I have a question on this topic if you have the time. Now that the issue of streaming in data in time will be much less of (if at all) an issue, what will be the biggest limiting factor in ncreasing the detail of game worlds? GPU and CPU horsepower, the amount of memory, development realities like time/money?
I'm hesitant to answer this, as there are a lot of factors (and every dev will have their own take :)), but this is just my take on it. From where we stand, this is the order of importance:
Development pipeline and workflow, Drive/IO speed, CPU, GPU, RAM.
Tools and realities of asset creation is still incredibly important. Now that our drives are about to go mach speed we can finally hit the dream of loading all those assets in a snap. CPU was unfortunately quite slow this generation and was a bottleneck for a lot of operations. GPUs have been doing a ton of heavy lifting for a long time now. I'm basically not worried about GPUs at this point. And finally, with SSDs about to pop off, RAM can potentially be used in a much more efficient way and doesn't need to grow at the same rate.
Short answer is probably tools, engines catching up, and artist workflows improving (for instance, the advent of procedural asset workflow was an enormous boon to getting high quality assets in a much shorter time. We wouldn't have even gotten the detailed worlds we have now without it). Even with all our improvements in hardware, it still takes a long time to develop assets!