We have urged beginners to read much, but also to approach the past without preconceptions. There is a paradox here. What exactly is the proper use of what one knows? Somehow, those with knowledge ought to be better at interpreting the past than those without; the old hand ought to be better than the tyro. But how do those advantages come into play, and how are the disadvantages to be avoided? There is a parallel with the art of asking questions: we should not let our question obscure the better question that the data may teach us how to ask. Our previous knowledge, similarly, is the answers to questions we have already asked. The best strategy is not to let the answers we know squeeze out the answers we do not know.
The art of balancing previous knowledge so that it does not inhibit the acquisition of new knowledge; the skill of letting the evidence in hand tell its own story, against the story we may have heard, is what we call Tact. It is the most elusive of the intellectual virtues.
A related virtue, a virtue of limits, is called for, even with the reading one does to acquire knowledge. The record of previous investigations, which is typically preserved in journals, is the "literature" of a subject. The literature of an active subject can be enormous. In 1934, by Beveridge's estimate, 20,000 articles a week were being produced by the world's scientific periodicals. As for the current year, well, ask your science librarian to show you the latest issue of Chemical Abstracts.
In international Sinology too, the expanding "literature" has reached serious levels. Not long ago, it was possible to require that advanced students begin a problem by "reading the literature." This meant the back files of the journals of consequence, which were three in number. Then journals multiplied, and standards were raised. Once "the literature" had been redefined so as to include "Chinese scholarly writings," which at that time was the largest body of humanistic scholarship in the world, Sinology had reached the Chemical Abstracts level. For such fields, and most fields are like this, the fact is that even if you can get access to a significant proportion of that material, there is no time to read it all. And reading even a substantial fraction of it puts you at risk of being intellectually flattened out, with your skills of acceptance hyperdeveloped, and your powers of discovery shriveled by disuse.
One radical solution is to read nothing; to go into the subject with only your own instincts to guide you, unprejudiced by other people's old ideas. This drastic solution is noticed by Beveridge, not without approval. Here is his quote from Bessemer, the Bessemer of the steel process:
I had an immense advantage over many others dealing with the problem, inasmuch as I had no fixed ideas derived from long established practice to control and bias my mind, and did not suffer from the general belief that whatever is, is right. (p2)
Such are the benefits of keeping previous knowledge at bay.
And even if you are managing your knowledge well, there are moments in the development of a subject when it becomes obvious that the whole framework of knowledge is in the way, and something outside the circle of previous understanding is needed. Feynman describes one such moment this way:
The next day, at the meeting, when we were discussing the tau-theta puzzle, Oppenheimer said, "We need to hear some new, wilder ideas about this problem." So I got up and said, "I'm asking this question for Martin Block: What would be the consequences if the parity rule was wrong?"
Lee, of Lee and Yang, answered something complicated, and as usual I didn't understand it very well. At the end of the meeting, Block asked me what he said, and I said I didn't know, but as far as I could tell, it was still open . . . I didn't think it was likely, but I thought it was possible. Norm Ramsey asked me if I thought he should do an experiment looking for parity law violation, and I replied, "The best way to explain it is, I'll bet you only fifty to one that you don't find anything." He said, "That's good enough for me." But he never did the experiment. (p248)
Those who did do it got the Nobel Prize for it. End of story.
There are other things to do with previous knowledge than either using it or ignoring it. You can also use it to see, not what has been done, but what has been done badly. It's hard to judge your own translation, but somebody else's translation may be so awful that you know it has to be wrong. The sense of wrong helps you see where the right may lie. An argument may be so manifestly fallacious that you know it must conceal an error. An error is something that we know is not the answer to the problem. Knowing that there is a problem, and knowing one thing that is not the answer to it, and knowing the point at which a previous investigator has shied away from it, are terrific hints on the way to a better answer. Here is Feynman again, talking to a senior colleague about a major discovery he had just made:
I went to Professor Bacher and told him about our success, and he said, Yes, you come out and say that the neutron-proton coupling is V instead of T. Everybody used to think it was T. Where is the fundamental experiment that says it's a T? Why don't you look at the early experiments and find out what was wrong with them?
I went out and found the original article on the experiment that said the neutron-proton coupling is T, and I was SHOCKED by something. I remembered reading that article once before (back in the days when I read every article in the Physical Review - it was small enough). And I REMEMBERED, when I saw this article again, looking at that curve and thinking, "That doesn't prove ANYTHING."
You see, it depended on one or two points at the very edge of the range of data, and there's a principle that a point on the edge of the range of the data - the last point - isn't very good, . . and I had realized that the whole idea that neutron-proton coupling is T was based on the last point, which wasn't very good, and therefore it's not proved. I remember NOTICING that! And when I became interested in beta decay, directly, I read all those reports by the "beta decay experts" which said it's a T. I never looked at the original data; I only read those reports, like a dope. Had I been a GOOD physicist, when I thought of the original idea back at the Rochester Conference, I would have immediately looked up "How strong do we know it's a T?" - that would have been the sensible thing to do. I would have recognized right away that I had already NOTICED it wasn't satisfactorily proved. (p254-255)
Feynman redeemed his missed chance later, and he got it right the second time. Not everybody is going to be that lucky twice. It is better to pay attention the first time to the flaws you notice: the discoveries that have not been made yet.
Knowing how to challenge what you have read is knowing how to read. Knowing how to balance your reading with your observing, to let your reading suggest but to prevent it from preventing, to know when to put your reading aside and just look at the thing in front of your nose; this is what we mean by tact.
- W I B Beveridge. The Art of Scientific Investigation. 1950; 2ed Norton 1957
- Richard Feynman. Surely You're Joking, Mr Feynman. Norton 1985
17 Mar 2006 / Contact The Project / Exit to Outline Index Page