Here are a couple of quotes from Williamson’s “summary of his book”:http://www.blackwell-synergy.com/links/doi/10.1111/j.1468-0149.2004.0356a.x/full/ in the _Philosophical Books_ symposium.

bq. Since Gettier’s refutation of the analysis of knowledge as justified true belief, one attempted analysis of knowledge after another has succumbed to counterexamples.

And later in the article…

bq. Knowledge is the norm of belief: a flat-out belief is fully justified if and only if it constitutes knowledge.

Now these don’t particularly look consistent. If Gettier is right, then there are (relatively nearby) possible worlds in which someone has a justified true belief that is not knowledge. Which is to deny, among other things, that a belief is justified only if it constitutes knowledge. But Williamson holds that Gettier is right, and that a belief is justified only if it constitutes knowledge. What gives?

One plausible solution (that doesn’t ultimately work) is that Williamson is making an order of analysis point. If the second quote is correct, then _knows_ is necessarily coextensive with _justifiably truly believes_. But that doesn’t show that the JTB *analysis* of knowledge is correct, any more than necessary coextension shows that we can analyse _being the number two_ as _being a member of the number two’s singleton set_. (The 2, {2} example is one Kit Fine has used in a number of places and I’m borrowing here.) That is an interesting position, one I actually have a fair bit of sympathy for, but it can’t be what Williamson wants. Because he’s argued explicitly elsewhere that Gettier cases show that there are _counterexamples_ to the JTB analysis, not just that it gets the order of analysis wrong.

From the article it looks like the way to resolve the apparent contradiction is that Williamson thinks that the Gettier cases only work if we interpret ‘justified’ as ‘justified by the best version of internalist epistemology’. He doesn’t think that there are counterexamples to an _externalist_ version of JTB. I don’t think this is particularly plausible. The intuitions supporting Gettier cases don’t turn on whether we’re internalists or externalists about justification.