What production methods should you use developing your game?

I'll work to come up with a more broad question next time. Dunno if I can top this one, though. :}

Allow me to preface these thoughts with the fact that my background is almost exclusively in AAA development. The smallest team I've worked with was about 25 people on the first Soldier of Fortune.

I know Agile methods are now such a hot topic that they've almost become mainstream, but I can't say even something like Scrum is required for successful development. I will say that my own limited experience implementing Agile methods has met with success, but it's still case-by-case as to whether that would solve more problems than it would fix. When I look at something like what Haunted Temple Studio is doing with Skulls of the Shogun I find it unlikely even a framework as almost universally applicable as Scrum would be more help than hindrance. I mean, they're four guys. And they're talking just a few months of development.
Is the importance of using a named method directly proportional to the size of the team, then? Well, going full on Wild West, code-what-you-feel with 200 developers seems unarguably questionable to me so maybe your team should be using something at that point, even if it's *gasp* waterfall. If you're working on the eleventeenth sequel to a particular sport game, though, maybe you can just do whatever you want. I haven't worked on more than a few sequels in my time, nor have I worked with a 200-person team so I can't comment authoritatively on that one.
That raises a point, though...does the familiarity of the project really determine how you do what you do? Or to put it in a more intriguing light, does the unfamiliarity drive your choice of methodology? If your team has primarily done shooters and now you're on a top-down action RPG, should that point you toward a more structured approach to your project? When we did exactly that at Raven we probably should've picked some sort of production methodology, I can tell you that. Instead, we drifted around a fair amount and took quite a while to create X-men Legends.

It may well come down to a case-by-case basis. The best general advice I could give is to ask the question of your project lead and/or studio management, "If someone brand new to the team came into your office and asked How does your team make a game, how would you answer them?" If you don't have an answer, or if it takes several tries over the span of several minutes, try to get someone in charge to sit down with you and do some research. Doing something new doesn't mean experimental or risky, but you should be implementing something with a wee bit more formality.