Ahh, so the software rounds first. The thing that saves this argument -- round before adding -- is that's what the government does. Add the numbers (at the left side) 12 times. Add them while rounding each one, then do it again and round at the end:
- Round each 1.0 > 12.0 > 12
- Round at end 1.1 > 13.2 > 13
- Round each 2.0 > 24.0 > 24
- Round at end 1.9 > 22.8 > 23
I was fretting over it because an error is still an error, and such a difference -- whether an actual error or it's that way because of methodology -- could land you in a different tax bracket.
