In general I agree with you but each scenario is different. It highly depends on how often you release and gain insights from real end-users. The shorter this cycle, the smaller the Product Backlog should be.

I keep something between 2-3 Sprints. Yet, I countinuosly adapt. But imagine a team with a slow release process that takes a month to go to production, they would need at least another 2-4 weeks to gain insights. In such cases, a backlog would need to contain more work. Of course we could discuss the release time and shorten it, but it's one step at a time.