How do hall effect switches not oscillate at the crossover point?

[e100]

I had run out of Allegro A3144 hall effect switches so I tried to make a similar thing using two 3503 linear hall sensors feeding into the differential AB inputs of a Max485 chip in receive mode.
I waved a magnet in front of one sensor and assumed that the 200mV hysteresis would give a clean signal, but instead I see a series of spurious pulses before it finally flips state.

Do real hall effect switches have the same problem internally, but just capture the first pulse and stretch it out to meet the fully flipped state?
The datasheet for the A3144 doesn't really explain what it does internally, other than showing a Schmitt trigger block.