From reading around the subject, IMO, I think people are over simplying the whole PS3/360 future capability issue..
I think the whole porting issue has come about because most developers idea of parallel programming only really works on the 360 architecture, i.e. just split your traditional design into 6 parallel threads, and off you go.. of course having a non symmetrical architecture puts the PS3 at a disadvantage that I think has been shown in shoddy ports, as this just doesn't 'work' on the PS3.
The downsides to this basic parallel design is that your CPU may be processing 6 times the code, but you just create a bottleneck at the CPU IO such as memory/GPU/BR/DVD/HDD, so as a 'system' it's still not getting the most out of it.
What developers are learning, and the PS3 architecture is forcing them to do, is to approach it from a different angle, with much more asymmetrical designs. If you do this, you are basically using some of the CPU power to ease some of the bottlenecks in the system (a good example being an SPE used to augment the RSX), with the idea that your system 'throughput' increases.
On the surface, especially using the SPE/RSX example, it would seem that the PS3 is far better suited to this then the 360, however, it actually transposes pretty well to the 360, since it has effectively 6 threads, and spare CPU capacity with a unified shader GPU, the actual system resources overall balance out fairly well. I think when one of the developers said that leading on the PS3 and therefore changing your design methods to produce asymmetrical designs, it benefits both platforms..
I'd say that game engine design now has a new 'avenue' to go down, we will see continual development on both platforms at a similar rate, most improvements will come from game engine design, not 'unlocking' hidden processor power.