Wednesday, March 09, 2011

Prediction

I have been interested in the social epistemology of so-called "fringe" science for a long a time. Back when I was doing my PhD, I even visited a parapsychology research institute in Germany, which was founded by the terms of a rich benefactor's will. He believed that we would one day develop the technology to speak to him beyond the grave. Quite reasonably (given this belief), he left all his money to research into the "boundary areas" of psychology. Not all the researchers I met there believed very earnestly in telekinesis, telepathy, or clairvoyance and such, but their research (in advanced statistics, say, or theoretical physics, or cognitive science) all in some way contributed to our understand of so-called "psi" phenomena, if sometimes a bit indirectly.

The institute also conducted actual experiments in parapsychology. One of my favorite setups (which I was sadly only told about) was intended to test for some innate capacity for precognition. Can we know what will happen in the future? The researchers assumed that any such capacity will resemble some more familiar capacity, and they quickly hit on memory as the best analogy. Precognizing would be a bit like remembering, except you would be "remembering" what will happen rather than what has happened. Now, this leads to two interesting consequences. First, like memory, we must expect precognition to be fallible. The fact that we often remember things differently, and often simply wrongly, does not mean that we don't have memory, it just means it isn't perfect. I'll get to the second consequence in a moment.

The researchers set up an experiment that tested both memory and precognitive abilities. In both cases, the subject would be shown a series of randomly generated numbers (between 1 and 9), one at a time, on a screen. In the memory task, they would be asked whether the number they are currently seeing is the same number they saw, say, five numbers back. If so they would press a button, if not, they would let it pass. They wouldn't get every number right, of course, but they would perform much better than chance because they would have actually seen the past number that they are comparing the current number with. Some subjects perform better than others, of course, but all demonstrably have memory.

In the precognition experiment they simply ask whether the number they are seeing now is the same number they will see five numbers further on. Here people obviously perform less well. (In fact, contrary to what some popularizes will tell you, the research at the institute I visited pretty clearly shows that we don't have parapsychological abilities. Sorry.) But here's an interesting thing about what they were also looking for (and also didn't find), and it's the second consequence of the analogy with memory. People who do the memory experiment will, not surprisingly, get better at it through practice. Would we not expect the same thing of a precognition experiment (if we had the ability)?

I thought all this was very clever. It really gets you thinking about how science works. But why am I telling this story here? Well, sometimes writers do actually balk at planning their writing on the grounds that they can't see into the future.

And sometimes people tell me that by making them plan their writing, and by forcing them to report on how well they stuck to their plan, I'm just making them feel bad about their lack of discipline. The plan is always so definite and clear and unambiguous and right, and then their actual week is a muddling through, ambiguous and wrong. They feel guilty because they are judging their performance according to the standard set by the plan.

The parapsychologists' attitude here might be instructive. Why, after all, do we imagine that it is our performance that was "wrong" and the plan that was "right". Why do we imagine that we are, from week to week, testing only our discipline? Could we not just as easily test, not your precognitive abilities perhaps, but at least your ability to predict what you will be doing next week. Try, sometimes, to let your actual performance be a judgment on the realism of your planning, rather than thinking of your plan as the basis of a judgment about the seriousness of your resolve.

Let me stress "sometimes". Obviously, we can't just make planning an attempt to predict performance. The plan must have some normative force. But a plan that is too rigorous or too ambitious is not going to be followed as precisely as a plan that respects the actual conditions you will be working under next week, and the state of your knowledge about what you are writing about. Imagine a memory experiment where the number is not between 1 and 9 but between 1 and 999, and imagine the question is whether the current number is the same one as the one you saw 20 numbers back. That would be much harder. But we would still imagine some improvement. In fact, we would imagine that if you start with the easy experiment, you would be able to increase the difficulty level over time. This, in fact, is more likely to improve your performance on the hard test than doing the hard test over and over again and largely failing.

When planning your writing, then, keep in mind that you are always learning about your ability to write. How much can you write, and how often? You are learning to predict next week's performance. But you are also improving your ability to write. As you write more and more often you also become able to write more and more often. So your plan can get more ambitious as time goes by without becoming unrealistic.

* * *

I sat down this morning at 6:00 AM knowing I was going write a post on this topic. By 6:46, I had written 920 words. I then had about ten minutes (after writing these words) to read it through before posting it. I can generally predict my blogging performance as follows: on Monday, Wednesday and Friday, I will start writing at 6:00 and post at 7:00. I will write between 500 and 1000 words. And I can even predict what I will write about. Sometimes I can predict content five days in advance, but in almost all cases I am right the night before about what I will write in the morning. I don't of course have any occult abilities. I just have a bit of discipline.

1 comment:

Jonathan said...

Right. If you get really good at predicting what you are able to do, then your planning will get much more realistic as well. All of a sudden you can uncannily know that you can finish an article in 5.2 hours. This is a tremendously powerful technique. I'm psychic and predict that I will end this comment with the word "end."