Peter Barrett

The takeaway: when democracies in a polarizing environment ban platforms, compliance is likely to follow partisan dynamics. Partisan identity shapes who stays and who goes, and effects outlast the ban itself. Platform governance can deepen fragmentation in the digital space 11/

Political violence is often treated as something extreme or exceptional. But much of the underlying psychology is not. It emerges from familiar processes: When people strongly identify with a group, threats to that group can feel deeply personal—sometimes leading to escalation rather than dialogue. I discussed these issues in a BBC radio special on political violence that aired today. I also explained how leaders and regular citizens can defuse violence and intergroup conflict. Understanding these dynamics is essential for anyone working in leadership, media, policy, and conflict resolution. You can listen to the full interview here with @aligoldsworthy : open.spotify.com/episode/7uWbmk…

@Evidencebbh @SullenSeagull Nous sommes malheureusement sur une plateforme qui ne permet que très peu l'échange constructif...même avec beaucoup de bonne volonté c'est très compliqué d'avoir de bons échanges ici lorsqu'on est pas d'accord.

@SullenSeagull Je n'abandonne pas définitivement, mais là ce n'est pas particulièrement judicieux. Hugo Clément est dans une contre attaque défensive, avoir cru qu'il pourrait s'ouvrir à un peu de remise en question ou même à une discussion était bien trop naïf. Et en face, chat échaudé blabla

Bon, je vais mettre un peu de côté mon envie de mieux saisir ce qui sous-tend les fractures épistémiques en me montrant compréhensif "avec les deux côtés", parce que je vais encore finir tout seul et canardé de partout :D Pis en plus, c'est certainement vain. Et j'ai du boulot.

@SteveStuWill Very similar to the percentage of social media users causing most of the disinformation and antisocial behavior. It should reassure us to know the vast majority of people aren't that antisocial.

A small number of criminals commit the vast majority of crimes. One famous study found that just “1% of the population [is] accountable for 63% of all violent crime convictions.” [Link below.]

@provisionalidea @danwilliamsphil Dan Williams' work is mentioned in the FT. Point is, AI having potential to build an epistemic commons is reasonable and since we already know LLMs are persuasive, makes sense to test directionality in AI-AI interaction for cost effective Hypothesis validation.

@PHSBarrett @danwilliamsphil what are you on about? who mentioned dan williams? and costello and pennycook use AI in interaction with human participants, not AI talking to AI.

I hate that the FT’s data guy is so consistently bad that every time I see something from him my visceral response is “something’s gonna be very wrong with this” and of course this is no different, piling bad work on previous bad work

@provisionalidea I was too bothered with the bots talking to bots thing but when you read more broadly on AI persuasion, it actually makes sense. I recommend looking at Costello and Pennycook's peer reviewed work. Also, @danwilliamsphil isn't "some blogger"...

he makes a strong assertion about the ability of models to persuade people but his method was to have chatbots talk to chatbots

Nous avons dépassé plus de la moitié de la campagne sur Ulule pour le #RECTLSE26 , merci de votre soutien ! Voici donc la deuxième "surprise" pensée pour vous remercier (il y en aura d'autres tout le long de la campagne). #TDAH #Barkley 👉fr.ulule.com/rec-2026/news/…

While social media is polarising, evidence suggests AI may nudge people towards the centre. This holds true of all studied models. Grok is more right-leaning than other models, but also has depolarising effects. By @jburnmurdoch.

J'ai le plaisir de vous annoncer la sortie de mon jeu « Fakemètre ». Le but est double : Apprendre aux élèves à évaluer la pertinence d'une source et leur permettre de faire évoluer leur avis en fonction des informations qu'ils reçoivent.

Could social media make us less polarized instead of more? We tested 5 algorithms on 3 platforms with 10,000 people for 6 months during the 2024 election, and found that the answer is yes. 🧵

Our recent PSPB paper showing that liberals and conservatives have different perceptions of victimhood (AoVs) is generating quite a bit of discussion on r/science. Short summary of the results (and 🔗 ) below 🧵

“Conflict entrepreneurs”--leaders who frequently use personal insults—are destroying social discourse and damaging democracy. An analysis of 2.2 million public statements from members of Congress finds that this type of rhetoric is linked to increased media coverage, but has no other benefits. This suggests a political incentive structure where the pursuit of media visibility alone sustains a form of discourse that may be corrosive to democratic norms, even without apparent electoral or financial rewards. #557388976" target="_blank" rel="nofollow noopener">academic.oup.com/pnasnexus/arti… If the media stopped giving these people attention, maybe they'd stop spewing insults and focus on trying to make the country a better place. Thoughts?

@SwipeWright I'd offer this as a self-supporting title : "When we miss something, we keep objects to remind us of it"

This new paper "explores the imbrications of queer identity" and is based on the concept of "diasporic longing," which it defines as "a sensuous ache of partiality and incompleteness that permeates my interlocutor's experiences of frayed histories and ephemeral homes."

Ulule démarre fort, merci à tous les soutiens. Tout le long de la campagne, il y aura des surprises, pour vous amuser ou vous faire sourire. La première est en ligne ! fr.ulule.com/rec-2026/news/…

@JohnHolbein1 You've got to admit that the four researchers and their families make for a pretty good sample.

12 people. 12. 6 males. 6 females. PNAS. Bonkers.

@jayvanbavel Interesting! Could we say there is an emotional amplification feedback loop? Emotional content -> high salience -> perception there is only emotional content -> self selection of users/creators -> belief that only emotional content succeeds -> creation of more emotional content

People consistently overestimated the average emotional intensity of social media content. This is associated with sharing more content on social media. But things are less intense than you think. psycnet.apa.org/record/2026-46…

People have really lost trust in their social media feed--algorithms have amplified the worst voices.

@AntoninAtger C'est un débat courant dans le petit monde des concours d'éloquence. Ce que j'ai longtemps apprécié dans cet exercice entant que prof c'est sa dimension émancipatrice pour les élèves. Ça leur donne un terrain de jeu d'exploration et d'expression. Je préfère encourager cela

Ce post fait polémique mais je pense que c'est un débat qui rejoint celui du langage. Faut-il que l'élève apprenne le "langage dominant" pour s'en sortir ? Ou faut-il faire évoluer l'acceptation d'autres manières de s'exprimer plus populaire ?

The World Giving Index tracks prosocial behavior such as helping a stranger, donating money, and volunteering time. In what seems like a world of woe, it's important to realize that these are increasing over time, contrary to most people's impressions. cafonline.org/insights/resea…

@daniel_dsj2110 Funny when you think the same guy hates it when french people use English words feering "englicising" the language... particular kind of circular thinking

French linguists on the English language