... except, of course, the fact that marketing people might think they could sell more N800s if they break compatibility intentionally and thereby force consumers to get the new hardware sooner.

The more I think about it, the more it seems to me that this is the only reason. You see, there's this Nokia executive, and his tech people tell him:
"We can make the OS2007 available for the 770. It will take some extra time and effort for the original N800-codebase to separate things a bit, clean up the mess. And of course the team will then need xy weeks to do the backporting and testing. We estimate costs of EUR xy for the whole project and a delay of xy days for the N800."

Then this Nokia-guy asks:
"Sounds expensive. What if we don't do all that?"

The techs:
"That would be really bad, software development would focus on the new OS2007 pretty soon and leave our existing 770-users without 3rd party updates sooner or later. They'd probably have to buy a N800, then."

Now the Nokia executive understands:
"Now, you tell me plan A is to invest money and sell less N800s. Plan B is to have the new OS2007 available sooner, not invest any extra money and make more people buy the N800. - Well, this is a tough decision. Let me think about it for a moment..."