Tube amplifiers that reflect sound changes when tube swapping occurs may not be designed very well.
A bit of an obnoxious assertion / wording on MC's part. It's true that if a circuit design runs its tubes in an egregious way (e.g. way out of spec, or far from optimal conditions for a given tube type), then tube rolling can have much more sonic impact than normal. But I've always found that tube rolling has audible effect, even in well-engineered components and for slots that have been advertised to "not" reveal tube rolling differences (e.g. cathode followers, grounded grid outputs with SS CCS).

