The fundamental problem in the philosophy of science might be called the demarcation problem: What is it that distinguishes science from nonscience or pseudoscience? What, for example, makes evolutionary theory scientific and creationism pseudoscientific?

Philosophers of science have taken three broad approaches to this problem. One approach is to look for some criterion that demarcates science from pseudoscience--like Karl Popper's criterion of falsifiability, which says that a theory is scientific if it is open to experimental refutation. Let's call this approach methodological positivism.

A second approach is to argue that science is demarcated from pseudoscience not by its methodology but by a sociological criterion: the judgment of the "scientific community." This view, associated with figures like Thomas Kuhn, Michael Polanyi, and Robert K. Merton, might be called elitist authoritarianism.

Finally, one might deny the very possibility of demarcation, arguing that there is no rationale that privileges scientific over nonscientific beliefs. This view is often called epistemological anarchism.

The most impish of the epistemological anarchists was Paul Feyerabend (1924­1994), whose methodological motto was "anything goes." His good friend Imre Lakatos (1922­1974) took an opposing view--a blend, as he saw it, of Popper and Kuhn. Instead of asking whether a single theory was scientific or unscientific, Lakatos examined entire research programs, classifying them as "progressive" or "degenerating." He used the contrast to show how scientific consensus could be rational, not just a matter of mob psychology.

Feyerabend found this unpersuasive. "Neither Lakatos nor anybody else has shown that science is better than witchcraft and that science proceeds in a rational way," he wrote in "Theses on Anarchism." But Lakatos never gave up trying to convince his friend that his views were wrongheaded--and Feyerabend returned the favor. Their correspondence from 1968 to 1974 (when Lakatos died of a heart attack) has just been published as For and Against Method(Chicago), edited by Matteo Motterlini.

The letters have their share of ribaldry. "I am very tired because my liver is acting up which is a pity, for my desire to lay the broads here (and there are some fine specimens walking around on campus) is considerably reduced," Feyerabend wrote from Berkeley. The friendly affection between the two philosophical antagonists is much in evidence. Lakatos, writing from the London School of Economics, often signed his letters "Love, Imre."

Philosophically, however, there is no detectable convergence of their positions over the six years of correspondence. That is not surprising, really, given how vexed the demarcation issue is. Take a seemingly easy case: astrology. We all think astrology is a pseudoscience (pace Feyeraband), but it is not easy to say why. The usual arguments are (1) astrology grew out of a magical worldview; (2) the planets are too far away for there to be any physical mechanism for their alleged influence on human character and fate; and (3) people believe in astrology only out of a desire for comforting explanations. But the first argument is also true of chemistry, medicine, and cosmology. Nor is the second decisive, for there have been many scientific theories that have lacked physical foundations. When Isaac Newton proposed his theory of gravitation, for example, he could furnish no mechanism to account for how gravity's mysterious "action at a distance" was possible. As for the third argument, people often believe in good theories for illegitimate reasons.

Surely, though, astrology fails Popper's criterion of falsifiability? This seems a promising line of argument, since horoscopes yield only vague tendencies, not sharp predictions.

Yet such tendencies, if they exist, ought to show up as statistical correlations for large populations. Indeed, attempts have been made to detect such correlations--notably in the 1960s by Michel Gauquelin, who surveyed the times of birth and subsequent careers of twenty-five thousand Frenchmen. Gauquelin found no significant relationship between careers and zodiac signs, which are determined by the position of the sun at the time of birth. But he did turn up associations between certain occupations of people and the positions of certain planets at the time of their birth. For instance, in accordance with the predictions of astrology, individuals born when Mars was at its zenith were more likely to become athletes, and those born when Saturn was rising were more likely to become scientists.

But if the scientific status of astrology cannot be impugned on Popperian grounds, perhaps it can be on Lakatosian ones. Some years after Lakatos's death, the philosopher Paul R. Thagard made a detailed case for astrology being a "dramatically unprogressive" research program, and hence pseudoscientific. Astrology has not added to its explanatory power since the time of Ptolemy, Thagard pointed out. It is riddled with anomalies, which the community of astrologers shows scant interest in clearing up. And it has been overtaken by alternative theories of personality and behavior, like Freudian psychology and genetics. (Not that the latter two aren't also vulnerable to the charge of being pseudoscientific.)

Lakatos himself clearly thought astrology was pseudoscience--as was much else. "The social sciences are on par with astrology, it is no use beating around the bush," he wrote to Feyerabend. ("Funny that I should be teaching at the London School of Economics!" he added.) As for Feyerabend, the only definition of science he was finally prepared to tolerate was "what follows from a principle of general hedonism." And what about the truth? "The truth,whatever it is, be damned.What we need is laughter."

Jim Holt