Here's an example. He is reviewing Nozick's Philosophical Explanations, which he dubs 'the Great American Novel of philosophy' (that is not an unadulterated compliment), and is wondering whether any philosophical works have successfully been set out as deductions from a small number of axioms.
At the very beginning of Western philosophy, Parmenides' poem (or half of it) may have tried to do that, but it is hard to tell from its ruins - except that they seem more like the ruins of a temple than of a tower.That's a marvellously suggestive claim and, typically, nothing more is said to explain just what Williams means by the temple/tower contrast. I wouldn't let a student get away with writing that in an essay without scrawling a big question mark in the margin. But it's a thought that lingers and might just turn into another longer and better thought later. I like philosophical writing that does that. Plain-speaking and relentless clarity and explicitness is all well and good but it's a chore to read and it leaves you dry-mouthed and gasping for water.