I would ignore that poll. You have to keep in mind that the q6600 has almost a year of very succesful and happy overclockers behind it, and that's going to push people towards suggesting it. It looks like a hard choice now, because yes when OC'd the performance difference is going to be small. But keep in mind that SSE4 instructions are relatively new and haven't been implemented in most software/benchmarks, and when they are the q9300 will outperform the q6600 even more than it already does. You can't get as high a % OC on the q9300 because of the higher stock FSB and lower multi, but at the end of the day that doesn't matter.
On the other hand I would agree with those who have said you might run into trouble with a q9300 on a board that can't handle higher FSBs. But as long as you pay some attention to your MCH cooling (generally, just replace the stock TIM with AS5 or something else decent) and up the voltage on it, this shouldn't be a problem as long you have a popular enthusiast board that supports 45nm chips.
For reference, I got my q9300 (not in my sig because it's a totally seperate rig from my home system) to 3.375 GHz at 3.125 volts and acceptable temps under a tuniq tower on an abit ip35 pro without a single hiccup. Massively, MASSIVELY easier than the OC you can see on my sig rig.
Yes, you will get higher clock speeds on a q6600. Yes, the q6600 has more cache. But the evidence out there points to the q9300 performing slightly better anyway because of the CPU architecture improvements on the 45nm line of chips. And that (currently) small performance difference will increase as SSE4 becomes more common.