So it's fair to say that despite my reservations, I've come to terms in one way or another with the various D&Disms, and want to use the system again. But at this point, more than at any other point in the history of D&D, which edition do you play is an interesting and relevent question.
The reason for this is, first of all, the OGL. The OGL has inadvertently opened the doors back to every single edition of D&D that's ever been in print. The third edition is, of course, expressly open and therefore impossible to stamp out, but by utilizing the OGL, gamers have managed to reverse engineer "retro clones" of various older editions.
And, of course, 4e is available as well.
Here's my personal opinion on each of the various editions of D&D, including a few that are technically not D&D at all. The attached graphic is kinda preview of what's to come, but let's not get ahead of ourselves just yet...
OD&D - Original D&D, retro-cloned via Swords & Wizardry. Honestly, I've never played this, except once as a kid before I considered myself a gamer in any fashion whatsoever. But… I also wouldn't. It's primitive. The roots as a tactical wargame are very apparent. It lacks something that I think is crucial to a good game today; rules to define characters. The simplicity of actually running the game is to be applauded, but it's mostly just simple because it's so primitive that it hadn't occurred to the designers to define things yet. The game was very poorly written, assumed that you owned other products (including products produced by a competitor!) and simply lacked options overall. The retro-clone version of it cleans some of that up, but this edition still offers me too little of what I want, and what it does offer can be had elsewhere in better form.
AD&D (1e) - Advanced Dungeons & Dragons, first edition, retro-cloned via OSRIC (Old School Reference and Index Compilation). I actually have bad memories of this game. While many D&Ders see this as the golden age of D&D, this is the edition of the game that eventually made me throw up my hands and give up on D&D. A fair amount of complexity was added to the game, although the complexity didn't address the issues that I personally thought were important. The game still suffered from a very high degree of arbitrariness, poor organization, poorly thought out add-ons and subsystems, and (although this is a minor annoyance) poor presentation and graphic design.
BD&D (all versions) - Basic Dungeons & Dragons up through Rules Cyclopedia, retro-cloned via Labyrinth Lord. I'm not honestly familiar with all of the many iterations of this "edition", which properly should probably be split out into several editions of very similar but not identical games. Basic was a bit of a misnomer, though---in no way was the game basic. It evolved almost as much complexity as AD&D (although much of its complexity had a different focus), it was in no way simpler to run, and it wasn't really less limited than AD&D either, for that matter. It was merely a concurrent game that was similar to AD&D, but… not. For my money, of all the TSR editions of the game, this is the one I'd be most likely to be willing to return to, but that doesn't mean that I'd be very excited about it. It was a continuation, if you will, of the tone of the OD&D game, but without the AD&D like mechanics. Therefore, I like and dislike mostly the same things about it; the ease and simplicity of running the game is good, the arbitrariness and lack of good definition around character abilities and traits is poor.
AD&D (2e) There is no retro-clone of 2e, because basically all 2e was was a re-write of 1e, a reorganization of 1e, and a removal of some of the stuff that made 1e good. Other than the organization aspect of the change, these were all "Bad Things"™ in my opinion; they didn't fix any of the problems I did have with the game other than the organization of the core books, and instead we got a watered down, "politically correct" version that had notably caved to pressure from some non-customers about what the game should and shouldn't include. There's little to recommend this version of the game at least until it became significantly revamped with kits, skills and powers, and other mechanical add-ons that made the game more robust. And even then, a lot of those add-ons created more problems than they solved anyway. 2e was also an era of innovative and unique settings. Not all of these were successes (I never could get past the fundamentally stupid high concept of Spelljammer for instance. And the execution, which had ideas like generalissimo space hippos, didn't help) but many of them are still well-loved. Looking at some of them that I've since picked up (I didn't actually play D&D during the 2e era myself) it's painfully apparent to me that many of these novel and interesting ideas were significantly held back by a clunky rules system at its core that was in drastic need of overhaul.
D&D (3e) The split between AD&D and D&D (or BD&D) was finally brought to an end after TSR went backrupt and was acquired by Wizards of the Coast. They released D&D 3e in 2000. This edition fixed many of my lingering problems and concerns about D&D as a game, but sadly, it left many others unaddressed, and in fact got worse in a few others. What I liked; 1) much of the arbitrariness was gone. There were now few mechanical subsystems that operated differently than the core rules, the core rules themselves operated under a very streamlined and consistent paradigm, and player options were brought to the fore. 2) character definition was significantly improved via a robust skill and feat system. You also were able to define your character many different ways via multiclassing and prestige classes. The classes no longer felt like bland straightjackets. 3) Subsequent games based on the same "engine" showed that it could be easily modded to allow for various different feels. D20 Star Wars, d20 Wheel of Time, d20 Modern and d20 Call of Cthulhu were all reasonably good games. 4) The OGL allowed for "third party" publishers to produce their own material for the game. At first, this was mostly just add-ons for D&D itself; adventures, new monster books, settings, etc. but as the movement matured, it paved the way for all new games, like Castles & Crusades, True20, and for that matter the retro-clones themselves.
