No one said 'roller coaster'. Part of quality software is saying 'no' to stuff, or having a coherent vision, not 'software by committee'.

Only if you have a quality engineer at the wheel though. Otherwise you just end up with horrible software.

I guess that’s why large enterprises hate it, since they’re all about reducing risk.

If they get middling software every time then that’s preferable to the two extremes.

Enterprise software isn't middling, it's usually only barely functional and falls just short of what it should do (i.e. it's shit). Everyone is just so pathologically used to it they think it can't be better. Probably due to the allergy to risk that you mentioned. Ironically risk is eliminated, but in doing so you've just ensured you have consistent problems that can never get better.