> Doing an entire benchmark on that basis without making that clear is misleading.

A benchmark which provides source code and test data should never be categorized as "misleading." At worst it puts too much trust in the average reader's understanding of things. But look at it this way: if you littered every page that included a benchmark with all the relevant caveats, it'd be a total mess.

Or as David Simon says, "fuck the average viewer."