Given how inaccurate polls have recently been, I think it'll be interesting and somewhat amusing to see just how far off they are this time, in 1992 for example, they were predicting either a hung parliament or a small Labour majority for pretty much the whole of the campaign yet the Conservatives ended up with 336 seats over Labours 271!
In the last general election, pre election polls predicted large gains for the Liberal Democrats which ultimately failed to materialise.
It'll be undoubtedly interesting to see how it all pans out, I'm expecting a Labour / SNP coalition (despite what Miliband says to the contrary) and then watch the fireworks once England wakes up to the fact their dog (Poodle?) has a very Scottish nationalist tail.....
I agree that historically polls have been pretty useless in the UK. But this year we have more polls than ever, and the agreement is extremely close (previous election often had 5-7% difference between pollsters). Polling methodology has also improved a lot. Even in 2010 online polling was scarce. We also have wide spread Ashcroft polls of individual constituencies of key marginal seats. Previously there was much less polling of seats, the pollster more aimed for population samples.