Rational Reconstruction

Definition

Lakatos's meta-criterion for choosing between rival methodologies of science: apply each methodology to the actual history of science and see how much of it comes out rational (internal history) versus how much has to be explained away by non-rational causes — psychology, sociology, politics (external history). "Philosophy of science without history of science is empty; history of science without philosophy of science is blind." The best methodology is the one whose internal history is largest — the one that recovers the most of great science as reason rather than accident.

In the Book

The book's provocation is structural: Lakatos writes history of science as "rational reconstruction" in the main text, with the actual deviations from that rational script relegated to footnotes — footnotes that note exactly where real scientists departed from what the methodology says they should have done. Critics called this "history as caricature"; Lakatos's answer was that the caricature is the point — it prices, in footnotes, precisely how much of the historical record the methodology can claim as reason, leaving the remainder for sociologists to buy. He runs this test against rival methodologies (inductivism, conventionalism, naive falsificationism) and argues each fails it worse than the methodology of research programmes does, since each is forced to declare too much of the best science — Newton, Copernicus, Einstein at their finest moments — irrational.

Why It Matters

This is a self-application test for any framework that claims to judge good decisions: does the framework, applied backward, endorse the decisions everyone already agrees were the organization's or field's best calls? If a prized decision-making methodology would have forbidden the three best calls a team ever made, the methodology is what's failing the test, not the calls. It also legitimizes footnote discipline as an evaluation practice — explicitly recording where reality deviated from a framework's official script, then treating the size of that residue as the framework's honest score, rather than quietly discarding the deviations as noise.