People have a habit of basing firm predictions on the exceptionally small sample size of generations that we've had thus far. The NA market is notoriously resistant to brand loyalty, for instance, seeming to hop from platform to platform be it Atari, Nintendo, sega, Sony, or Microsoft. The only consistent trend I've ever really noticed is that Nintendo tends to dominate handhelds, and I think much of that can be attributed straight to Pokémon.
Who could have predicted that Microsoft, with all the momentum they'd accrued from the previous gen, would absolutely face plant out of the gate this time around? It's just hard to predict these things as something as simple as a bad price point can derail a console.
You make an excellent point. We only have had a few generations so far. And yet people limit their data even further by only looking back 1-2 generations. On top of that, people often look at the home market only and ignore the handheld market. So while data is limited, it seems people making predictions often go the extra mile to ignore most of the information we actually do have.