Simple Code

AWOL:
GoForSmoke: The Data Conversion Handbook tends to disagree with you, so I'll stick with Analog Devices' view.

Fundamentals of Sampled Data Systems on page 2.2 they explain "normalization".. It's rounding.
It has to be to be linear digital. It takes VREF/1024 to go from 0 to 1 out of 1023. Zero is given.
Read that short paragraph also cover scaling among the DP methods.

How is that different from me scaling? They say the same thing that I do right in the fundamentals.