Something I meant to add to this weeks ago, was that benchmarks wise the E and Q are relatively the same, it's almost a moot point.
The more important factors are the compatibility, overclocking flexibility, heat etc.
I find the the E to be much more efficient and manageable, stressable and has a lower diminishing return than the Qs.
Your only option around this is to use the g0 stepping revision for the Qs, if you can obtain one.
The only thing really going for the Qs is the 'future proof' arguement, however it's flawed.
First thing to consider is that no application(most likely) is going to REQUIRE more than two cores for quite a long time. We're still waiting for a market influx of applications that even make use of more than two cores. C2D have plenty of life left in it.
Second, and something that most people do not consider is that the current Qs are two pairs of cores. Instead of 4, it's more like 2 + 2; they communicate through the system RAM and FSB, because they're located on two physically separate dies. This is a whole lot more convoluted than passing information around, within the CPU, like "true" quad core should.
I imagine by the time 'true' quad core based chips arrive, there will be a better number of applications that can make use of them at that time, and any previous Q versions will be dirt cheap.
So with that said, I would vote for the E technology because it can be tinkered with more, and by the time it's at it's 'rope's end,' so to speak, the better Qs will be on the market.
EDIT: Addition. I imagine a good plan would be to wait for those 'true' quad cores, and even if you did buy them when they were partially 'new,' you would at least be truly future proofing. To compliment that idea, prices of the newer GPUs, like 8900s and such would have come down, and you'd be able to properly upgrade. Rather than Q now, 8900 later, another Quad chip later and get stuck in a leap frog process.