That's not true, as of now far more people have 5 series radeons than the "new" 68XX VLIW5 GPUs. Only VLIW4 is a really new design. The whole situation just reeks of planned obsolescence. The sad thing is that their competitor has even worse business practices.
I doubt AMD, or anyone else, has the resources to pour into performance enhancements for product that is EOL. It's a lot of work keeping the current product operating at it's peak.
As for the claim that was put forth that they actually sabotage their older equipment with inferior performance on newer driver releases to accelerate it's obsolescence, we'd need some proof of that. They'll sometimes have driver releases that reduce performance a bit on their current models as well. I'm assuming it's for the good of the overall performance and bug squashing.
I personally don't understand the perceived need for new drivers every month. It's nice that AMD offers this (They actually offer more than that, but it's one certified driver per month.), and if there's a particular performance issue that gets fixed, then great. If your equipment is working fine though and the next drivers don't have anything to offer for you, what's the drama? If I install a new driver that breaks something, I roll back to the previous driver, and do a bug ticket. Wait for the next one and if it's still there tell them again. In the meantime, if I have to use the same driver for 3mos. no big deal really.