1/

means that companies are often the end users, so they aren't "selling" the software, so they can still benefit from it being out there. And then there's "public domain," which I guess is more broadly applied than software, but kind of similar in effect (for me) to MIT license.

And then there's licenses that charge money, and this can be the case even if the "source code" (the means to reconstruct the software's effect) is available. There are lots of views and some laws, but these are the basics as I interact with them.

All this is to say that it's hard to move in an effective way, from a permissive license to something with restrictions (because adding restrictions retroactively would be a weird unstable clause if it was in the first license).

One additional thing, and relevant here, is that this code exists in places (repos/package managers/registries), and if someone can at least temporarily pull their code there, then it can break other people's code until a backup is found and links are restored to code copies.

Oh wait, unless you're asking the normative question of "could you, if you don't mind unpublish your code and break whatever palantir uses?" and that seems like a fantastic idea and a thing we should all do if we can.

Rather than licensing, I wonder if it would make more sense for open-source software to detect at runtime whether it's running at Palantir and break in subtle ways. Sure, Palantir could hack around it, but maintaining a local fork and integrating with upstream is annoying