It's complex, I'm sure. I don't KNOW, but I'm imagining the ZWave radio, which is of course half duplex, getting inundated with receive data (every motion device sending LUX, temp, IR, humidity readings) every x mins. The transmit side gets to wait for a turn. I'm guessing (again) but I can see specific minutes where many devices have something to say.. and one controller vs N chatty devices yields a bottleneck. Now pile on ZWave antique devices using the series 300 chip (vs series 500 for Zwave+) and it seems like with 80 randomly purchased devices, bottleneck is approaching certain. 
I've read comments that say something along the lines of: "It only polls every N seconds, I don't see that being a problem." It doesn't matter what value N is, eventually N aligns with everything else using any other value (M) because N * M = collision.
I have my mental image of how difficult it is to cure and maybe I'm close in my guesses, maybe I'm far, but I didn't see anything in the spec that allows the Hub to manage the chattyness of 30/60/90 devices. But I'm hoping I'm completely wrong and Hubitat pulls the rabbit from the hat.
