Perhaps because Larrabee wasn't going to be a competitor? The last thing the industry needs is a player who can push monsterous marketting muscle behind a product that we would laugh at as a $20 part. That hurts all of us long term, and using massive marketting to push that horribly designed part is the only option they had to move it.
		
		
	 
If we believe the numbers from Intel's Siggraph 2008 paper, we need approximately 16 Larrabee cores running at 1GHz frequency to achieve 60 fps on FEAR at 1600x1200x4AA.
http://www.techspot.com/review/163-radeon-hd-4770/page7.html
Radeon 4670 gets 47 fps with 1600x1200x4AAx16AF, so its comparable. If they got 32 cores working at 2Ghz it might have had AMD/Nvidia's last gen high-end parts within arms reach.
It's not that bad, but they probably don't have the hardware working at 2GHz with 32 cores and even then its a >600mm2 part competing with chips at half the size like the HD5870, which beats the theoretical Larrabee SKU.
In essence, the RealWorldTech article's conclusion seems accurate. Had it been launched at that spec at early 2009, or even mid-2009 it might have been ok, but
#1. Writing high-performance drivers for gaming is different from having it function as a CPU
#2. Even in mid-2010 it might not have reached the maturity in the hardware they wanted to reach a year back
They will probably focus more on their IGP driver development. Rumors of Clarkdale IGP supporting video transcoding in hardware with upcoming drivers: 
http://www.xbitlabs.com/news/video/...ext_Generation_Platform_to_Support_GPGPU.html