Duke Nukem Forever would beg to differ. The risk with delaying is that you may not be prioritizing the important things. Because you just bought yourself more time, it's easier to do the part you enjoy, instead of the part that's needed.
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).
Unfortunately, at least my tastes and expectations in games have changed in the intervening 15 years, and it no longer suited them at all.
If only the game was as good as the 2001 E3 trailer.
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.
For a puzzle game, I would say that there's less risk for that, although I do recall other games copying the 4D mechanic since.
[0] https://www.vice.com/en_us/article/vbp9m4/what-happened-to-r...