I think with Deus Ex:HR, it was more a case of the ending being rushed than anything else. I have no idea if it actually was, but I can't think of any other reason to essentially throw out the established gameplay at the very end of the game other than they were simply running out of time and had to get it done. The gradual tapering off of the HUB cities would also seem to support that conclusion.

Which is actually where I see a lot of big games ending up these days. Uncharted 1 and 2 actually suffered from this a great deal as well. Because they simply wrote the game as they went along, the final sections are by far the least interesting, and the final bosses were also underwhelming at best and annoying bullet sponges on the highest difficulty.

A lot more games fall into the category of ran out of time so the ending suffers as a result. And in large part I blame the sky high budgets and the difficulty in developing on present hardware. It is very hard to do a good amount of prototyping and playtesting, particularly when it's crunch time and you have under a year to finish the damn thing and have it on shelves. The only developers not really affected by this are either those who had a clear vision from the start and accomplished it relatively smoothly, or those with all of the time in the world to finish the game. But since not every developer is Valve, most don't have that luxury.

Some of it could also be that writing a satisfying ending requires a good writer, and frankly, many video game writers aren't that good. Even a lot of the stuff from companies known for making story and character driven games like Bioware and Bethesda I find completely underwhelming. Though admittedly, it's not always the writers fault. Many companies are getting better at having a writer involved from day 1 so things go smoothly there, but not all of them. And the ones that don't usually suffer the most for it.