More mature startups may be able to simply abolish the long-term incentive that the 90 day window provides, but I suspect younger (<30 employee) startups need added turnover protection. Perhaps backloaded vesting would be a more comparable replacement? Transparent, predictable, fair, and not as harsh as cliffs. You keep what you vest, but 70% of it vests in years 3-4.
[1] (Although there's certainly unfair variability based on personal financial circumstances in that average.)
However, 2- or 3-year cliffs could make sense as an alternate way of promoting a long-term mentality.
If you want to require a minimum commitment from employees, just increase the cliff. For one thing, cliffs are well-understood.
My chief objection to minimum service periods (whereby you only have 90 days to exercise if you leave "early") is that they actively hurt disadvantaged groups. An upper middle class engineer often has resources or access to resources which will let them exercise options early. In contrast, a disadvantaged employee is unlikely to be able to come up with those funds—which is particularly problematic because they often are leaving to escape a hostile work environment.
At least cliffs apply equally to everyone, regardless of their resources.
Is this different from a cliff?
Greater portability could in theory lead to higher turnover even among happy employees. They might go on to found their own company sooner. They might see good financial sense in diversifying their options portfolio. Yet young companies need the team to stick together for a certain time. Especially very small startups at the YC stage -- turnover is very harmful.
Note: In Adam's example, nobody leaves the pre-IPO company in under 4 years of service.[2]
[1] https://github.com/holman/extended-exercise-windows
[2] "imagine a company takes 10 years to IPO. Employee A works at the company from years 0 to 4. Employee B works there from years 4 to 8. Employee C works there from years 8 to 10."