Spiralling costs of game development.

There has been a lot of discussion by various parties on the expected increase in cost of game development on the "next generation" platforms.

A good current gen title developed by an external developer is probably in the $5-$10 million range, and some of the larger companies are spending $30-$50 million internally on what they consider to be potential AAA titles.

The first thing to note here is the cost disparity between internally developed and externally developed titles (although it's not an apples to apples comparison), this exists for a whole host of reasons. External developers are more focused on development, they have to work much more closely within budgets, larger franchises are usually developed internally, internal teams have more management overhead and have to deal with much greater scrutiny, typically internal teams are larger, which leads to significant development issues.

So why develop product internally rather than externally? The argument has always been that you have greater quality control on internal product. How much that really reflects in real shipping quality is somewhat debatable. The other less often mentioned reason is that you have greater control over the money you spend, If a publisher spends a million dollars externally, he has no real guarantee that the money won't be used to bail out some other publishers title rather than being spent on his.

I've built a lot of games over the years and frankly ignoring the issue of parallelism (which is another blog) the technology problems are well understood. If anything it's gotten easier to write systems like graphics engine runtimes than it was in the PS1/N64 days.

So why are products going to get more expensive?
It's largely to do with the scale of the content. Consumers have continually rising expectations from a game, and those expectations dictate a certain quality level and quantity of content.

Given that project durations aren't getting longer that means more people, unfortunately a lot of existing technology is simply not well designed for very large teams. The problems with putting 100 content people on a system designed with 5 in mind are significant.

I've worked on legacy code bases in teams of 20+ engineers, where the majority of engineering time is spent propping up the technology and tools as the content teams try to work with continually broken builds.

The solution is changing the way we develop, putting real infastructure and process in place, designing patterns for problems that scale. But I don't see it happening anytime soon, a lot of the scalable and robust patterns go against the cycle counting/to the metal mindset that still exists in console game development. There is nothing wrong with this mindset and at some level it's a necessity for technical excellence, but with very large teams you can't focus on it to the exclusion of everything else.