I've worked with a couple of good people who would style themselves "information architect". In my experience they do pretty well understanding the whole "lifecycle" of the information and help optimize information flow between various parts of a business. They shine well when there are multiple places data lives, and when there are many uses for shared information. Then they are good at figuring out where data lives, what is important for just an app or subset of apps, and when that information is a global requirement. They aren't DBAs, but they seem like it sometimes. They can figure out where there is bad duplication, where duplication makes everyone's life easeir, an when to use what representation for something. Basically a good IA can help ensure that your data flows well.

Programmers have a tendency to optimize the data architecture towards their specific project, while IAs optimize it between various business units. This does create a tension sometimes, but a good IA goes a long way to smoothing that.