APPENDIX: RANDOM VERSUS UNSYSTEMATIC OBSERVATIONS

Sir Francis Bacon’s first rule of induction was that we must free ourselves of all preconceived notions, conscious or unconscious; and second, that we must observe, and record our observations quite unsystematically.

Bacon stressed that only by following both rules can we achieve a random collection of facts. And, he said, when that collection be large enough to be representative it will be ripe enough to enable us to squeeze theories out of it. His third rule, consequently, is that we must collect a lot before attempting to theorize.

The most obvious criticisms of Bacon - namely that we never have enough data - led thinkers of the more intellectualist methodologies, from William Whewell to Popper, to conclude against unsystematic observations as wasteful and pointless. It mattered to them little that unsystematic observations are not fully random. They did, in criticism of Bacon, declare it impossible to free ourselves of all preconceived notions. They sometimes even went further and concluded that hence even unsystematic observations are not random. Especially in our century they sometimes concluded further that these, therefore, merely reflect the preconceived notions of the observer (Russell, Popper).

Does all this make unsystematic observation valueless? Let us assume without debate, that observation is better systematic of some sort than unsystematic. Suppose, however, a case where systematic observation is precluded on technical grounds. What then? Can we declare that there is no future use in unsystematic observation? Though containing much random (and hence superfluous) information and much biased (and hence distorted) information, it may perhaps be of some future use and so it is certainly better than none, and so when the cost of gathering it is not too high it may be commendable.

Instances abound. Charles Darwin, by no means a Baconian, collected while on the Beagle as much information as he could - for an obvious technical reason; he did not hope to arrange a second visit.

No doubt, there are different functions to unsystematic observations. In their very limitation due to the bias of the observers, we may find in them the biases. A psychoanalyst encourages unsystematic observations, or free association, not on their scientific merit, but on the very contrary for their revelation of blind spots and biases. Similarly with anthro­pologists asking their informants to observe freely.

If this might be so, then bias which cannot be criticized and eliminated, should be given free rein. That is to say, even while making a concession to Bacon and admitting unsystematic observation, we should not go so far as to encourage random observation. Not only is random observation in principle - psychological, methodological, epistemological - impossible; the very attempt to make our observation more random than it naturally is may, indeed, back-fire. The natural unsystematic observation is already so much encumbered with irrelevancy, that the significant in it may - for merely technical reasons - never be retrieved for good use; when the attempt to make the unsystematic observation even more random takes over, the swamp is too much.

To take one interesting example, I quote in full one note, the final one, from Edward Sapir’s Language, Chapter III, on phonetics.

The conception of the ideal phonetic system, the phonetic pattern, of a language is not as well understood by linguistic students as it should be. In this respect the unschooled recorder of language, provided he has a good ear and a genuine instinct for language, is often at a great advantage as compared with the minute phonetician, who is apt to be swamped by his mass of observations. I have already employed my experience in teaching Indians to write their own language for its testing value in another connection.

Sapir’s proviso, namely that the recorder have a good ear, clearly indicates the risk that the unsystematic observations of the recorder may be of no interest whatsoever. Yet, clearly, for technical reasons, these are to be preferred to the ‘minute’ recording of all facts.

There is little doubt that the ‘minute’ observer is historically a des­cendent of Bacon even when, as usual, his historical and intellectual awareness is too poor for this fact to be conscious with him. There is little doubt that the very reason Bacon recommended minute observation, namely the attempt to achieve a random collection of facts, is what makes the venture useless. It is obvious that once induction was given up as a bad job the pendulum swang full swing towards the highly directed fully designed experiment - as a test of a theory. There was admittedly some indulgence towards the odd nature-lover who collected specimens but his status was deemed as no higher than that of a stamp-collector. And so a lacuna was created, in which various writers, including Darwin, Evans-Pritchard, and Sapir, can be found. These are not inductivists and they do not preach random observation. But they do see room for un­systematic, though not random, observations, for one reason or another. What I have ventured to claim is that the lacuna was created by the reasonable, yet too violent, response to Bacon. Since Bacon was in error, we can differentiate the unsystematic from the random observation, and reject fully only the latter. There is no risk in becoming Baconian by the sheer advocacy of unsystematic observations - for one reason or another, as the case may be.