I'd buy that if it sent out the exact same code that it would without that UA. But it doesn't.

What don't you buy? They test with a certain set of supported platforms. If a client isn't one of those (whitelisted) platforms, then they may send out a bunch of inefficient polyfills so they know XYZ functionality exists, even if it's slow.

Except that it works just fine if those 'unsupported' platforms are treated in the exact same way as the supported ones. To make it seem as if a whole pile of work was done in order to make unsupported platforms work less well but this is a good thing is something that I find very hard to reconcile. And if this were a company that did not have a history of exactly such tricks it might be more believable, MS - with me at least - does definitely not have the benefit of the doubt in cases like these.

No, a whole bunch of work was (hypothetically) _not_ done to verify if they were supported or not. I'm not suggesting that it's a good thing, I'm suggesting that laziness, sloppiness, or a simple oversight are just as plausible as malice.

It is _less_ effort to say "We support these specific platforms, and rather than doing actual feature detection, we'll just load a bunch of polyfills to make it work on anything else".

How thoroughly do you figure that guy posting the question tested it compared to whatever QA process MS does for supported platforms?

Good enough to detect a problem.

lol.