And the polls nearly got it right... except for that newcomer, Survation, which totally sucked. Much like our own newcomers.
Labour: 14,718 - 42.1% (+10.2)
Lib Dem: 11,160 - 31.9% (+0.3)
Cons: 4,481 - 12.8 % (-13.6)
UKIP: 2,029 - 5.8% (+1.9)
Nazis/BNP: 1,560 - 4.5% (-1.2)
Others: 982 - 2.8% (+2.3)
So what does this mean? Well, Labour has slight momentum, the only kind you can get from winning a seat you've held since (it's) creation; the Lib Dems save face, sparing Nick Clegg more knifes in the back, though expect more soon; and the Conservatives go down in total defeat. Fun!
What is kind of the kicker here is the fact that the Conservatives, by sparing their Lib Dem allies by not campaigning against them, it totally backfired, and not in the way you think (though it did in that way as well).
One, by not going after the Lib Dems, they allowed themselves to fade into the background as the third-party. Because they focused on only one of the frontrunners, namely Labour, they allowed all their attacks against them to naturally drift over to the natural alternative in Oldham, aka. the Lib Dems. So in reality, what probably happened was that the Lib Dem vote actually drifted over to Labour, and the Conservative's votes coalesced around the Lib Dems, seeing as how they would have had a member of the governing coalition either way. Result is that Labour wins, the Lib Dems remain credible, and you're totally tarnished as any sort of alternative.
Two, by allowing themselves to fall into the background, they played their hand worse because they could have won the seat. Think about it: had the Conservatives run a full campaign, they could have effectively gone after Labour and allowed the Lib Dems to defeat themselves, opening up the possibility that some Lib Dem voters, seeing their candidate faltering, could have drifted to the Conservatives, and some of those voters who went Labour may have gone Conservative because they became a more credible alternative.
Ah, well. It happens.