Prologue: Three Jews on the Internet

I had just finished my lunch at the college canteen. As I walked along the long tables to put away my tray, I joined the conversation of another small group of people. A fellow freshman—a guy with a prominent nose—was steaming with rage, his cheeks and forehead purple, his tone of voice confidently jolting like a sawtooth, because apparently some ignorant peasants, somewhere, believed it was possible to keep an artificial intelligence in a box.

I felt confused. Why was he so profoundly worried about this, uh, “problem”? So, I have a computer in a box… Uhm… Yes I have a lot of old computers in boxes, always stayed there, whatever. Maybe the cardboard will catch fire if the computer is kept on? No, wait, now I get what he’s talking about: the cooling system won’t work properly because air circulation is constrained in the box! He must be one of those who have a big custom machine with colored leds on the fans that leave a fancy afterimage and strange ports nobody ever uses and everything is overclocked and by shorting R435 with pin 99 and overfeeding (important) an additional bit gets written between each two other bits on the disk doubling the capacity.

Instead, he rage-explained to me that it was about a man named Mr. Yudkowsky whom people paid (?) so that he would convince them to press a fake button (??), under a confidentiality agreement (???).

I was even more confused, but this man was clearly on to something—he sounded so confident!—so I followed along. As we crossed the central bridge, he was wrath-expounding an experimental test of optimal agency:

The rules were SIMPLE, they would be given X dollars if they predicted the amount Y that the partner had predicted they would get given the expectation Z of a third party who would lose X – Y from an initial Q amount of dollars known to the first but not the second, and OF COURSE all they had to do was compute d/du P log(P), but these SIMPLETONS utterly FAILED because they would rather convince themselves that they really DIDN’T CARE about losing ten dollars because they were FREQUENTIST MORONS.

All the details in the quote above are wrong, because I was not getting anything of that. But I understood that there were these poor people who were handed a complicated puzzle I couldn’t even get in my skull in the first place, and by statal decree I was supposed to be good at math, and if every one of them got it right, they would probably gain 10 dollars, and of course they couldn’t really not want those 10 dollars so they were incredibly stupid to not be perfectly intelligent. For real? What contrived mind would conceive such a vexatious experiment? What’s wrong with this guy? Why am I even paying attention to what he says?

Five years later.

As I write these words, the frequentists are amassing at the gate. I don’t know how long we can resist. I’m reading translated foreign articles in the newspaper. I arrive at the general culture section, the one where I can be sure to find a “physicist” explaining that the “quantic” mysteries prove wrong the naive materialistic reductionists that pretend to determine what is true and false (as if!) when the possibilities of the mind are actually not bounded by such trivial preconceptions as logic, or a “philosopher” (quotes possibly redundant) highlighting how the weird syntax rules of the defunct language of three derelict guys in some remote country open up a new profound understanding of the universe that is forever forbidden to us due to our excessive reliance on technology which, let’s be honest, can’t really do anything for real, or an “intellectual” (quotes redundant) showing how teaching too much math makes people think in a rigid way such that they believe any religious bullshit that the current dictator wants them to believe.

The fool of the week is some “Pearl” guy who has discovered the secret of causality [sound of quivering flamelights]. Mankind went through an intellectual dark age where no one could discern cause from effect, but now, thanks to the genial heterodox work of Pearl, a causality revolution has sparked through the world. All paradoxes are solved. All doubts are cleared. By invoking the power of the do-operator, the most insurmountable of problems becomes a trivial exercise to the initiated. The article avoids ever mentioning what this do-operating consists of, it just states very clearly with a lot of words that you are supposed to do-operate stuff and you will gain final knowledge. I am so overwhelmed by the light of this revelation that I put away the newspaper, take a mental note to cancel my subscription (to avoid damaging my feeble mind with such a stream of profound truths), and never think about this again.

No! Wait! I’ve already seen this somewhere… Uh-uh-uh-uhm… Found, it was mentioned in that blog, Andrew Gelman’s blog. I like that guy! He’s friendly, once I used his gang’s statistical software because I couldn’t manage to do a complicated data fit for the weekly laboratory assignment, it took some time but it spat out the answer that I wanted to be true, so he’s a good guy. He also always gives useful advice, like how he always recommends being wary of implicit choices in your procedure that you will take to bend the result of the statistical analysis to give out precisely the answer you want, such as doing the fit in a different way until you think it works. Anyway, if this was featured on Gelman’s blog, it means it’s ok! It’s not something silly! It’s just the newspaper that—by chance—made Pearl’s work sound silly like the other silly (for real) stuff they usually write about.

After having determined as indisputable fact that Pearl was a reputable source of knowledge, “understanding anything by repeated application of a do-operator” sounded dignified and worth of investigation. In fact, it sounded much like “understanding anything by repeated application of Bayes’ theorem”, an advice originated from that weird man, Mr. Yudkowsky, that had been proven useful so many times. Indeed, I had been curious at times about what exactly that shady character had been pursuing with such knowledge; however, whenever I inquired about his writings, I was directed toward an incredibly long revised version of the Harry Potter novels, depicting a by magnitudes more wearisome protagonist than the original, and left it at that.

Now, Pearl appeared to have written various books on the topic, luckily in a very limited number of two. The titles were “Causality” and “The Book of Why.” After obtaining my pirated copies, I could see at once that the former was very long and full of “obvious proofs left to the reader,” while the latter was rather short and filled with funny drawings. My profound love of knowledge compelled me to opt for the most efficient means of acquiring new information, and thus I set forth on reading “The Book of Why.”

The book proved interesting in a lot of ways. It blended historical narration with a clear presentation of a mathematical theory of causation and many remarkable anecdotes. It would be a waste to spoil for the reader the pleasure of discovering herself all the gems of this work, so I will limit my recount to the three themes that most stuck in my mind.