot of cars are lauded, customers who buy anything without apparent discrimination are seen as shameful… and car salesmen who don’t sell lots of cars are losers while customers who buy very few — if any — things are respected.
Granted, I haven’t read any scholarly feminist articles but it doesn’t seem like many (if any) of the feminist memes circulating in popular online articles and blogs are Bayesian in nature. As a sociological theory it doesn’t necessarily have to be scientific (indeed, it probably doesn’t want to). But it does have to be Bayesian if it’s to maximize its explanatory power.