Lakatosův výzkumný program

epistemic

Lakatosův výzkumný program

Definition

Lakatosův výzkumný program (Lakatosian research programme, MSRP) is the methodology of science developed by Imre Lakatos in The Methodology of Scientific Research Programmes (1978) as a refinement of Popperian falsifikace. Lakatos accepted the Quine-Duhem point that single hypotheses are not tested in isolation, but rejected the relativist conclusion. Instead he reconstructed scientific practice as the development of research programmes with three structural elements:

  1. Hard core — a small set of fundamental claims treated as not directly subject to refutation; the identity of the programme.
  2. Protective belt — auxiliary hypotheses that absorb empirical conflict and can be revised without abandoning the programme.
  3. Heuristic — positive (paths of theory development) and negative (which moves are forbidden) rules that constrain belt revision.

The decisive innovation is the progressive vs. degenerative distinction:

  • A progressive problem-shift introduces auxiliary hypotheses that predict novel facts not previously known — empirical content increases with the rescue.
  • A degenerative problem-shift introduces auxiliary hypotheses solely to neutralize counter-evidence, with no novel predictions — empirical content decreases (or stays constant) with the rescue.

The same protective move ("we add the assumption that X") can be either, depending on whether it is content-increasing or content-decreasing. Lakatos's criterion is therefore diachronic (judged over time, by track record) rather than synchronic (judged at the moment).

Within the Kompendium epistemické a kognitivní anihilace, MSRP is the canonical tool for distinguishing legitimate epistemic conservation from EA-1.3.1 immunization. EA-1.3.1 patterns are exactly degenerative shifts: every counter-evidence triggers a content-empty belt-revision that preserves the hard core but generates no new predictions. Legitimate science also revises auxiliaries — but the revisions are progressive: they specify new observable consequences that were not implied before.

Context

Within the Prismatic Platform, Lakatos's distinction is operationalized as a content-delta metric on belief revisions:

  • The Decision Engine tracks every revision to a belief's auxiliary structure and computes the content delta — does the revision generate new testable predictions, or only neutralize the prior counter-evidence? Revisions with negative or zero content-delta are flagged as candidate degenerative shifts and weight the belief's calibration score downward.
  • NABLA Infinity maintains per-belief auxiliary-hypothesis ledgers with timestamps and prediction registrations, so the progressive/degenerative judgment can be made post-hoc against the actual record rather than against narrative reconstruction.
  • The Color Teams Red Team explicitly seeds platform reasoning with degenerative-shift patterns to verify that the content-delta gating intercepts them before they harden into kognitivní anti-rekurzivita.
  • The complement is the honest hard-core declaration: the platform requires explicit identification of which beliefs are core (not subject to single-evidence revision) versus belt (revisable). Concealed hard cores — beliefs that behave as hard core while being presented as evidence-driven — are a documented EA pattern.

The MSRP framework also clarifies why naïve falsificationism fails as a defense: any practitioner can rescue a theory by belt-revision, and the question is not whether a rescue is permitted but what kind of rescue. The Decision Engine therefore does not flag rescue per se — it flags rescues whose track record is degenerative. Czech-language treatment: /kompendium/04-mechanismy/lakatos/.

See Also


Connect & Contribute

Created by Tomáš Korcak (korczis) | Open Source under GHL