I mostly agree. Though, you'd get some interesting reverse engineering related problems.
Ie look at the patent, come up with a problem that would make this patent obvious, lock some engineers in a sealed room with the problem, and wait for something like the patent to fall out. Was that independent invention, then?
Probably not, although you could write the rules in a way that avoids this I think -- e.g. the knowledge of the person who looks at the patent is imparted to the engineers because they're all part of the same organization.
You could probably work around this by e.g. introducing enough layers of separation between engineers and the person who read the patent/formulated the problem. It would be an interesting form of... knowledge laundering? ;).
Ie look at the patent, come up with a problem that would make this patent obvious, lock some engineers in a sealed room with the problem, and wait for something like the patent to fall out. Was that independent invention, then?