I would not check vs. 1-2-3 block unless its a premium quality and quite new.
I am guessing my scales are within +-0.001 in say 10 inches but I can play around with it to get it more - the actual display is 0.0001 but just tightening the gibs has the effect of moving things around by 0.001.
I think we chatted about this on Kevin's post but I cant locate it right now. The longer the standard is, theoretically the more accurate you can calibrate the DRO. It doesn't have to be 123 or 246, those are just common blocks people may have. Because 0.001" grind tolerance over 3" block is a bigger discrepancy number than 0.001" over 6"block. But if you happened to have a longer rod of any arbitrary length like 14.923" that's perfectly fine because you are just indicating off that, comparing to DRO reading whatever it is & then entering the compensation factor to fudge them together. The trick is, how did you know the rod was exactly 14.923" to begin with? It all comes down to trusting the most accurate ruler.
I think this what people are counting on with 123 block stackups. If the the manufacturing tolerance is +/- 0.001" and you got a 50% random distribution with 2 smalls with 2 bigs, you have exactly 12.000". But unless you measured them with something better to validate, they could also be 11.996" - 12.004".
I doubt any of us are physically cutting to these tolerances in the real world of our mills anyways. But we can dream until a surface grinder shows up under the Christmas tree LOL
Yes a DRO does not lie when you lock down tables & see drift. Or our typically non-counterbalanced handles freewheel rotate a smidge & move the leadscrew a thou one way or another. Thats the beauty of DROs. They are just the messenger, they just read displacement independently.