Do you have any evidence to support that accuracy is not effected? Plenty of companies out there that say it does and have proven it. just like to see some factual evidence to say that it does not.
Its a simple enough test, it requires a reasonable size machine to test the motors better , ive done this, found problems by doing the same test myself. simply set up a dti step back and forth into it by 1mm beginning with full step all the way up through your microstep settings, calculating backlash error and correcting and testing for correction accuracy between changing microsteps. If you are right the dial on every setting will be within the mechanical accuracy of the machine every time.
It wont It will drift out of accuracy as you go higher up the settings i promise you that.
And if you were correct, when i say my machine was struggling at 10x, by which i mean it was running unsmoothly and it would not stop correctly, Then tell me if this problem doesnt exist, Why at half step does it run perfectly?