From the Author

The first book ended with a single verb in the second person plural: check. A convenient position, I admit: hand out the homework and leave the classroom. Since then I have been checking myself, and the main result fits on one line: the technical preprint that book referred to has slimmed down from seventy-eight pages to sixty-six. In part this book is about where the twelve went.

Only in part, because my losses here are a special case. While I was cutting my own sections, it dawned on me that crossing out is the mechanism everything else rests on. The eye keeps in focus a patch the size of a fingernail at arm’s length and fills in the rest of the field from memory, and not once in your whole life have you noticed. A memory is rewritten at the very moment you turn to it, so that reading memory and corrupting it are a single operation. The environment leaves a quantum system a handful of stable states and takes everything else for itself. Evolution, whose rules were written down long ago, has been refusing to start up in the laboratory for the fourth decade running.

That is where the title came from, along with the substitution hidden in it. At seminars on managing your own life you are sold “conscious choice,” and sold it, one must admit, successfully. By the moment of your choice, selection has already run several times, and you were not invited to any of them. What remains conscious is the last step, and even that only on holidays.

As in the first book, almost everything here belongs to other people. In the sixties Yarbus fixed a suction cap to his subjects’ eyeballs and recorded where they looked, and on one and the same canvas he got several dissimilar paths, where the only thing that changed was the question put to the person before the showing. Attwell and Laughlin calculated what a single signal costs the cortex and discovered a budget under which only a small fraction of neurons can be active at once. Bekenstein bounded the amount of information that fits into a region of space, and the limit is set by the area of its boundary. Nader showed that a memory brought up from storage is consolidated anew and requires protein synthesis to do so. Friston described an organism that moves in order to be surprised as little as possible. Zurek explained why, of all states, only a few become classical. Cronin and Walker figured out how to measure with an instrument the number of steps needed to assemble a molecule, and got in return a scandal that goes on to this day. The papers are peer-reviewed, the references are in place; my part is to lay them out in a single row and see what comes of it.

I let in a little of my own, and only at the very end, and all three times these are stories of how something did not work out for me. First, a quantity I derived so beautifully that I had time to believe in it, and which diverged from the Planck data by a hundred sigma, and diverged in my own calculation at that: when your theory is buried by other people’s hands, you can still survive it, but by your own it is far more educational. Then an engine I built so that lying to myself would be technically inconvenient: the fingerprint of a prediction goes into a registry before the measurement begins. After fifty-two runs it dutifully reported that it was working better and better, while nothing new was being added to it. And finally a machine to which for several years in a row we fed even graph paper instead of an eye, and then wondered why the same part of it came out now the best in the system, now the worst.

There are four storeys of reliability in this book, one more than in the last: “verified” (other people’s papers in peer-reviewed journals), “hypothesis” (my model, which has not yet been demolished, but not crowned either), “personal experience” (I lived through it, I cannot prove it) and a newcomer, “verified and did not work.” The last one was started out of necessity, because “not yet proven” sounds almost like “ran the experiment and got a rap on the knuckles,” although the difference between them is roughly that between “dinner is getting cold” and “there will be no dinner,” and holding on to this confusion benefits exactly one person: the one who wrote the book.

There is still no answer at the end, and by the thirteenth chapter it turns out that the question itself costs more than I thought when I asked it.

A book in which the author spends three chapters out of thirteen telling you where he went wrong sells worse than a book in which he is right. I have checked that too.