Back when The Observer first unveiled Chris Wylie, its “whistleblower” on Cambridge Analytica, I wrote that while there were various interesting and worthwhile questions to explore – What happens to our data? What is it really worth? Who might use it illegally? How can we protect it? What legitimate campaigning uses for it are there? Do they work, and if so, how well? – it would be advisable to take claims of elaborate conspiracies with more than a pinch of salt.

There was an obvious risk in the fact that Wylie’s interviewer, Carole Cadwalladr, and her audience desperately, desperately wanted to hear him say particularly things about political defeats that they cannot stand and refuse to either acknowledge, accept or analyse.

That risk was evident in Cadwalladr’s previous reporting along similar lines.

In early 2017, she wrote a series of stories in which she unquestioningly repeated Arron Banks’s self-aggrandising pronouncements that he was the man who “bought Brexit”, despite the fact his fringe group failed to win the designation, resulting in drastic limits on its spending, and then became a byword for farcical blundering.

As I wrote almost exactly a year ago, Cadwalladr’s relationship with Banks became unhealthily symbiotic – he wanted to make Brexit all about him because it salved his ego, and she wanted Brexit to be all about him because:

‘…Banks embodies all of what [The Observer] would like Brexit to represent – namely, a British version of Trumpism – but also because his habit of implying he managed to secure victory by pushing “the boundary of everything, right to the edge” plays into its wishful thinking that the Leave vote was somehow unfairly achieved and therefore doesn’t really count’.

It was exactly such wishful thinking which gave rise to the untrue conspiracy theory among some of Cadwalladr’s fans that Banks’s Leave.EU and Vote Leave were secretly in cahoots – an idea completely laughable to anyone who had any dealings with any of those involved either at the time or since. I personally witnessed the absolutely furious dislike and disdain from both sides during and after the designation battle, including in private settings, and it was completely genuine.

The Banks saga wasn’t Cadwalladr’s only incident of somewhat over-ambitious theorising about sinister political influences. During the General Election, she wrote about the power of Facebook ‘dark ads’ during the General Election, only for her case study – the Tory campaign in Delyn – to promptly lose. She moved on from that inconvenient example, and so far as I can see doesn’t seem to have returned to the topic of Delyn, despite having declared that the tactic used there “undermined the entire democratic process” before the democratic process turned out to be ok after all.

It’s unsurprising, if troubling, then, that various holes have already started to appear in the reporting of the Wylie-derived allegations from the last couple of weeks.

As Dr Chris Kavanagh, a postdoctoral researcher at Oxford University’s Institute of Cognitive and Evolutionary Anthropology, wrote here, Cadwalladr’s personal investment in her source led her reporting to go extremely heavy on what she saw as his many wonderful character attributes, and simply to entirely accept his narrative about his work and its power – but apparently failed, perhaps as a result, to challenge him on essential technical questions about his story.

As Kavanagh puts it, “the most important” of Wylie’s claims, about the alleged effectiveness of CA’s approach, is “lacking almost any supporting evidence”:

‘To be crystal clear, I’m not arguing that Cambridge Analytica and Kogan were innocent. At the very least, it is clear they were doing things that were contrary to Facebook’s data sharing policies…What I am arguing is that Cambridge Analytica are not the puppet masters they are being widely portrayed as. If anything they are much more akin to Donald Trump; making widely exaggerated claims about their abilities and getting lots of attention as a result.’

That’s an exaggeration that we have seen Cadwalladr indulge before, of course, in her reporting of Banks. Might she be too invested in trying to help one side of the referendum campaign? “This isn’t about Remain or Leave,” she claimed on the Marr Show – while signally failing to answer Isabel Oakeshott’s question of whether she would be investigating the actions of the Remain campaign or just Leavers.

Then, yesterday, even as The Observer covered new claims from the same source, the newspaper published a correction (buried on page 50) to no fewer than three of Cadwalladr’s articles from the week before. Alongside getting facts like the job titles of a subject wrong, and even getting the work responsibilities of one of her own sources wrong, there was a very significant ‘clarification’:

‘…we are happy to clarify that we did not intend to suggest that AggregateIQ is a direct part and/or the Canadian branch of Cambridge Analytica, or that it has been involved in the exploitation of Facebook data, or otherwise been involved in any of the alleged wrongdoing made against Cambridge Analytica. Further, we did not intend to suggest that AIQ secretly and unethically co-ordinated with Cambridge Analytica on the EU referendum…’

The supposed part of AIQ in the grand conspiracy theory was central to the previous week’s reporting – and yet here is The Observer ‘clarifying’ that the company did not in fact have that role after all. The idea that they “did not intend” to suggest this is a strange way to put it – the ‘unintended’ assertion appears to have been picked up by various of their readers, and has been quite widely repeated since (including by Wylie, according to Channel 4 News). Having to ‘clarify’ it suggests that the reporting went further than the paper now believes the evidence can support.

Perhaps greater scepticism and an accompanying willingness to ask tougher questions, even if they threatened to puncture comfortable and wishful thinking, might have been wise after all?