Originally Posted by jjpjimmy
Thank you for taking the time to draw the image, that's precisely what I was trying to say.
No one has shown sufficient evidence/ convinced me that this is NOT due to human error. What do you think it means to me when I can recreate BOTH results on many different mouses? It SCREAMS human error. The timing problem addressed in the video may be a problem but you need to show me that there are no human errors in the testing method before
moving up to a more complicated object to try troubleshooting.
2. How would you explain if one were to be able to reproduce both
the error and correct behavior on 3 different mouses?
I ask you, why am I seeing / lines, especially prominent in OP's image AND most other 'error' results.
You are right to assume human error being at work when there is no hard evidence to assume otherwise. But there was "evidence" provided that can cancel out human error.
1. Your explanation was based on the assumption that drifting is predominantly going to the left, due to most people being right-handed. But the direction the cursor is dirfting towards is determined by the direction of rotation.
2. Prior to us knowing from the video what exactly causes the behaviour, and even more so now, differences between mice are understandable (differences in timing). Different results for the same mouse on the same PC with the same settings I would have explained by people subconsciously compensating.
You can rotate the mouse as much as you want, even without a centered sensor, if you are not sliding the mouse across the mousepad physically, there is no way for the cursor to drift that far. That cancels out human error. And while I agree that the "results" are not reliable, because we don't have a common method of testing/don't know what exactly the tester did or how he did it, the test where you simply look at the screen, stack circles on top of each other and wait for your fingers/wrist/arm to move or grip to change is pretty fail-safe. After a while, it is easily noticeable and continuing to draw circles on the same spot without changing the grab/position becomes practically impossible.
/ behaviour of the drift can be explained by the circle not being not fully closed, but overshooting beyond the point of the circle's start. This could actually be some kind of counter-measure manufacturers have taken against circle drift, my OMB does drift in a \ fashion.
Originally Posted by Gidra
The timing issue he explains in the video as the cause of circle drift is in in reference to start up /wake up time - the time it takes for the system to pick up movement after rest. By this explanation doing stops after every circle would cause more offset, not less as you observed. The sensor "shouldn't" stop tracking in a continuous circle motion.
It could be that his explanation simplifies this issue, though. Polling rate is an interesting topic. You think that you can reduce the problem with a low polling rate?
That's what I wondered as well. He said registration of first motion, the "wake-up" of either the mouse or the PC should be at fault in that case. But as you rightly pointed out, when continuously moving the mouse, there is no wake-up, in which case there should also be no drift. Since there is though, there has to be continous delay. Maybe wake-up refers to creation of every count in the mouse processor and registration of every report on the PC? I mean, even constant motion practically has to wake the chain. Higher polling rates would then reduce the issue, since they reduce delay. A lot of speculation though. The only thing I know right know is that delay causes circle drift, meaning that the worse the drift is, the more timing issues there are in the chain. Trying different settings and checking for the amount of drifting is all we can really do right now. All my mice drift less at native CPI for example.
@ Ino.: Everything with drawing circles does indeed reek of human error. But until now everybody has failed to explain how a cursor could be practically infinitely moved to the right without the mouse being moved to the right. No matter how you move, rotate or grab the mouse, there is simply no way to do that.