Remember the general election, when most reports on voting intention turned out to be total tosh?
Well, here we go again.
The general election hopelessly wrong-footed most commentators for two reasons: dodgy polls and shouty lefty Twittervists.
The polls created an illusion that Ed Miliband and Labour were a nose in front. Labour's voluble activist base on Twitter then leapt on every iffy poll and each tweet describing yet another great session on the #Labourdoorstep to amplify and broadcast the narrative that Ed Miliband was about to become prime minister.
Understandably, most journalists looked on and followed the crowd. The pollsters and the Twittervists seemed to be saying the same thing.
A self-reinforcing spiral of delusion took hold that was only broken when the public's actual votes shattered the Westminster's conventional wisdom on the evening of 7 May.
Now, it's happening again in the Labour leadership race.
YouGov have provided the poll and the Twittervists have been hard at work since news of it broke last night (though in truth, this process was already under way, with the equally misleading CLP nominations being used as the metric of choice by Corbyn's online barmy army).
The problem, as at the general election, is that the polling is misleading.
In the case of the Labour leadership race, the capability of any polling company to accurately sample members is highly questionable.
For online polling, the problem is particularly acute.
Online polls, such as YouGov's, are based on people opting in to take part in a panel and answer questions. The self-selecting aspect of this process is normally mitigated by the sheer size of the panel and the rigour that can be applied to make the samples statistically representative and significant.
But when polling comparatively small groups - like members of political parties - problems inevitably emerge.
Because there are a limited number of Labour party members in the population - roughly 0.4% - it's difficult to ensure there are enough participants in the panel, let alone that they are representative of the group being polled. In fact, it's not clear how a polling company would know the right demographic profile of Labour members to adjust its sample, even if it could, without Labour's membership lists.
The unrepresentative nature of the sample is evident in the detail of the poll. The figures seem to suggest that over half of members and supporters have joined since 2010, a third the election in May!
This is ridiculous.
Yes, there will have been churn, but over half new in the past five years? One third new in a couple of months? A Labour staffer I spoke to this morning laughed out loud when presented with these figures.
It seems likely that the element of self-selection has led to a massive over-representation of fervent new recruits, eager to share their views.
The YouGov poll, in common with the previous emphasis in reports of CLP nominations, seems to have been skewed by representing activists' views, rather than those of typical members.
In contrast to street-pounding, leaflet delivering, phone-banking stalwarts of activist endeavour, most members are passive, verging on the inert.
They don't fall over themselves to give their views. For most, politics is pastime not a passion; the last election was a disappointment not a disaster.
Members are normal people and as such, they're not that different from - god forbid - the legions of ex-Labour voters who opted for the Tories in May.
They are a silent majority, similar to the silent majority of voters who regarded Ed Miliband as wholly unelectable and the Labour party as unable to manage the public purse.
And as at the election, it's a silent majority that appears to have been largely missed in the polling.
These people are not going to vote for Jeremy Corbyn. He is going to finish fourth, just as everyone expected when he first squeaked into the contest.
The silent majority is also why Liz Kendall will surprise many people with her result - the concerns about electability that she is articulating, echo what the vast majority of typical Labour members think.
And it's why Andy Burnham's baffling welfare hokey cokey earlier this week, will have lost him votes - flip flopping is not leaderly and tacking to the left on welfare is neither in line with mainstream public opinion nor builds confidence in Labour's electability.
The real unwritten story of this leadership election is the widening gulf between the majority of passive party members and activists.
This is a division that exists in all parties, but not since the early 1980s has the chasm been so great within Labour. Large swathes of Labour's activists are much closer to the Greens than the views of the ordinary members, or indeed, the leadership.
The depth off this divide will be evident when the results are announced on 12 September and yet another media inquest begins on how the polls and social media got it all so wrong.
Again.
Atul Hatwal is the editor of Uncut
This blog first appeared at Labour Uncut, and can be read here