I don't see myself working in any game studio. There is not much innovation going on, the pay is not very good, etc.
That depends a hell of a lot on the studio.
There are good places to work, and they do plenty of innovative stuff, with competitive pay rates with non-gaming programming jobs. They're hard as hell to break into, sure, but they do exist.
It's the same in any industry or any career. If you want to work for the top 1% of companies, you better be a top 1% employee. Life in the first two standard deviations is not all roses and rainbows in any business. I've had my share of experiences like frob's in the non-games software world where the quality of life was utter rubbish - both as a contractor and as a full-time employee. Overall it comes down to how hard you're willing to work to get into a good position, and how qualified you are to stay on track in that position over time.
Compared to the horror stories I hear coming out of the supposedly "good" employers in the non-games software world, I'm pretty damn happy to be working for a studio.
Well, I know Valve is pretty good, and they pick up projects like Portal... but overall I dunno, I just don't see any of big guys making something particularly new. Even something as old as generating unique faces is omg super innovative there. The other software industry is also not very good of course. It's all mostly redoing old shit over and over again everywhere with few exceptions (Google's self driving car for example of real exception).
I just don't see how a big name studio would end up paying me as much as I got on Polynomial. Someone might, now, after I've released a competent game, but now I can make a next game and avoid the mistakes I made, and that should pay off better. (Polynomial was my first game. There's one thing I am sure I did right - going very obsessive on polish - and it was correct to make very basic gameplay for the first game - but it'll be better to have more diverse gameplay in the second)
Regarding role of luck: it shouldn't be either overstated or understated... something like angry birds, well, there's a lot of such games, many of them good, just 1..2 uberpopularity slots, whichever takes them is up to luck because nobody's really doing some sort of careful comparing like in sports. On the other hand, mmorpgs, that's heavy monetary investments, someone puts in more money than anyone else (including into marketing), has actually a good game, and captures the niche, not a lot up to luck.