Scientific inquiry proceeds according to certain strict social constraints which, for example, ensure that it is rare for the natural scientist to deliberately fake data. Such toeing the line is enforced by voluntary and widely accepted norms, violation of which would most likely result in traumatic and rapid loss of reputation and projects as trust is withdrawn by other members of the scientific community.

     So argues Gordon Tullock of public-choice-theory fame, co-author with James Buchanan of The Calculus of Consent, in a book first published in 1966 and recently reissued by Liberty Fund as part of a series of Tullock's works. The Organization of Inquiry is all about how scientific work gets done and promulgated in a socially coordinated but mostly not centrally-planned way. I have no review of the book, just a few questions and observations. Well, first an observation, then a couple questions. Well, not questions really so much as complaints. Maybe half and half.

     1. Tullock considers two basic motives of scientific inquiry. Per C.P. Snow, "one is to understand the natural world, the other is to control it." Or as Tullock puts it, we investigate "because we are curious, or because we hope to use the information obtained for some practical purpose." (Obviously a person can be motivated by both desires.)

     These motives produce, respectively, "pure" science and "applied" science, which have no reason to feel superior or standoffish toward one another given how interdependent they are. Sometimes while trying to make something work an engineer will achieve a synthesis of scientific insights overlooked by the "pure" scientists entrenched in traditional specialties. The "applied" insights can inform the "pure" research and perhaps even spawn whole new fields of inquiry. And of course many practical inventions routine abet "pure" inquiry: microscopes, telescopes, satellites, rockets, the giant donuts that spin subatomic particles around. Indeed, many "pure" scientists travel to work in motorized vehicles. I may be straying from Tullock's exact argument here.

     2. I'm not sure I understand his view of induction.

     Induction is the process of drawing generalizations from particular cases; as opposed to deduction, the process of relating particular cases to already known generalizations. If Joe is a man, we deduce that he is mortal from the previously known fact that all men are mortal. We need not investigate Joe's specific biology except insofar as needed to confirm that he is a man rather than a mannequin. We already know that any particular human being's biology is the same in the relevant respects as that of all other men. We don't know this because we got it from an index card, but because of our prior study of the biology and history of human beings. You can't do any deductions about the world unless you've done some induction first.

     Tullock, under the influence of Karl Popper, says that how we reach a hypothesis is less important than whether it survives predictive tests. He does not much discuss how hypotheses might be formulated, but presumably these could include methods quite futile and time-wasting, and not only with respect to one scientist's idiosyncratic effort. A bad hypothesis might well survive sundry inappropriate "tests" buying into the faulty assumptions of the hypothesis. On the other hand, there may be many truths we can learn about the world, arrived at by observation and thought, that cannot really be "tested" except according to a very generous understanding of the concept of testing. Are economics and history subject to the testing requirement? Do we really not know that all other things being equal, a rise in supply of a good on a market will lead to the same or lower price of the good, simply because we cannot control for other factors in real economies? When we read more history or discover more documents or artifacts from a particular era, do these constitute "tests" of our ideas of what happened?

     Certain "proper procedures" in reaching hypotheses would seem to be indispensable once we get past free association or brainstorming. For example, that the assumptions of a hypothesis must pertain to reality. To be sure, our current understanding of certain facts is often the very issue at stake. But what about, to pick an example at random, the hypothesis sometimes bandied about that causality is routinely violated at the quantum level of being, a judgment allegedly supported by experiment?

     Causality is an implication of the fact that things are what they are, that they cannot simultaneously be and not be at the same time and in the same respect. You can't investigate at all unless you take for granted that the subject of your investigation has an identity that can be investigated. By what identifiable tests could one show that an entity somehow acts in flat contradiction to the possibilities of action entailed by its very nature (i.e., its actual nature, as opposed our perhaps mistaken understanding of its nature)? And how much scientific thinking-time goes down the drain if such a "hypothesis" asserting the flatly impossible gains currency and is incorporated into many other hypotheses and theories?

     Yet certain experimental puzzles in combination with certain untenable philosophical assumptions have indeed historically led to such persistent "inferences" about the nature of very small things, the conclusion that somehow causality is being violated. In reason, once a train of experimentation and thought leads you to such a self-contradictory conclusion, you know you're on the wrong track without any further "testing" at all.

     3. The importance of acknowledging the self-evident in governing our thought leads us to Tullock's citation Jerrold J. Katz's "strict proof" in The Problem of Induction and Its Solution "of the impossibility of logically justifying 'inductive' reasoning," perhaps because induction must ultimately be grounded by what is self-evident, not what can be proven. Which would be a fair enough observation.

     But justification or validation is not the same as proof. The justification of inductive reasoning starts where all reasoning starts, with acceptance of the self-evident: a world with a specific nature, of which we are conscious. The world and our awareness of it are given to us as unavoidable starting points. Normally we act on what we self-evidently know without making a big fuss about its being self-evident. But carefully attending to the difference between that which is and is not self-evident helps us avoid such claims as that somehow the quanta "violate causality" simply because we macro-entities lack a perfect understanding of the conduct of micro-entities.

     What can be granted is that inductive generalizations, no matter how scrupulously formulated, cannot be reached from a standpoint of omniscience or infallibility. Any tenable theory of knowledge must take into account the context-dependent nature of knowledge-acquisition, the need to be open to relevant new evidence.

     Tullock also brings up the old bit about the possibility of a black swan. Why the sudden appearance of a black swan would not annihilate our knowledge of swans is left as an exercise for the reader.

     4. So much for the nature of scientific inquiry as such. Let me briefly turn now, finally, to the main part of The Organization of Inquiry, how inquiry gets promoted and organized socially. Despite many fruitful suggestions from Tullock I will touch on only one.

     Tullock says that investigators in the natural sciences are subject to stronger incentives to "get it right" than investigators in the social sciences, where "the possibility of practical applications is very limited, and curiosity is likely to be directed at non-scientific ends."

     If such was largely the case in 1966, I wonder if it is still as much the case decades later, given how bureaucracy and politics have affected scientific interpretations in areas like cancer, AIDS, global climate and all the other natural-science inquiry co-opted to service dubious political ends.

     Earlier in the volume, Tullock proposes that natural scientists are every bit as susceptible to prejudice and emotionalism as the rest of us once they leave their professional arena with its strict social controls and enter the realm of, say, political debate. Even though, when reporting lab results, they remain conscientiously honest and objective. But if it's still true that data cannot be outright faked without incurring high professional costs, what about minimal logical standards of interpreting the data? Can those be faked?

Copyright 2005 by David M. Brown. Brown is a freelance writer and editor. To view previous installments of this column, click here.