But it's worth noting that on this reading the reason "water is H20" is necessary. i.e., you can't imagine a scenario where it seems true to say this is false, is *because* our language practice is messy and isn't purely built up from observational terms via definitions that always succeed so it can end up just meaning nothing.

Thus, the right response to your example about with the D20 communities is that, if they really have our practice with water talk, they should respond "ohh shit…guess we have to rethink things" and yes they end up reusing the characters/sound "water" to refer to the stuff in the rivers and streams but it's fair to say it's no longer the same notion because you couldn't go back to their textbooks written while they thought water was defined to be H2O and plausibly understand them to be speaking falsely about what they now use it to mean (D2O)

Or maybe I'm just reading my view into Kripke. It's been awhile since I read him so that's totally plausible.

How about this. What Kripke is trying to suggest with the necessary part here is that *if* our talk about water makes any sense at all then water is H20. I'm suggesting it's like a stipulative definition in mathematics, let x be the unique solution to f(x) = 0, where we can say that it's necessarily true that f(x) = 0 even though if we later discover that f(x) has no solutions or multiple ones then we learn that our talk about x simply doesn't make any sense at all.

Similarly, I'd suggest that what's going on in the water is H20 example is that we wanted to stipulatively redefine water to be H20 to render our talk more specific in certain contexts but that stipulative definition only makes our water talk make sense if (since we aren't willing to totally change our practice of water talk) it doesn't make too much of our water talk false.

So it can be necessary in the sense that if our (error prone) speech patterns are close enough to something to be meaningful at all then water is H20 but has a postererio content: namely that the conditions under which that stipulation leaves us meaningfully talking about anything is a contingent a posteriori truth.

So maybe the brief version would be: I'm suggesting Kripke is packing everything into the idea that we can find out our language practice is so far from anything coherent we were just talking nonsense so *conditional on some redefinition succeeding* you get a necessary truth but it's a posteriori that we aren't talking nonsense.

But yah I don't think the way he described it was very illuminating nor like the way it seems kinda forced into a very particular metaphysical framework.