"The search is the most important part of the process..."
I think that you might have a hard time finding anyone to disagree with this statement. When the examiner finds good art, the rest of the examination could almost take care of itself. (Not quite, but almost.) Good art is certainly a prerequisite to an efficient examination.
"...but the powers-that-be would rather offload that to the applicant."
I've seen the ESDs that are submitted with some of the Accelerated Examination cases, and I've spoken with examiners who have examined them. I personally wouldn't accept the art submitted as the best art; I'd still perform my own search. Hopefully the ESD would give me some clues as to the best places to direct my search, so at least it would still hopefully save me some time, though.
"...either Examiners work to become true experts in the art, or we're better off with a registration system..."
This is a good plan. But there are problems that stand in the way of making it happen.
First, after an examiner has built up a substantial docket, the constant flow of RCEs will make it possible (or more accurately, necessary) for the examiner to pick up fewer new cases to examine, because they need to spend most of their time working on amendments. The fewer new applications taken up for examination, the less opportunity to study an area of the art with which they are not familiar.
Second, there are many art areas that are broad enough that it is virtually impossible for examiners to gain expertise in all of the art that they might potentially examine (because they just don't see applications from all areas of their art). I often find myself spending a substantial part of my initial search finding and reading background information so I can get a good understanding of the invention and the related art. I honestly don't know how the examiners in the Business Methods arts do it. That area is seemingly so wide ranging that I imagine they need to do that almost every time they pick up a new application.
I completely agree that a good initial search is central to good examination, but it's tough to do when you've got only a very few hours in which to do it. For me, it's worth spending the extra time on the first action because of the time it saves me in answering the amendments. It saves me from writing many second non-final actions (which is definitely a contrast with my first few years examining).
The thing is, it invariably (from what I've seen) takes some hard experience and some time invested writing second and third non-finals and withdrawing finality for an examiner to figure this out. This will only happen if attorneys at least occasionally force examiners to do it, instead of filing an automatic RCE even when the art is not as good as it should be.
(And this is not a criticism of the way attorneys prosecute their clients' applications, really. It's simply one of the reasons why things are the way they are.)
So what is one to do? One solution would be to give examiners more time for their first actions.
Another possible solution? How about further subdividing the art that examiners handle? In order to make them true experts, it might be helpful to allow examiners to examine a much more narrow area of art than they do today (in some areas of art, anyway).