It would be *more* susceptible to burnout than the current model, actually, since there's no provision for the scenario where none of the five randomly-selected arbitrators are willing to do any case work. In a system where the full Committee hears every case, a few active members can keep things moving (to some degree) even if the bulk of the Committee isn't actively participating.
(Or, in other words: we can currently close a 5-0 case if we find *any* five active arbs, but, in the new model, we would select five arbs first and then expect them to be active -- which is a rather dangerous assumption to make.)
You make a good point. It should work, though, as long as the arbs are honest about when they are and aren't able to be active. They'll be occasions when something comes up and an arb becomes unexpectedly inactive, but in most cases they should know far enough in advance whether they are able to take on a case or not. An inactive arb can be replaced by one of the others if necessary, of course.