I agree, I think the "variance" between labs has been decreasing over time and that benefits us.
As for terminology and definitions, I use the various Oxford Dictionaries and depending on context, here are some definitions of
Reliability:
The sampling
variance of a
statistic. For example, the reliability of where σ2=Var (
X) so, as the sample size increases,
X̄ n becomes a better estimator in the sense that the variability decreases;
From: "
reliability, in
The Concise Oxford Dictionary of Mathematics"
From the Oxford Dictionary of Academic English:
the quality of being able to be trusted,
the quality of being likely to be correct or true,
the quality of being able to work or operate for long periods without breaking down or needing attention.
The Last definition is the definition with which we generally associate with
reliability, but given the definitions above, I still do not think the word reliability applies to any of the OP's charts. Obviously, I do see "variances."
We may agree to disagree but I think we need to choose words carefully when describing technical data.
In today's environment, too many times proper word definitions are re-defined or re-interpreted to suit a particular method of Marketing, or to suit some ideas or strategies of a particular group.