I disagree to some degree, Mikecr ...
I find that my RCBS CM does drop a .02 grain overcharge on occasion; about ten percent of the time. It never registers final approval to an undercharge. The key is to watch the first digital read indication and wait for it to validate that reading with the second "beep". If there's an overcharge it is revealed in that load verification phase and it's a simple task of pulling out a few kernels to achieve the desired load. But the amount of error in any charge using the CM depends on the powder I'm using. Stick powder offers difference challenges than ball powder. Heck, even with a beam scale I find myself making adjustments to the final weight from time to time when the trickler moves faster than I expected. I don't believe we can reasonably expect scientific laboratory performance out of a digital powder measure and scale at the price the CM 1500 is sold for.