Additionally, it's far from clear the examples reviewed were representative vs written to support the conclusion, so in addition to the problem of how variance is framed as a bug rather than a feature, it's not even really clear how much is actually occurring.

"Elections have consequences" is a well-known phenomena, and the idea that a branch other than the legislature might have influence over how the law is interpreted or executed shouldn't be news to anybody, least of all a member of the judiciary, many of whom will in fact be no doubt less qualified to make judgments vs agencies with specific technical and legal expertise.

It doesn't matter if the example is representative. It doesn't matter if it is 1%, 10%, or 50%. If agencies are able to act as a political branch and use Chevron as a cover to make these political decisions immune to judicial review, then the system is broken and needs to be fixed.

This is like defending qualified immunity for police by saying "well the arrests you presented where the police used the QI defense to cover over egregious rights violations are not a representative sample of all arrests made by the police."