you know, turnout may be up absolutely, but it's relatively low at under 70%, so i think it's reasonable to suggest that a lot of these soft tory but really undecided voters that they got over the phone just stayed at home, in addition to a clear shy labour bias.

the online polling may not be scientific, but it at least controls for apathy better than the phone polling does. i mean, you can imagine how this goes.

you get the voter on the phone, and she's watching the telly, barely listening.

"yeah."
"tory, i guess. that's what the papers say."
"i dunno."
"listen, i'm watching the telly, so bugger off."
*click*

then you get these 10% tory leads in the polls that are based on frustrated people trying to hang up the phone.

randomization is irreplaceable, and you're going to eventually run into online polls that get it horribly wrong (we've already had some in canada) because they aren't random. but, if you're lucky, the online polls are at least self-selecting for likely voters. it's one of those situations where something is wrong by method, but in being wrong by method is also accidentally consistently right in outcome.