I don't think Microsoft cares that much, but I'm sure they view it for what it is: illegal, unauthorized usage of their intellectual property. (I'll hold off on my personal views of 'intellectual property' and whatnot; I'm not a lawyer, and those opinions are hardly relevant to the reality here.)
Still, even if Microsoft doesn't care, Wine Project and ReactOS care. My employer might care. There are unfortunately legal implications that exist, and if people want to play by the books they have to at least do the basic due diligence.
If you genuinely just don't care, you can lie to them and claim you never saw the original code, and attempt to cloak any code theft to make it indistinguishable from clean-room reverse engineering. Nothing stops you from doing that. Hell, the trouble is that it really can't be proven for sure. Even holding the positions I do on forums wouldn't prove my own innocence in such a situation, though maybe it helps build a case.
What stops someone from viewing that repo over Tor, on a different git server, or at a public library, unsecured WiFi network, etc and being considered tainted? At what point is someone considered untainted?
This seems like a grey area that slows the pace of R&D in countries that believe code is copyrightable & patentable (the US and select Western countries) compared to say New Zealand & China, where these aren't an issue.
Where did you get that idea about New Zealand copyright? Code is copyrightable here, see e.g. [1] or [2].
And it's not about belief, but definition. Copyright and patents are human inventions defined by countries' laws (which is why different countries have different rules: different definitions). In NZ, as in many countries, "Computer software follows the rules for literary work" ([2] again).
It was more that we clarified what was and wasn't patentable, as I don't think software was exactly patentable before that either; its patentability and the ability to enforce said patents were ambiguous. Patents on software are not quite completely banned, either, as, from your link's linked source, "Processes will still be patentable if the computer program is merely a way of implementing a patentable process."
