I agree with gmaxwell's statements, and I'll also add that in the existing Bitcoin system p2pool will never be strictly competitive even if it were cost-free to run because p2pool will always have a higher orphan rate vs. well-run centralized pools. We're fortunate that the difference in profit is currently small enough that it's lost in the noise of variance, but increased importance of transaction fees and a larger - especially uncapped - blocksize will change that for the worse in the medium to long term. Additionally in the short-term efforts that improve block relaying efficiency by transmitting transaction lists rather than full transactions, and even just dedicated connections between large pools, will also harm p2pool by making the orphan rates of large pools less than it.

But it's not all doom and gloom you know: sure p2pool and Bitcoin itself have some fundamental flaws, but the impact of those flaws won't be fully felt until some time in the future. For now we can learn a lot about crypto-currencies and when the flaws do become critical the next system will have a chance at fixing them. I don't think efforts to improve p2pool are going to have anywhere near the impact people think they will, but they aren't going to be wasted either.