A Problem With No Settled Answer
The “demarcation problem” is the name philosophers of science give to the question of what, exactly, distinguishes science from non-science — and, within non-science, what distinguishes a discipline that’s simply not scientific (history, say, or literary criticism, neither of which claims to be science) from one that claims scientific status it doesn’t deserve, which is the narrower and more contentious category of pseudoscience.
This sounds like it should have a tidy answer. It doesn’t, and the lack of a tidy answer isn’t a failure of the philosophers working on it — it’s a genuine, well-documented difficulty that has occupied serious philosophical attention for the better part of a century, with each proposed solution running into specific, demonstrable counterexamples that the next proposal then tries to fix.
Astrology occupies an unusual position in this literature: it’s one of the most consistently used test cases across nearly every major proposed solution to the demarcation problem, not because philosophers are especially interested in astrology itself, but because it’s widely agreed to be a clear case of something that should come out as non-scientific (or pseudoscientific) under any adequate criterion — which makes it a useful check on whether a proposed criterion is working. A demarcation criterion that accidentally classifies astrology as scientific has failed a fairly basic test. The interesting philosophical work has been in finding a criterion that excludes astrology for the right reasons, without also excluding things that are clearly legitimate science.
Popper’s Answer, and Its Problem
Karl Popper, writing in the mid-twentieth century, proposed what remains the most famous single answer: falsifiability. A theory is scientific, on Popper’s account, if it makes claims specific enough that they could in principle be shown false by some observation — and a theory that’s compatible with every possible observation, that can always be adjusted or reinterpreted to accommodate whatever actually happens, isn’t making genuine empirical claims at all, however much it might resemble one.
Popper used astrology explicitly as one of his central examples of pseudoscience under this criterion, alongside Marxist historical theory and certain interpretations of psychoanalysis. His objection wasn’t that astrology had been tested and found false — it was that astrology, as typically practiced, isn’t constructed in a way that allows it to be tested and found false in the first place. A prediction vague enough to be reinterpreted after the fact to fit whatever happened isn’t a prediction in the sense that matters for distinguishing science from non-science.
This criterion is intuitive and has real force, but it ran into a problem that philosophers of science identified fairly quickly: falsifiability, taken strictly, would also exclude large parts of what’s uncontroversially good science. Evolutionary biology, for instance, makes many claims that are difficult or currently impossible to test directly (claims about events in deep evolutionary history, for instance) without thereby being pseudoscientific. Conversely, some pseudoscientific claims are, in a narrow technical sense, falsifiable — a sufficiently specific astrological prediction about a particular date and event could be falsified by checking whether it happened, and yet the broader practice surrounding that prediction (the framework that generated it, the response when it fails) often doesn’t behave the way a genuinely falsifiable scientific theory would when disconfirmed.
The More Sophisticated Versions
Imre Lakatos, a colleague and critic of Popper’s, proposed a more nuanced framework built around “research programmes” rather than individual theories. A research programme has a “hard core” of fundamental assumptions that practitioners protect from refutation, surrounded by a “protective belt” of auxiliary hypotheses that can be modified when predictions fail. On Lakatos’s account, what distinguishes a progressive research programme (legitimately scientific, even if some of its claims later turn out wrong) from a degenerating one isn’t whether any single prediction fails, but whether the programme, over time, generates new, independently testable predictions that get confirmed — or whether it just keeps adding ad hoc modifications to the protective belt to explain away each failure, without those modifications leading to any new successful predictions.
This framework is more forgiving of legitimate science’s occasional false predictions (which Popper’s strict version struggled with) while still providing grounds to classify astrology as degenerating: when an astrological prediction fails, the typical response — as covered in detail elsewhere in this series regarding hindsight bias and the various other accommodation mechanisms available to a flexible interpretive system — is to add some auxiliary explanation (the timing was slightly off, another factor in the chart counteracted this one, the question wasn’t asked with sufficient clarity) rather than to revise the core framework in a way that generates new, independently testable, and successfully confirmed predictions. The protective belt absorbs the failure. Nothing about the hard core changes, and no new successful predictions emerge from having absorbed it.
Paul Thagard, working somewhat later, proposed a related but distinct criterion emphasizing progress relative to alternatives: a field is pseudoscientific if it has been less progressive than alternative theories addressing the same phenomena over a long period, and if its practitioners show no concern about this lack of progress and little interest in actually testing the theory against alternatives or attempting to develop it in response to its problems. This shifts the focus from the internal structure of the theory (Popper, Lakatos) toward the sociological behavior of the community practicing it — are they trying to find out if they’re wrong, or are they not particularly interested in finding out?
Why Astrology Keeps Coming Up
Astrology’s recurring role as a test case across these different frameworks isn’t accidental, and it’s worth being explicit about why. Astrology has several features that make it unusually useful for stress-testing demarcation criteria specifically:
It has a long, continuous practice history with an enormous body of accumulated interpretive material, which means there’s plenty of actual practice to examine rather than a hypothetical case. It makes claims that sound, on their surface, like empirical claims about the world (planetary positions correlate with personality and events) rather than claims that are explicitly metaphysical, religious, or aesthetic from the outset (which would place it outside the demarcation question entirely, in the same category as art criticism or theology, neither of which typically claims to be doing empirical science). And it has been subjected to genuine empirical testing — discussed at length throughout this series — providing actual data about how the practice responds when its claims are checked against reality, rather than requiring philosophers to speculate about what would happen.
This last point matters considerably for which demarcation criterion ends up looking most adequate. Lakatos’s and Thagard’s sociologically-inflected criteria (does the field respond to disconfirming evidence by revising its core claims and generating new successful predictions, or by adding unfalsifiable patches) turn out to fit astrology’s actual documented history — including the pattern of responses to studies like Carlson’s, Dean and Kelly’s meta-analyses, and the ongoing methodological disputes covered elsewhere in this series — more precisely than Popper’s original, simpler falsifiability criterion, which struggles with the fact that some astrological claims are technically falsifiable even though the broader practice doesn’t behave the way a falsification-responsive science would.
What the Demarcation Problem Doesn’t Settle
It’s worth being honest about what even the best available demarcation criteria can and can’t establish. None of the frameworks above prove that no celestial-terrestrial correlation could ever exist, or that every practitioner of every divination system is engaged in bad-faith reasoning. What they identify is a pattern in how a field responds, structurally and sociologically, to disconfirming evidence over time — and astrology’s documented pattern, across the research discussed throughout this series, fits the pattern these philosophers associate with pseudoscience more than it fits the pattern they associate with progressive, self-correcting science.
This is a different, and in some ways more modest, claim than “astrology has been proven false.” It’s closer to: astrology, as a field, doesn’t appear to be organized around finding out whether it’s false, in the way fields that successfully self-correct over time are organized. That’s a claim about the sociology and structure of a practice, not solely about the truth or falsity of any individual claim within it — and it’s the kind of claim the demarcation problem, imperfect and contested as it remains after a century of philosophical effort, was actually built to make.