However, 3e came with its share of the bad, too. 1) the combat system made it difficult to run combat without a graphical representation of the characters and their environment. Realistically, this means minis and a battlemap. While I like minis for their own sake, and I don't mind playing a tactical minis combat game, I'd rather not have my roleplaying game practically demand minis. 2) The game gets increasingly complex and difficult to run, especially at higher level. Higher level games, I've noticed, include an awful lot of flipping back and forth through rulebooks to see how exactly did that rule work again? 3) It didn't really address a lot of D&Disms. Levels in particular and the incredible difference between a low and high level version of the same character, don't model anything I know from fantasy fiction at all. 4) this isn't a problem for me per se, but it's one that I've heard a lot and I can see where it comes from. I took seriously the motto "Tools, not rules" and it fitted the GMing style I'd been accustomed to anyway. So, to me, the fact that there are a lot of rules covering a lot of situations was a nifty toolbox that I could dig into as I saw fit. Other players, however, saw it as a straightjacket and a complexity inducing nightmare, as they felt duty-bound to accurately reflect every single rule. As an example, calculating Jump skill Difficulty Classes: I saw the examples as just that---examples that gave me an idea what an appropriate DC should be to get the effect I wanted. Others spent all kinds of time actually calculating all of the inputs into a DC for each and every Jump check. So, although it's not a problem for me, I can see how this is a problematic aspect of this edition.
D&D (3.5) To be perfectly honest, I'm still pretty ticked off about 3.5. 3.5 was written, allegedly, to fix some problems that had come up with 3e. It did do that. A bit. It also "broke" as much as it fixed. And mostly, it just changed stuff that was fine to something else that was fine, so that you got no improvement, but you got a lot of confusion. Then to cap it all off, you were supposed to buy all the books over again that you'd already bought. Or at least a great deal of them; the three core books were re-released with minor changes, and the original run of class-based splatbooks was also all re-released. I've grudgingly made my peace with the changes, since it's easier to find 3.5 players and books than 3e these days, since there is a small improvement, and since backwards compatibility and online SRDs means that you don't really have to rebuy the books if you don't want to and you can still play just fine. The subsequent splatbooks are a significant improvement, too, which went a long way towards convincing me. But the entire deal left a sour taste in my mouth.
D&D (4e) I've never played 4e. I have no interest in 4e. As far as I'm concerned, it came about at a time that I had no interest in migrating to a new system. Also, whatever lingering problems I have with the system don't seem to be the ones that 4e addressed; in fact, if anything they made some of my personal concerns worse rather than better. I'm not a hater; I don't have a personal beef against 4e like many people on the Internet. I just have no interest in it.
Pathfinder. While not strictly a version of D&D, Pathfinder could be considered an "alt.4e; what 4e could have been like if it were more like 3.5." Or, you could call it 3.75 maybe. It's not published by Wizards of the Coast, nor is it published under the D&D brand, but this, clearly, is an edition of D&D nonetheless. The reasons it came about were Paizo being put over a barrel somewhat in terms of their business model. Wizards of the Coast dithered on the OGL for 4e, and even sent out some hints that there may not be one or that it could be crippled by restrictions. Paizo needed to keep publishing stuff, but couldn't get distributors to sell it if it was support for an out of print game. So, they needed to put 3.5 back in print somehow. Since 3.5, as a complete game, is open content under the OGL, in theory, all they had to do was package the SRD and print it, and sell it. Of course, if they were going to go to all that trouble, why not see about tweaking it and making a few improvements? This is what lead to the Pathfinder RPG. While I applaud Paizo for acting decisively to protect their business and make sure that the edition that they (and I) prefer is still in print, in terms of what the game actually offers, Pathfinder is plagued by the same problems that 3.5 was… it doesn't fix enough to justify rebuying all my stuff. The good news here is that Paizo suspected that might be true, and offered it, in addition to a hefty and expensive full color book with beautiful artwork that really pretty much justifies the price by itself, as a $10 pdf. The Bestiary (i.e. Monster Manual) is available as another $10 pdf. At that price, like I said, even if you ignore everything except the art, you've got a pretty good deal.