The question is whether Alaska polls overestimate Democratic support or underestimate incumbent support. (Too bad there were no 2012 polls to test this hypothesis)
Funnily enough, the 2010 polls both overestimated Democratic support AND underestimated incumbent support.
http://en.wikipedia.org/wiki/United_States_Senate_election_in_Alaska,_2010#Polling_2
Also, just a reminder, Research 2000 was proven to be a fradulent polling firm, so I'm not sure why Wikipedia lists it. Just nobody cares to edit old election articles?
Well, CNN was pretty damn accurate.