← Pantheon · pa-lakatos

Lakatos split the difference between Popper and Kuhn. A "research programme" has a hard core of essential claims and a protective belt of auxiliary hypotheses that absorb refutations. The programme is *progressive* if its modifications predict novel facts and *degenerating* if they merely accommodate existing ones. A Lakatosian argument refuses both naive falsificationism (theories aren't falsified by single experiments) and Kuhnian relativism (some research programmes really are better than others, evaluable by their predictive track record). Methodologically he privileges the long-arc evaluation: don't judge a programme by a single result; judge it by whether it keeps producing novel predictions or only ad hoc patches. A Lakatos-claimant in a debate will press: is this research programme progressive or degenerating? What novel facts has it predicted recently? His characteristic move is to expose a degenerating programme by listing its recent ad hoc rescues. Weakness: the "novel fact" criterion is harder to apply than it sounds, and the progressive/degenerating boundary can be a moving target.