Going off of my own simulator session and videos on YouTube, it is normally short bursts of one or two seconds on autopilot.
In manual flight it would have only been expected if it was STS. I don't know the answer for that system but it certainly doesn't keep trimming to the point where the stick force becomes very high. I can only assume it would be similar. Based on the report plots with flaps still deployed any automatic trim was short bursts which would be consistent with that.
Also, we've seen reference to the FAA expecting pilot response to a runaway to be 3 seconds. Therefore, it is a logical conclusion that over 3 seconds of continuous trim is not normal.
The combination of long bursts of trim combined with the great increase in stick force should have indicated that something was wrong with the trim system. Since it obviously wasn't a jammed stabilizer the only other logical conclusion is a runaway.
The logical follow up for this is though, that electrical trim is still functioning and not the cause of the run away. This only leaves the options that actually manual trim does run away when electrical trim is not on (so switching off electrical trim is really counter-intuitive in that situation) or the computer that controls electric trim (that is active parallel to the thumb switches) is malfunctioning and is trimming but at this point it was not 100% clear to the pilots that with AP off there is actually a computer aided system (besides STS) working in the background that could malfunction (or better function as intended, to press the nose down no matter if it kills everyone as long as AoA is too high). It was probably known that MCAS was a thing but not how it works, where it is implemented and how it is shown and felt if it goes rogue. That would have been an item for difference training in a sim but yeah Boeing thought better hide it.
The fact that electric trim is still functioning does not indicate that it is not the cause of the runaway. In fact, the way the NNC is written, it seems that it would be expected for electric trim to be able to counter a runaway in many situations. The manual wheel/manual trim system has priority so the fact that the electric trim worked would indicate that the manual (non-electric) part of the system wasn't the likely cause.
In the electric trim regime I see 5 possible failures that lead to run away trim:
1. The manual electric trim inputs have a short or are stuck or any other failure
2. The actuator fails and just trims till the max/min
3. The wiring fails/shortcuts
4. The computer fails
5. The manual non-electric trim fails
Now how does this impact the pilot, or how is it felt:
1. The trim is continuous and can not be stopped if smacking the control column does not release the switches. Nothing can be done except hit the cut out switches
2. Will be a continuous trim that will not stop if you flick the thumb switches
3. same as 2.
4. This is actually the tricky part because the manual electric trim does still work and if activated will kill the signal from the computer because it has priority over the computer
5. This would be a death sentence because no matter what both ways of trimming would be tarnished and there would be no real way to use electric trim if manual is prioritized over electrical
Now for Point 4:
If the computer fails, or in case of MCAS actually does what it supposed to do (trim with unlimited authority no matter what as long as AoA>X) then it does not present itself as a runaway because:
1. Electric trim does still work and
2. After you did trim with electric trim there is no immediate
restart of trim caused by the computer, there is a 5 second break. This actually means that the trim action is deliberate by the computer and not a malfunction.
The problem is that it was never stated that a false input from the AoA probe will lead to step-wise
nose down trim.
The problem could also be a broken feel computer. At the end it is, with no knowledge of the engineering and function of the software, almost impossible to conclude that a broken AoA sensor leads to the computer deliberately apply step-wise nose down input.
The real problem seems to be that a broken AoA probe leads to a cascade of problems in the 737 without ever telling the pilot that the probe is broken. The trim did not run away or had a failure, all the trim inputs from the computer were legit and working as intended. From the aircraft perspective only a full nose down input was the solution to a problem that actually was not there.