The reviewer decides who is invited of course and so what if there are weak members of the panel? The breadth of opinion is what you want. You can find clear points where the whole panel agrees and some where no one is in agreement. You can assign likely characteristics to the ones with the strongest statistical agreement and the more speculative characteristics to the ones where people agreed less. A table with traits and agreement level could be quite useful for someone to see what most people who heard the piece of gear in that system thought it sounded like.How do you handle the issue of opinion quality? Who decides which audiophile friends have enough experience and critical listening skills?
All I can say is try it out and see what works. Happy to discuss with you privately about a framework for conducting the tests and tabulating the results.