I'd like to see a series of tests that proves that OS and apps benefit from SSD but data does not. The OS X library and apps are far more like repositories of pictures, music, etc. than you suggest - most of that code just "sits where it is," too. I keep seeing this same assertion over and over, but my empirical feelings are just the opposite. Certainly, SSD allows the OS and apps to load in a flash (or Flash), but once they're loaded, a fair amount of that code remains RAM-resident. OTOH, data is constantly being manipulated (especially if you're working, as I often do, with image files in digital asset managers).
I believe that Fusion (not splitting into separate SSD and HDD) is the correct approach - we don't manage RAM manually, we depend on the OS to determine what belongs there, and what doesn't, and when. Effectively, Fusion's management of Flash storage is little different. The OS is in a far better position than we are to know exactly what is and is not in use, and how it's being used, and the time we spend consciously managing SSD distracts from the task at hand. (Is it more important to creatively manipulate our work, or creatively manipulate our system storage?)
I know that there will be times I'd forget to move a data file to SSD prior to opening (losing all benefit of SSD for that file). And what about versioning issues when I'm done with the file? If I copy, rather than move, I may have to waste time figuring out which of the versions is the correct one. When I edit images, I'm constantly opening and closing files - no freakin' way I'm going to manually manage that.
As to "more speed..." Pure SSD consistently benchmarks around 10%-20% faster than Fusion, and that's nothing to sneeze at. But splitting SSD and HDD only buys the benefit of "pure SSD" for those items that are on the SSD. This is kind of like the casino gambler who counts only winnings, and never counts losses.