DNF was fine, it just wasn't the amazing masterpiece people expected. Miyamoto's quote can be true, but it can also be true that while delays can get you to 'good', that doesn't mean they can get you to 'great', and 'great' might sometimes only be achievable through aggressive cutting to the core experience. Not necessarily rushed, but not without deadlines and hard choices either.
For games pushing a new gimmick this seems even more true, and especially the idea of iterating on things that maybe aren't even quite "good" in order to learn more and eventually understand how to make something great even if a lot of the time greatness seems accidental. Two examples come to mind... Doom was made in a year, but was a culmination of even shorter projects successively refining what FPS meant. Portal was done in a little over 2 years, but was also built on the experience of Digipen students' project game that introduced the gimmick mechanic along with others at Valve with experience making FPSes (along with some writers who knew what they were doing).
Duke Nukem Forever was far from fine. They'd spent over a decade switching engines, rearranging progression and changing scenarios until they had no more money and nothing even close to a finished project.
Then Gearbox stepped in, wrapped it up in near passable form and shipped it. But the final product is very clearly not a good game. Level design is bad, graphics are terribly inconsistent, ranging from looking modern for its time to looking ten years dated. Gameplay is poorly balanced and the pacing is just not there.
I loved Duke Nukem 3D and was very much looking forward to Forever when it was announced. By the time it launched I had no expectations on it and bought it as a fun thing. It's amusing that they wrapped it up and published it, but it's not a finished game.
What it is, however, is fodder for some great conversation at work when we're discussing project management.
DNF had nothing to do with what made D3D actually good: amazing level design, creative weapons and enemies, fantastic music, tons of secrets, and a variety of weird environments. D3D took all the joys of shareware, episodic DOS gaming and brought them to 3D. DNF took all the juvenile bits and wrapped them in a half-baked CoD shell.
If only the game was as good as the 2001 E3 trailer.
For games pushing a new gimmick this seems even more true, and especially the idea of iterating on things that maybe aren't even quite "good" in order to learn more and eventually understand how to make something great even if a lot of the time greatness seems accidental. Two examples come to mind... Doom was made in a year, but was a culmination of even shorter projects successively refining what FPS meant. Portal was done in a little over 2 years, but was also built on the experience of Digipen students' project game that introduced the gimmick mechanic along with others at Valve with experience making FPSes (along with some writers who knew what they were doing).