I know the design goals of the door handle... to allow you to quickly and easily open the door. It's surprising, actually, how many door handles get it wrong (the excellent book "Design of Everyday Things" spends quite a bit of time on door handles).
For one to seriously do an overall evaluation of a site revamp, it helps tremendously to know what they were attempting to do, and why. Using your Mini analogy, if the 2009 removed the door handle altogether, and used some other mechanism for getting in the car. It would really help to know what they were going for.
You suggested CMS should do a serious evaluation, I was just giving a few reasons why I doubt it would happen. Even if it were to happen, I suspect you'd discover some people struggled finding what they were looking for, while others found it much more efficient. One of the things it seems you almost always find in usability testing is that no matter how obvious you think something is, lots of people do it some wholly different way. I suspect the reason it's more or less working for me, and I see improvment, while you're getting just the opposite, is that you and I use ebay in different ways.
It WOULD be fascinating to put you, me and 10 other people in a room, with 5 or 10 goals (find all '96 BMWs, or Saabs in your area, etc) and see how the different people honed in on that info. That's real usabiility testing, and I'd sure hope a company ebay's size did it, rather than just trust the geeks to know what's best.
I do agree with you, the listings are hosed with the new version. Things aren't showing up where they should, and are showing up where they shouldn't. But I suspect that's an issue in the database layer, not the new design. But even beyond that, there's several other bugs you and I have identified. There's no doubt it's got problems.
Now if you're asking if I think it was prudent to roll out design changes and database changes at the same time? No. I don't. Then again, despite the recent total overhaul of CMS, I personally prefer incremental changes when possible. A slight visual thing here, a new feature there, a database change where needed, while keeping a sense of continuity. There's been changes on Classic pretty much every day since the relaunch. Most folks haven't much noticed, which is a good thing.
There's a design term that basically means "staying out of the users way", but I can't remember it right now. It's early, and I'm cold.