Originally Posted by Erknie
I would like to see that test video done again, this time with some consistency in rpm used, and perhaps some monitoring with an oscilloscope showing the time delay between the signal sent from the gyro, and the signal sent to the motors. You have the issue of how long it takes for the motors to actually change speed after the signal to do so has been received. For all we know that delay could have been substantial, and a simple change of rpm may bring it into a phase where it stabilised the arm. I am not convinced by the footage, it's not to say it isn't true, it's just not all that scientific.
I also wanted to point out, on a different subject, that at 6000 RPM asking a motor to change faster than 100 Hz doesn't really seem realistic. This would equate to changing the speed more than once (at higher than 100Hz) during a single rotation.