As Charles says, from what I've read of the CERT advisories, there is nothing proactive one can really do for these DDos attacks, besides securing machines from being hacked, correct?

A site can use anti-spoofing filters on their router/firewall (even if the ISP can't or won't do it on their end) to make sure that their machines don't forge source addresses. This might stop any of their machines which have been compromised from really doing participating in the attack. (I say "might" because the slave daemons don't have to forge addresses.) Other misc. ideas are in: http://www.cert.org/reports/dsit_workshop.pdf (BTW, a "advisory" version of unicast RPF-type stuff would be immensely helpful in deploying URPF, "source validation," ingress filtering, or whatever you want to call it.)