There's an interesting post on Slashdot discussing people who get paid to game.
The most interesting comment, from an actual game tester:
The Natural Order of Things in game development goes more or less like this:
1. Finish the game, ignoring all bugs unless they crash more than 50% of systems.
2. Publish it and hope for...
4. If planning to do a sequel, release a patch to the first game, to make people think "you care"...
5. Publish the (buggy, untested) sequel and go back to #3.