A few weeks ago the launch of Rome 2 brought back certain memories of the Elemental launch. Basically they'd stuck to a release date that they should have pushed back, and the result was a game that was several patches sort of decently playable, along with various other problems that will take longer to sort out. Now with the launch of GC3, I'm kind of thinking about the lessons learnt from those sort of debacles. I'm sure SD have done this to death in-house, but speaking as someone who was with Elemental from pre-alpha, here's my thoughts about the dev process and things SD must watch out for.
1. Don't commit to a release date too early (no matter how much your marketing team begs you for it). Development is an unpredictable business, which my day-to-day job is making me painfully aware of at the moment. Also, is far better to let a date slip at short notice than release a bug-fest. Sure, you'll get some hate and frustrations for it, but nowhere near as much as the alternative.
2. Keep someone high-up independent from the process. I think this was one of the fundamental issues in Elemental, which I remember Brad talking about before. When you are tied up in something you really care about, it is easy to have your vision clouded by wanting it to be as good as you want it. In Elemental everyone high-up was involved, and it meant that there was no devil's advocate to point out the flaws that were deep in the game. You need someone independent to be honest and forthright about issues.
3. It is better to ditch an unworkable feature and try something else than endlessly try and tinker it to save it from oblivion. There were a few features early on in Elemental that were 'sounds like a good idea at the time' but totally unworkable. The prime example I remember was sovereign death = game over. A cool idea but a gameplay killer as who wants to risk a game-over in something you've played 20 hours on, especially when it is hard to retreat from battles and there's a large element of chance. This stuck around and stuck around with umpteen tweaks, when it was never going to work. This relates a bit to the above point - you need a sharpshooter on bad ideas (as well as the alpha and beta testing masses).
4. Testers can give much better feedback if they know what is coming. This is a hard one, and may only really work for alpha if at all (as it would require a certain amount of keeping stuff out of the eye of anyone not on the alpha). One problem I found with Elemental was that the only source of info on updates was public journals by Brad et al. This meant that, as there were cool features they wanted to keep under wraps from a 'keeping the public interested' point of view, it was very hard to get an idea of the game except from what was in front of you. I could report bugs, and say 'this works, this doesn't', but often I was left with 'this is rubbish, but I'm sure it will improve' without really knowing this way or that. In the end I found the pre-alpha stage simply didn't work for this reason - the game was too incomplete to test from a 'are these features fun' point of view, and since the process was very long we all ended up falling into the traps of points 2 and 3.
I'm still debating whether to sign up for the alpha and all the cool stuff - it depends whether I think I will be able to devote enough time and energy to make it worth it. But I have confidence this time that SD have learnt from that saga.