Barely a week goes by without government ministers or MPs warning Facebook, Twitter, Google, YouTube (a subsidiary of Google), Instagram or WhatsApp (both owned by Facebook) that they must do more to prevent radical or dangerous ideas being spread. A “crackdown” is always just around the corner to protect users from harmful content.

Oddly, MPs never wonder whether they might be victims of the same effects of these tools that they, too, use all the time. Why not, though? We keep hearing that it’s a big problem for people to be repeatedly exposed to radical ideas and outspoken extremists. It’s just that for MPs, those tend to be within their own parties rather than on obscure YouTube channels.

If you look at the literature around radicalisation, and then at our politics, it’s hard not to think that social media – in particular WhatsApp, the messaging service that lets you communicate with one or many people in closed “groups” – is not helping. WhatsApp groups mark Westminster’s tribal lines; the Labour and Tory MPs who left to form the Independent Group were apparently thrown out of their respective party-oriented WhatsApp groups in a move as ceremonial as the breaking of a cashiered soldier’s sword.

“Every faction has a [WhatsApp] group,” one MP, who is concerned about the effects, told me. “The key point I think is it makes people immediately react, and also pick up on every slight [insult], combined with tweets.”

What if using the WhatsApp messaging service means the European Research Group is in effect radicalising its Brexiter Tory members, so they egg each other on to take more and more extreme positions in pushing for no deal? What if groups on Facebook are giving people the chance to say things they wouldn’t consider saying aloud in public – so that members of the Tory party make overtly Islamophobic comments, leading to suspension? Or encouraging even the leader of the Labour party to post in support of retaining an antisemitic mural (he subsequently said: “I sincerely regret that I did not look more closely at the image”). What if being on Twitter leads to MPs who make provably incorrect statements to stick to them, because to admit error is to lose face publicly and – more importantly – with your peers? Or leads to others pushing Islamophobic content on Facebook?

The dynamics of closed groups have been understood for years. The clearest finding is that they tend towards the most extreme position of their participants, something known as “the law of group polarisation”, described in a seminal paper in 1999 by Cass Sunstein, then at the University of Chicago. As Sunstein observed, it “helps to explain extremism, ‘radicalisation’, cultural shifts, and the behaviour of political parties and religious organisations”. And, he added, “it is closely connected to current concerns about the consequences of the internet”– which was not then nearly as pervasive as it is now.

Social science experiments show that if you pose a question as simple as “which of these lines is the same length?” to a group that has been secretly infiltrated with actors trying to steer the group towards an obviously wrong answer, then hey presto, you can get people to disbelieve their own eyes and agree with the actors up to 70% of the time.

Polarisation is the most dangerous form. That happens naturally when everyone has a broadly similar opinion about the topic in hand (for example, they all belong to the same political party), but some have more extreme views.

Sunstein pointed to experiments where a group was required to give a unanimous answer to whether a person with a secure, lifetime job should take a new job at a new company with an uncertain future. (You may be able to think of a political equivalent.) Almost every time, the group’s final advice was riskier than the advice the individuals themselves believed was best before the session. Crucially, after the decision, some of those who had previously been cautious became “radicalised” – picking the less safe decision when given the choice privately as an individual. Sunstein noted that “if a group decision is required, the group will tend toward an extreme point, given the original distribution of individual views”.

Try to read his paper and not hear echoes everywhere of what now happens. Who would want to be the MP in the ERG WhatsApp group saying maybe they can live with the backstop? Instead, the ERG has splintered from most Tories – and grown more extreme.

Equally, what sort of online discussions went on in Labour backrooms that meant a Merseyside councillor thought the phrase “Jew process” would be an acceptable pun, not calamitously insulting? And while increasing the disapproval of Christopher Chope is probably good for civilisation, the Tory chair Brandon Lewis was quickly met with opposition last August over the prospect of formally investigating Boris Johnson for his remarks about burqas. Via a Tory WhatsApp group, of course, where the former London mayoral candidate Zac Goldsmith (whose campaign was criticised for its anti-Muslim overtones) backed Johnson, and quickly saw his praise echoed and amplified by Andrea Leadsom and Nadine Dorries – neither widely known for their calming approach to controversy.

Is there a solution? Not an easy one. But the first step to solving the problem is recognising that you have one. Facebook already knows that the viral spread of information on WhatsApp is a problem which has led to deaths in India. MPs aren’t at risk of taking it that far, but they should at least recognise that the views they’re seeing in those WhatsApp groups could, shockingly, be completely wrong and that disagreement can be good for the health of the group – and, by proxy, all of us.

• Charles Arthur is the author of Cyber Wars: Hacks That Shocked the Business World