It does in terms of "reality overrides perception" and "what we want doesn't change whats real". Our ability to set hard limits in law doesn't change the fact our ability to actually test for those limits always have some margin of error.
In this case, the argument is the devices may be wrong because they're up to 6-10% off in some cases. Is that enough to invalidate measurements because of the margin of error? That's something the law doesn't deal with properly, but it really should.
What is the maximum margin of error allowable in a device used to determine legal status of an individual (in terms of drunkeness in this case)? without that being specified in law, what is the biggest margin of error socially acceptable? 0 is not practical or possible, so it has to be some sort of number.