In every class I teach I try to sometime remember to tell my students that I am presenting a lie. A favorable teaching evaluation might say ``He makes it all really clear'' and that is what I try to do (often far from successfully [smiley]. ) But ``making it really clear'' can sometimes be a disservice to students, because they go away thinking ``I could never think like that'' -- because I have not shared with them the reality of the hours and somedays days or years spent working and reworking material to the point that it can be presented on the board in ten minutes and made to seem obvious.

There's an anecdote that when John von Neumann first came to the Institue for Advanced Study in Princeton in the forties (I guess) he hired a maid. And someone asked the maid what it was like working for the famous mathematician. Her answer was, ``He's a fine person. Just a little strange. All day he sits at his desk and scribbles, scribbles, scribbles. Then, at the end of the day... he scrunches up all those pieces of paper he's been scribbling on all day and throws them all away!''

That's the reality of doing science. In our graduate courses we teach the myth -- that science is neat, organized, rigorous. Yes, the final product should have those properties. But that's just the tip of the iceberg, the end result of a lot of activity that's very unneat, disorganized, and not rigorous. First off, there's the very crucial and subjective decision as to what questions to investigate and which ones not to. In my opinion, it is almost always the decision at this level that makes the difference between great science and mediocre science.

And the one rule that is paramount in deciding what questions to investigate is... We investigate the questions which we know how to and do not investigate the ones we don't know how to. This is a trivial observation, but I think it's important for scientists not to start believing their own bullshit and thinking that the topics on which research is done are the ones which have the most intrinsic significance.

Whenever we get into these arguments about NLP it seems impossible to break through this attitude that only the scientific bottom line matters, that only the tip of the scientific iceberg is important. Over and over again, when I try to explain that something is interesting the same response comes back: ``Your opinions and your personal experiences are not evidence.'' Shit, man, I know they're not evidence. I'm not claiming that I'm presenting a neat, rigorous, and organized piece of science worthy of being published in a refereed journal. I'm trying to direct your attention to an area where there is something interesting, an opportunity for research.

Every once in a while, the thought occurs to me that almost all academic research (except one's own, of course!) exists primarily as a form of busy work to keep scientists occupied.

I used to think that this failure of vision, this inclination to continually pursue minor refinements of existing lines of research instead of looking for significant new areas to investigate, and the inability to see beyond the scientific bottom line... was a result of the pressures of the academic world. And to a large extent I still believe that.

In principle, the ivory tower idea sounds great: Free scientists from the pressure of having to produce commercially useful results so that they can work on questions which are really important. The only thing is, the pressure to produce publishable papers proves to be even more insidious. The primary consideration in choosing what to investigate becomes having a fair assurance that one will be able to get publishable results fairly quickly.

But now it occurs to me that another factor in this tendency to play it safe scientifically may be the lie that we present when we teach science. We are so busy being neat, rigorous, and organized in our teaching that we never have time to tell students about all the wild speculation, and all the mistakes and blind alleys and the bullheaded stubborness that real scientific progress has always come out of.

What is so amazing to me in these discussions in sci.psychology is the lack of curiosity on that part of scientists. The idea of looking into a subject for the purpose of learning something, of increasing one's understanding and insight without necessarily getting a publication out of it... seems to have become hopelessly old fashioned. It seems that back at the time when the scientists we now venerate were wild and crazy kids, sometimes doing incredibly risky things and outraging their scientific elders by their disrespect for established methodology... it seems that if one told one of those scientists about something very remarkable he might have said ``That seems very strange to me. I don't think that should happen. I think I'll try that out myself.''

Whereas now, at least in sci.psychology, when one tries to tell people about something remarkable psychologists fall all over each other trying to explain why they don't need to have any interest in it.

There's an anecdote about the Indian number theorist Ramanujan. While he was in the hospital he was visited by his friend and colleague G H Hardy, who remarked that the cab he had come in had the number 1729, which Hardy though as an exceedingly uninteresting number. And, as Hardy tells it, Ramanujan immediately replied ``Not at all, my dear Hardy. 1729 is an extremely interesting number, being the smallest number which is the sum of two cubes in two different ways.'' (1729 is 10 cubed plus 9 cubed and is also 12 cubed plus 1 cubed.)