I am somewhat sensitive to the doubt that philosophers have our intuitions corrupted by our training and so are not to be trusted, so I sometimes try to check what the inuitions of the more philosophically innocent are like. I never much liked Putnam's water/H20 case, so I tried that out on an innocent victim. Apparently, according to a detailed survey of a large sample size (1) which I assume is representative, XYZ is not water. Heavy water is water. Water ions are not water.
I introduced heavy water and ions into the equation because I alway found the view that water is H20 because we always intended to mean stuff having the same microstructure to be dubious in light of the fact that not all water has the same microstructure. Varying the number of neutrons doesn't matter. I had previously thought that varying the number of electrons didn't matter either (I thought it was generally held that water ions are water), but based on my exhaustive survey, both electrons and protons matter. But neutrons don't. So it's a matter of sufficiently similar microstructure, not a matter of the same microstructure.
That leads nicely into my theory that we don't identify natural kinds in the world, in the way that all these new realists and reference theorists like to insist; instead, we identify similarity clusters, which we like to have as big as possible. The more different similarities we can bundle together, the happier we are with a concept, and we tend to assume that once we've got a bunch of them together, there are probably more we don't know about, and we get unhappy when there turn out to be unexpected lacks of similarity in new areas (as with jadite and nephrite). But there's just a continuum of more and less extensive bundles of more and less tight similarities, not any sharp distinction between natural kinds and anything else.
Oh, my extensive surveys have also concluded not only that Gettier was wrong, but so was Plato. True belief is apparently enough for knowledge. No justification needed, and obviously accidents are just fine. Lots of philosophical problems are solved by that one.