According to Eddie Sefko, doing great work as always, Brendan Haywood's contract contains a SCANDALOUSLY underreported detail---any team that waives Haywood CAN SPREAD OUT THE PAYMENT OF THE REST OF HIS CONTRACT UNTIL 2026. In other words, they could pay no more than 2 million dollars a year.
What this means, in short: The Mavs continue to work their contract magic with contracts that find creative ways to provide huge amounts of cap space for other teams. Mark Cuban is a firm believer in getting players for JUST cap space, something he is almost certainly wrong about (see, for example, LeBron James, Dwyane Wade and Chris Bosh), but it does mean that the Mavs now have two contracts (Odom's being the other one) that provide huge amounts of matching salary for trades that magically, on the other side, turn into free cap space, which is how you get players like Tyson Chandler (underappreciated, uncertain, possibly overpaid, but sometimes immensely valuable) for essentially nothing.
It also means we're probably wrong about the Mavs amnestying Haywood, at least as a first option. He is apparently much more valuable as a trade chip than we had realized. The fact that Mark Cuban didn't bring this amazing clause up, when pundits were slamming him for overpaying for a center, goes to show how much interest Mark Cuban has in justifying himself to the media. If I'm Cuban, I'm now thinking of trading Haywood and Odom and keeping the amnesty option available for Marion, if it becomes necessary. I'm also thinking, if possible, that I can get one more year out of Marion first.
Update: Over at Dallasbasketball.com, Mike Fisher and David Lord, who is a cap genius, say that what's been reported about this isn't true---that it doesn't change the Mavericks, or the receiving team's CAP figure but, instead, it saves some owner personal money. So, unlike the case with Odom, which is a buyout, you get Hay you get an 8 million dollar cap hit---but if you're a penny-pinching owner, you don't actually have to write him 32 million dollar checks over three years.
So, if they're right and Sefko's wrong, this isn't as good, but it might still be something.