As someone noted upthread, QA (Quality Assurance) is the publisher's responsibility, not the developer's. The normal scenario is that the publishers will run QA, send fixes back to the developers along with a deadline and budget to deal with them, and this will take care of most of the problems. If the game has particularly notable problems, the publishers will again do another QA run, this time for a patch.

The problem with Obsidian is that they are an independent studio which is either hired to work on a particular game, or creates a game and sells it to a publisher. This used to be commonplace, but is now extremely rare. Most studios have either been bought out by the big publishers (Blizzard and Infinity Ward by Activision, BioWare by Electronic Arts etc) or become monstrously rich in their own right through other schemes (Valve via their Steam service, for example). Obsidian's position means that companies often hire them, pay them less than what they would pay their own teams, shorten deadlines without warning and generally treat them in a disproportionately unfair way.

To whit:

With KNIGHTS OF THE OLD REPUBLIC II LucasArts suddenly shaved a third off the development time and budget of the game, as they suddenly decided they had to get the game out before Christmas. Obsidian did the best they could, but no company could have achieved that objective in the time they had. It's actually a miracle that the game hangs together as well as it did.

With ALPHA PROTOCOL Obsidian sold the game idea to Sega, who were initially excited about it. However, for reasons that are still unclear (but speculated to be related to the departure of some of Sega's game development team, possibly leaving their replacements with a game they weren't interested in), Sega did not pay Obsidian properly and did not communicate with them much during development. When Obsidian delivered what was effectively the alpha build of the game, expecting Sega to run QA and get back to them, Sega instead simply cut off all funding and communication and announced the game's release date. When Sega then delayed release by six months, Obsidian offered to work on the code and do a bug pass, but Sega did not respond. They also didn't advertise the game very much. It sounds very much like an insider politics problem, with the game getting dumped on people who didn't care about it and got rid of it quickly.

NEW VEGAS didn't actually have that many issues (certainly not compared to FALLOUT 3 itself) and was a superior game on almost every level to FO3. Bizarrely, however, a lot of magazines and sites that had completely ignored similar (or even worse) bugs in FO3 and gave it high scores ragged on NEW VEGAS for it and gave it a slightly lower score, to the point where Obsidian missed out on a promised bonus payment because the game failed to hit a Metacritic score by one point: the fact that Bethesda, paying millions in advertising for the game and effectively buying good reviws from the press, were not as concerned about bad reviews for NV as they were FO3 because it would not reflect as badly on them may have played a role in this. NEW VEGAS outsold FALLOUT 3 by a huge margin (almost double its first month sales) but Obsidian didn't see any of that extra money as, per their agreement, they took a one-off flat flee for their work on the game. The rest of the profits - which were extremely considerable - went to Bethesda.

DUNGEON SIEGE III, on the other hand, shipped with absolutely no problems whatsoever. Obsidian were given a good enough budget, a good enough development timetable and had publishers - Square of all people - who understood the game-making process and backed up their decisions. The game was a little dull creatively, but it was released bug-free.

The only game that Obsidian really caused their own problems on was NEVERWINTER NIGHTS II. Using established technology and having both reasonable development time and a decent budget, they really shouldn't have produced a game with that many issues (technical and creative). However, they did fix most of them pretty quickly and their two expansions are generally regarded as being hugely superior to the base game.

Obsidian have had extremely bad luck, but I think that a lot of their problems stem from remaining independent but still making big-budget, high-profile games. That leaves them at the mercy of publishers who want to dump on them at a moment's notice.