Turning out the lights on conditional "might"

This week's Style Guide entry at Johnson, the Economist's new and prolific (multi-author) language blog, is on the distinctions between may and might. Naturally, I was happy to see that some of the civilized world still shares my belief -- rooted not in rules but in a lifetime of using my language -- that you don't write "If he caught the ball, we may have won," when, dammit, he didn't catch the ball. Or as Johnson puts it:

Conditional sentences stating something contrary to fact ... need might: If pigs had wings, birds might raise their eyebrows.

Cheered though I was by such conviction, however, I'm more certain than ever that this use of might is circling the drain. When I blogged about it (at the Globe's Boston.com website) last December, I noted that the New York Times officially supports my position, but the Cambridge Grammar of the English Language thinks it's time to let it go.

And now comes The New Yorker, pounding another nail into the may/might coffin. In the Aug. 2 issue, Atul Gawande's article* on end-of-life treatments has this:

Almost nothing we’d done to Sara -- none of our chemotherapy and scans and tests and radiation -- had likely achieved anything except to make her worse. She may well have lived longer without any of it.

 Might well have, you mean. Oh, well.

*In the printed version of the piece, there was also a more obvious problem: A girl's name was given as both Ashlee and Ashley within the space of a few paragraphs. The web version has fixed that, but left "may well have lived longer" as is -- which must mean that even New Yorker readers don't mind it.