According to Marlinspike, it's not the SSL protocol that's the problem, it's how so many site implementations of SSL that are insecure. This includes many of the major banks, email systems, social networking sites, and more. Even most software update mechanisms.
One of the concerning aspects of Marlinspike's research, just isn't how banks, email systems, or social network sites are running vulnerable implementations of SSL -- it's that software update systems are doing the same. This needs to be a trusted process. As most users have their systems set for automatic updates, and if a malicious someone were able to spoof the connection then malware could be easily sent to the system, rather than a patch. The solution for this is simple: software makers need to digitally sign their updates.
When it comes to the weak X.509 certificates we've known about this for some time, and the industry is moving away from MD2 hashes.
Second, regarding Marlinspike, one needs to understand that SSL merely creates the secure tunnel between your system and whatever web site you're interacting - and does nothing for the security of your end point or the Web site itself. That means most any SSL attack that would result from this flaw would be a man-in-the middle attack where the bad guy would capture traffic flowing between your PC and the Web site.
Unfortunately, the Web sites you're connecting to are already wide open and vulnerable, and has been pointed out many times. Same is true for most end user systems. It's just easier to drop malware on Web sites that will infect systems, and then capture that data, or infect end points directly -- than it is to conduct man-in-the-middle attacks. So, while interesting, and serious, and a hole that needs to be plugged, it's simply another hole in the hull of a ship already taking considerable water.