Why the pollsters got it wrong - and why their mistakes matter to the parties

He compared the snap polls with broader annual surveys and he crunched numbers. Lots of numbers.

And at the end of an immensely detailed 20 page report, he said the pollsters got it wrong because they asked the wrong people.

Forget "shy Tories," "lazy Labourites" or a late, undetected swing to the Tories, the pollsters believed David Cameron and Ed Miliband were neck and neck on the eve of last May's election because the people they spoke to were unrepresentative of the population at large.

They spoke to too many Labour supporters who, for some reason, are easier to get hold of on the phone or by email, and too few Conservatives. They also spoke to too few young people, who are less likely to vote.

Fatally, they failed to correct the errors during the usual "weighting" process which is designed to turn the responses of the first 1000 or so people they contact into a more representative snapshot.

Old-fashioned, time-consuming (and expensive) random sampling cannot be beaten, Professor Curtice said.

The pollsters will have their say next week when the initial findings of an independent inquiry set up by the industry are made public. But it's unlikely there will be wholesale changes to their methods given the pressure they are under to take the political temperature in just a couple of days or so.

How accurate will the polls be on May 4, the day before the Holyrood election? We'll have to wait and see.

The question of why the polls got it wrong last year is probably not causing sleepless nights for most people.

But the wide margin of error (the Tories were nearly seven percentage points ahead in the end, far from neck and neck) raises a series of other questions that should concern voters.

Are we paying too much attention to the polls?

Are the polls shaping the way parties campaign and, by extension, public opinion rather than simply trying to reflect it?

Ed Miliband's former spin doctor Tom Baldwin complained bitterly about the influence of the polls on Newsnight this week.

At the last election, the polls raised the possibility of a minority Labour government propped up by the SNP, a scenario that was unpopular among large sections of voters in England. It was ruthlessly exploited by the Tories who produced posters of Ed Miliband in Alex Salmond's pocket.

The election became a "giant process story," said Mr Baldwin, making it hard for Labour to get a hearing when it tried to raise concerns over Tory plans for the NHS or tax credits.

George Foulkes, the Labour peer, agrees. His gripe with the pollsters goes back even further, to September 7, 2014, to be precise, when a YouGov poll put the Yes campaign in the lead for the first time in the independence referendum.

Lord Foulkes, who is campaigning for the polling industry to be regulated, believes that poll changed history, panicking the three main pro-UK parties into making their "vow" of extra devolution and, ultimately, producing the Smith Commission and new Scotland Bill.

Not everyone would agree all that was down to a single poll. But even if YouGov only nudged Messrs Cameron, Clegg and Miliband a little, the point is still made. It's worth thinking about with the referendum on Britain's EU membership hurtling into view.

Even the latest Holyrood polls have implications for how the parties campaign. You might think, with the SNP 30 points ahead, the polls don't matter. The pollsters might be a bit off target, one way or the other, but they can't be THAT wrong about the outcome. Yet the SNP will have to manage such a commanding lead or risk shedding list votes to other parties. Labour will have to find a way of making its policies the story rather than its collapse or its battle with the Tories for second place.

In the end, the media commission the polls and the media must take responsibility for presenting the findings. Perhaps we need to be a little more circumspect. Adjust our own weighting, if you like.