I don't remember them all anymore -- I keep only loose tabs on Wikipedia these days. I was a very active editor and a founding member of a WikiProject in the 2004-2005 timeframe. I thought the blanket removal of most fair use images and the policy banning them was inappropriate and one of the least pragmatic things from an editorial quality perspective that could have been done.
That's the only big policy change since I quit that I can remember right now, though I know there have been several other that I felt were misguided or inappropriate. When I browse talk pages these days, I see very many new policy links, and that's part of the problem. Wikipedia has a pretty steep learning curve just to format an edit in a way that won't be reverted under a cavalcade of acronyms from the WP pagespace.
Edits on controversial topics are nearly impossible, primarily because articles will often be locked in the first place, or if they're not, each member on each side exhausts his/her three reverts, someone complains to an admin, and the page is locked so that it can all pick back up in a few days. It becomes especially difficult on pages that are hand-sanitized by a politician's staffers or other editors paid to keep WP articles favorable, because they literally have all day and they make their money by being combative, intimidating opposing editors, and otherwise staving off legitimate unfavorable edits.
And that's part of what makes this all so frustrating. Wikipedia has no real conflict resolution mechanisms. Open-source software is supposedly an inspiration for WP, but OSS has very clear, authoritarian project leaders that don't tolerate silliness. WP is an absolute free-for-all; there are policies, but in general they just mean you have to avoid specific phraseology and can still do whatever you want. As such, editing WP is very inconsistent but almost universally frustrating and it's often clear even to readers which group has established turf over a given article or set of articles.
On top of all of this, Jimmy Wales has an extreme aura of arrogance and ethical pliability. See the incidents around "co-founder" and "founder" of Wikipedia for just one example; there have been some other good ones over the years. Wales misappropriated Wikipedia, really the brain child of Larry Sanger, for his own and hasn't known what to do with it since. Sanger, meanwhile, started another wiki-based site that evolved on the Wikipedia model with approved revisions and a few other small tweaks (Citizendium, I don't know if it's still alive or not).
It also seems to me that Wales has been trying to carefully straddle the community line regarding policies, etc., (except a few cases where lawyers specifically said "this part is not optional, sorry") because he doesn't really know what's good for WP. It's all some kind of anomaly to him and he's just trying to keep a light touch so he doesn't scare it away. As such, WP yet languishes in its morass of "whoever-is-the-most-obsessed" methodology for resolving editorial conflict and lacks any serious drive or leadership.
WP is already like a big corporation because it already has a CEO that got there through business school instead of innovation. That's always the sign that you shouldn't expect many interesting advances from a given group anymore -- they are interested in maintaining their model and milking it for money until someone comes around and obsoletes them. Hiring a chief executive trained in milking money out of existing infrastructure instead of a chief executive trained in contributing something new is a huge red flag from every non-financial perspective (though the finances usually languish along with the invention and research).
Basically, it boils down to frustration over the lack of conceptual or procedural improvement in Wikipedia and the general belief that Wales is blatantly unqualified in almost all relevant respects and sort of wandered into his position accidentally. I don't necessarily blame him for that (though I do blame him for the morally dubious stuff, like Bomis, Sanger, and that girl), it just doesn't bode well for the organization.
And, as I'm sure we all know, it's generally a pretty bad thing when confused people get a whole bunch of money, as Wales just procured for himself. Wikimedia already emanates these astonishingly corporate vibes, I think we should only expect those to increase.