I think the article was a little slanted toward a commercial viewpoint. If IBM were writing EVMS as a commercial product to fill some niche it's one thing. But this was a project designed to further kernel development as a whole. When a different idea proved to be the more popular one they simply decided to work on something that may be useful instead of beating a dead horse, so to speak. Whether EVMS got accepted or not IBM would not have directly profited off it. They profit off having a system they can use on their big-iron, and the other product will help them on that as well.
So by my view point IBM is just continuing to be a useful part of the Linux community instead of just throwing their weight around. That's how I look at it anyway.