In our continued refinement (occasionally to the point of absurdity) of defining the term "old school" with relation to D&D we're continually pushing the "old school" envelope back. I'm not quite sure where the end point of this is, but it's already absurdly far as it is, in my opinion.
While there is absolutely a style difference between D&D of the late 70s and AD&D as it developed, I still think that the notion that AD&D is the anti-"old school" and that old school has to, by definition, pre-date AD&D is kinda ridiculous (especially given that the OSR was specifically kicked off when the first retro-clone, OSRIC, managed to pass legal hurdles and become a reality—OSRIC, of course, being a retro-clone of AD&D, not OD&D. The later focus on OD&D was just that—a later focus, and largely a rediscovery of that style of play. But saying that only OD&D is old school, while AD&D is post-old school still seems like a quixotic splitting of hairs to me.
But this is nothing. The other day Jeffro made the argument that the Thief class was the end of old school—and in old school discussions, that's hardly a unique position. [NOTE: Since typing that, Jeffro has taken exception somewhat to my characterization of his position, so take that for what it's worth. Personally, I still don't know how else to read that post other than, paraphrasing, "The Thief class introduced a number of elements that were the roots of the New School changes to D&D, which is why Old Schoolers are always trying to 'fix' it." But although I don't know how else to read it, Jeffro has claimed that he does not believe that and he doesn't know where I got that conclusion, so... there you have it. Follow the link and read it for yourself and make your own conclusions. For purposes of this post, I'm going to talk about that argument whether or not Jeffro specifically made it, because whether he did or not, it's still a position that is held by some in the OSR.] While one can say that the way the Thief class was implemented may have had an unintended cascading effect that changed the tone of the game over time, that's not really the issue. The Thief class was being extensively used (pre-publication) at the very first Gencon that post-dated the publication of D&D—mere months after it was published. Greyhawk, the supplement that included the thief officially, was in print a mere year after the first printing of D&D. To suggest or even imply that the only old school game predates the thief, as can reasonably be inferred from both Jeffro and Maliszewski's posts (and many of the comments that follow) means that old school becomes a vanishingly small window of gaming, and begs the question; why not suggest that the publication of D&D in the first place was the end of old school! Gygax and Arneson really sold out when they printed the game up, man!
I disagree with the notion that the Thief class caused a cascading effect of limitations, though—or at least that if it did, it was the fault of stupid players, not the mechanics themselves. The argument goes a little something like this, and if I'm not making it with sufficient force and attention to detail, you'll have to forgive me, because I don't take it very seriously anyway: before the thief class, any character could do anything. After the thief class, it became common to believe that only thieves could listen at doors or pick locks or climb walls, etc. Also; it was the prototype for what later became skill systems, which even further eliminated the ability of any character to do anything. Before skill systems, there were no rules at all, and if a character tried to do something, the GM just told him how likely it would be based on common sense, his experience, his whim, etc. Most likely he rolled under his ability on 3d6, or was given a percent chance that he had to beat on a percentile roll (although that may have been an idea that started with the thief abilities anyway.)
This argument is of course absurd, because if a character could do it before without any rules for it, he could still do it without any rules for it. The GM could make the same ruling that he always made. Skills gave the GMs a framework whereby any character could attempt anything and he didn't have to make an arbitrary roll for it. Those who say skill systems stifle GM creativity are on a slippery slope to "why have any rules at all; they all stifle GM creativity." As I said, I have little sympathy for this argument. It can also be a reducto ad absurdum; if thieves are the only ones who can do those things, then why is every other class besides fighter (or fighting man) capable of fighting? Or why can only magic-users and clerics cast spells (Oh, whoops, that's a different can of worms.)
I've seen some players (Maliszewszki's post above, and the comments therein make this argument) that there's no problem with skill systems, but that skills and classes are fundamentally at odds and a game should either be a skill-based game or a class based game. I also think this is a reducto ad absurdum and actually requires the notion that each class is a straight-jacket and that only members of the class that are specifically designed to be the only one that can do stuff. I understand academically the concept of using classes to make easily digestible archetypes, but I disagree with the idea that that's important. I don't think that it's too complicated to understand the idea that people can dabble in a variety of different skills. They can have one profession, yet pursue other hobbies. They can excel by natural inclination at certain tasks and struggle with others, in spite of some grand designers attempt to couple both tasks together. After all, that's exactly how real life works.
Rather, I think that the Thief exposes a problem with D&D itself at a very fundamental level; that of the dungeon. The archetypes on which the Thief class is based (Bilbo, the Gray Mouser, Cugel, various picaresque-type characters) are not ones that appear in a dungeoneering environment. In fact, the dungeoneering environment is so unique, unprecedented (and monotonous) as to have literally no connection to anything that any normal person can understand. So yes, while there are very limited examples of it in the literature (Fafhrd and the Gray Mouser in the Thieves' Guild, the Fellowship in Moria, John Carter or Tarzan breaking out of various prisons, etc.) those are very limited examples, that are over with quickly. Primarily because if they dragged out any longer than they did in the stories in which they appear, they'd very quickly become extremely tedious and boring.
Almost every complaint I hear from OSRians about the Thief class have to do with its interaction with the dungeoneering environment, or how it (purportedly) changes other characters interaction with the dungeoneering environment—at least in the perception of many of the players. But what if this isn't a problem with the class itself, but rather one with the constant dungeoneering assumption of play?
For what its worth, this isn't unique to the thief. What's the ranger without wilderness travel, anyway? The ranger is another of my favorite archetypes (that I also rarely like the interpretation of) which always struggles because so seldom to D&D players actually spend any meaningful game-time in the wilderness. For me, since not only the archetypes that the thief (and ranger) model are among my favorites from the source material literature and because I find the dungeoneering paradigm tedious and ridiculous, naturally, I don't have problems with including them in the game per se. That doesn't mean that I don't have a lot of issues with many of the specific iterations of them that have appeared over the years.
Speaking of which, another curious curiosity with regard to the thief class is that as non-weapon proficiencies and their descendant, skill systems, spread through the game, the designers, in trying to protect the role of the Thief and give it a chance to shine, focused more and more on the backstab ability, which evolved into the Sneak Attack ability. This had the added side effect of transforming the Thief into an Assassin; even though the terminology never caught up. So this begs another question to; exactly what is the role of the Thief archetype in the game? What is it supposed to look like and how should it be modeled? Of course, there are various equally valid opinions on this question, depending on what kind of game you like to play and what you expect the character to actually do, which is one of the reason why these two classes in particular tend to get more customized, more house-ruled, and more "fixed" by amateur designers than most other classes.