Given all the electronic scales already on the market are getting cheaper, all I can think is that he is talking about a high speed weighing system. These are used in industry. The problems with this application, however, have already been mentioned, and that is multiple competing overlapping error sources. I might have, say, a .45 Auto case that weighs 3 grains less than another. If that same case happens to have 3 grains too much powder, then the finished cartridge will look perfectly good on a scale but actually be dangerous.
As far as selecting cartridges for accuracy, there's is a statistical problem. Let's suppose your cases have 3 grains extreme spread. The standard deviation will then be in the range of 1 grain. Further suppose you have a powder measure that throws the charge to an extreme spread of 0.6 grains. It will have a standard deviation not far from 0.2 grains. Standard deviation interrelation adds as the square root of the sum of the squares, so the net cartridge standard deviation from these two factors will be the square root of 1²+0.2² or √1+.04=1.02. That's not much different from the case weight SD alone. So you could have the weight look pretty close to normal for the case brass spread when the powder is actually varying all over its extremes. In other words, the case weight variation will interact such that it will cover up all but really gross charge weight deviations most of the time. So weighing finished cartridges as an accuracy selection strategy won't likely work, and accuracy freaks are the main market for expensive tools.
As far as weighing for safety rather than accuracy, primers all weigh about 5 grains, and good brands of bullets are very consistent. But the only way you could detect a safe load vs. and unsafe load or even a missing primer is going to be to pre-weigh each case and keep track of that case through loading to isolate the powder weight as a variable.
Gunsite Orange Hat Family Member
CMP Certified GSM Master Instructor
NRA Certified Rifle Instructor
NRA Benefactor Member