Technically if you ddosed all the pools and went on to hit individual users with a coordinated strike by multiple botnets, you could then very easily 51% the network.
If you did that yes, but that part is not "very easy". You are talking about denying thousands of nodes instead of mere 10. It's just a matter of scale. The point is not if these networks can survive in their infancy, it's about which technology will be more resilient when they are big enough. That's why BCX's attack is not very useful at proving the point as well. When controlling these networks will be worth enough, DDOS will work for neither, just like DDOS'ing paypal wouldn't. The only point to be made, which is enough in my opinion, is that fewer points of failure is worse in principle. There is nothing stopping powerful entities from physically intervening with these nodes. I wouldn't want my money to be subject to the judgement of a few controlling entities.
Think of it this way... You could create a network of issuers and servers with OpenTransactions. Let's say, there are tens of these. There can be proxies and whatnot to protect their privacy, so the same methods apply. This way, you don't need to waste electricity on proof-of-work at all, and get the same protection of integrity. There are no 51% attacks. No more being subject to the "unfair distribution" argument as well. Distribute the money with free development bounties, so that all will be earned by hard work and issuers won't unjustly make money off of the currency itself. How is this not better?