Gettier, Instability, and the Structural Condition on Knowledge

Since Gettier, epistemologists have repeatedly attempted to repair the justifiedtrue-belief model by adding further conditions designed to exclude accidental truth.Yet the most striking feature of post-Gettier epistemology is not the ingenuity ofthese repairs but their failure to converge. This paper argues that the persistentnon-convergence of post-Gettier analyses is best explained by a mislocation of theexplanandum. Proposition-level theories evaluate completed beliefs and thereforeunderdescribe the structural fragility in the representational route connectingbelief to world. The paper proposes a structural reconstruction: knowledge isthe stabilization of representation under admissible variation. On this view,epistemic luck is representational instability. I first diagnose the repair cycle ofpost-Gettier epistemology and argue that proposition-first analysis systematicallyomits formation-path fragility. I then introduce admissible stability as a conditionon objective epistemic success and reconstruct epistemic luck as instability underadmissible transformation. Finally, I apply the framework to three canonicalcases—fake barn, stopped clock, and Smith–Jones—and argue that these cases areunified as distinct forms of representational instability. If correct, the lesson ofGettier is not that knowledge requires one more clause, but that epistemology hasbeen analyzing the wrong kind of object.