I look at them as a useful way to plan and estimate. It's okay if priorities shift or if you realise something will take longer, or that you actually need to work on something else right now and some KR will be pushed to next Q, etc. And it's perfectly fine not to meet some of your OKRs. I think of them as a compass rather than a paved road.
If it's ok to shift priorities 3 weeks into OKRs and push them to the next quarter why was it important to have them at all? Do you now need to measure the new work that's prioritized?
It's totally fine (and encouraged, in fact) to not pass every OKR, but I don't see any point to them. All of the failures of waterfall that everyone has always been aware of, but with the additional work of stating how you'll measure it, so that finance knows who to pay more.
If you have no idea where you're going, and no metrics at all, how do you reasonably pick any tasks?
So, clearly, you have some goal. It's likely also measurable. If you pick it so specific that you can't stick to it, yeah, OKRs are nonsensical. But "make money" or "increase conversion rate" or "find market fit" are clearly meaningful goals, no?
The difference from waterfall is that OKRs don't prescribe a "how". They describe a desired outcome. Guardrails within which you can be as agile as you want, just get some results.
As I said, look at them as a compass - a map will be a better analogy maybe - this is what I want to achieve at the point in time I'm making my plans, and here's how I'm going to achieve it. As long as that doesn't change, it's useful to break down a vague goal into milestones and evaluate on occasion (my team does it every 2 weeks) how you're progressing.
But it's okay for that to change, and then you just write new OKRs.
