My understanding (again, uncited, like almost everything in these threads) is that one might /perceive/ a different result than this because the /software/ calibrates what 0 and 100% mean based on observation.

If you never observe a zero (or, presumably, a 100%), it might recalibrate incorrectly. Again, this is based on old and uncited knowledge. There's surely someone here involved in writing power management software that could chime in.