Question on how step-counter type trackers work

Saw discussion of FitBit in another thread, about it being movement related so it got me to thinking. Does one have to be upright for a step counter to count steps? If not, how much movement would be needed to count as a step? Just wondering if there's any role for these devices in identifying limb movements during sleep.

The "good" ones work with three-axis accelerometers--(not a noisy/cheap pendulum/spring.)
So they can differentiate one position from another.
Different processes can count steps--walking, running, or stair climbing.
Some can calculate strokes swimming or bike pedaling.
Other calculations assess movement while we are sleeping.
Accuracy depends on the individual's physiology and the programming.
Remember these items are intended for sports--not medical use.

I don't know about the others but the FitBit doesn't need actual steps to tell you that you "stepped".
I logged over 7,000 steps one day mowing the yard....I use a riding lawn mower.
So I don't think it would be particularly useful measuring limb movement.....it lies.

I accidentally washed an Oregon Scientific pedometer.
It thought I walked a heckuva long ways.
It still works!
There was an article a while back, suggesting that Fitbit tends to consistently over-count.

Playing guitar really gets my step count up!
When you understand how they work they can be a useful guide to what's happening, but they are by no means a scientific measurement device.

Linking the devices to smartphones, (and the Web,) increases the temptation to cheat
by adjusting the programming to add steps, giving users of one brand a "leg up"
on those with other, more honest brands of counters.
I prefer a pedometer to be self-contained, as it is easier on battery life.
Bluetooth kills batteries.
As self-contained pedometers are quite scarce now, I have stopped feeding the ones I have,
and will not buy any with that "feature".