Research Programme

Definition

A research programme is a developing sequence of theories, not a single claim, held together by three parts: a hard core of commitments the community has decided not to question, a protective belt of adjustable auxiliary hypotheses, and heuristics that tell researchers what to avoid and what to pursue next. Lakatos's central move is to relocate the unit of scientific judgment from the isolated theory to this whole structure evaluated over time: "The basic unit of appraisal must be not an isolated theory or conjunction of theories but rather a 'research programme'."

In the Book

Chapter 1 sets up why the shift is needed: Popper's rule that one refutation kills a theory would have condemned Newtonian mechanics at birth — the Moon's motion misbehaved for decades and Mercury's perihelion never fit, yet the Principia became "the most successful research programme ever." Chapter 3 gives the structure in full dress using Newtonian mechanics as the showcase: hard core = the three laws of motion plus universal gravitation; protective belt = geometrical optics, planetary-shape assumptions, perturbation theory, atmospheric refraction; positive heuristic = the planned march from point-planets to spinning oblate spheroids to interplanetary perturbations. Anomalies hit the belt, never the core, and the heuristic tells each generation which idealization to relax next. Lakatos frames this explicitly as a rebuild of Kuhn's paradigm — the same tenacity and shared commitment Kuhn described sociologically, reconstructed here as a rational strategy with logical parts.

Why It Matters

Judging a single claim in isolation invites both false death sentences (killing something on its first anomaly) and false immortality (defending anything indefinitely with no standard at all). Treating the unit of judgment as a structured, evolving sequence lets you ask a sharper question than "is this true right now?" — namely, which part of the structure is under strain, and is the sequence as a whole still generating growth. The same move applies to any long-running commitment: a company strategy, a scientific field, a technology bet. Most "is X failing?" debates conflate a belt problem (fixable, expected) with a core problem (existential) because they never separate the structure in the first place.