Yeah, I can't read anyone's mind here, but I really don't see any destructive motives from Haze (or anyone else, for that matter). If you look past some of the earlier posts, which tended to be less accurate and more angry, I think he at least makes a plausible argument. Basically, it seems that the original sin of MAME licensing was switching to a non-standard license, rather than a standard OSI license that was written up by IP experts. Then, as various people found loopholes in the license, it kept being updated. But, with too many contributors to actually get the permission of every one, it's not clear that MAMEdev had the ability to take code published under the old license and republish it under the new license without violating the old license. Now, as I've said, I don't know if this is true. But Haze says he's gotten legal advise on the matter, and I haven't heard anyone else say they've done this.
If, for the sake of argument, I accept that conclusion, then some of Haze's suggestions make a lot of sense. He's perfectly correct to not attribute partial copyright to himself in cases where he's made major modifications to code already in MAME. If the code wasn't properly licensed to begin with, then he has no legal rights to it. Perhaps, with a lot of work, they could get a completely clean version, but with such a long and complicated history of the code, it would be difficult to say this with a lot of confidence. And if they end up publishing code under GPL or BSD when they don't have the legal right to do so, then they're giving museums and academics a false assurance that they can use MAME code legally. Sanctioned commercial use of the program could perhaps bring unwanted attention to the project.