Yes, and all it provides is anecdotal evidence. Do you have a link to Microsoft's tech specs for the Vista / 7 graphics stack where I can find factual evidence? Probably no, because they're only made available to graphics chip manufacturers under strict NDAs, hence "Where's the code?".
My point with the Windows graphics driver example is that if you plug untrusted / untested / uncertified / crappy software into otherwise stable software, you can't hold the latter responsible for the unpredictable behavior you may get. It's not specific to Windows, X or any other software or platform.
'My point with the Windows graphics driver example is that if you plug untrusted / untested / uncertified / crappy software into otherwise stable software, you can't hold the latter responsible for the unpredictable behavior you may get.'
I disagree. Stable software is well designed enough to not let unstable software break it - you have evidence of this in the article, although you as you've mentioned you chose not to believe it.
No, actually, they usually can't. As the article states, drivers that crash don't stop the presentation server from working. Did you read the article?