To have a valid claim, Carmack must have had copied some code or data he produced during those hours. If he merely recreated the work from memory, Zenimax have nothing.
Unless recreating non-patented work from memory is actually forbidden? That would be horrible.
Yeah, forgot about that. Such Faustian clauses are evil, and shouldn't be possible in the first place.
Nevertheless, this doesn't look like a Non-Compete Agreement violation, or we would probably have seen those words in the article. My guess is, Carmack haven't signed one (75% probability).
I agree that Carmack probably didn't not sign one. But they are very common among companies in the tech sector and they can be just as horrible as you imagine.
Their claim will be that any of Carmack's work on those topics, because he originally began his work under Zenimax, is recreating based on Zenimax's property, because Zenimax has a claim to the early prototype Carmack thought of when pitching the ideas and there's no good way to show that Carmack isn't using the knowledge of those topics created during that period for his current work.
It's essentially writing new code looking at your old source code, and hence directly derivative of work that Zenimax has a reasonable claim to.
No no no, there's a difference from re-creating work from memory, and re-creating work while looking at an old copy.
It is also possible to work back from first principles. Heck, many Free Software was written like that: by rewriting the proprietary parts, possibly one by one.
Unless recreating non-patented work from memory is actually forbidden? That would be horrible.