If you want to talk quantitatively about measurements, you need to do your statistics homework. This sort of reasoning is related to hypothesis testing and is part of the realm of the science of statistics. The other 1% of the time they are different based on dumb luck. That is to say, if two measurements are more than 1% apart from each other, then 99% of the time this difference represents truly different distances. For example, I might be able to measure a distance to within 1% precision given a confidence level of 99%. One can never know for sure if two unequal measurement results are just caused by dumb luck or if they are “really” different, but one can say with some quantified confidence level that the difference wasn't a fluke. It's a parameter that includes repeatability and random noise that prevents us from being able to decide if two different measurement results are different because they are measuring two different true values of the measurand, or because random errors have produced the difference. In that sense, ordinate resolution is a specification of the precision in the ordinate. When speaking just about an ordinate in isolation, resolution can be defined as the statistical blurring caused by measurement uncertainty. The minimum interval possible between two unequal ordinate measurements by a measurement system when measuring a DUT.
Nadovich, in Synthetic Instruments, 2005 Ordinate Quantization Interval