iiinova — where your digital self keeps evolving

Alberto Acerbi @acerbialberto.com · 17/09/2026

The framing matters because they shape policies: platform regulation, detection and takedown, and media-literacy “vaccines”. While these measures may be useful, we argue for a broader approach that also considers institutional accountability, media pluralism, and citizens’ participation and trust.

130

Alberto Acerbi @acerbialberto.com · 17/09/2026

Speakers often use a self-absolving and fatalistic narrative. Disinformation is portrayed as a “weapon” or “poison”, driven by hostile states, bad actors, and digital platforms, and citizens as vulnerable subjects who need protection.

110

Alberto Acerbi @acerbialberto.com · 17/09/2026

New paper out with Eliana Fattorini, Massimiano Bucchi, and Enzo Loner. “Weapon and poison? Framing disinformation in European Commission speeches, 2016–2024” We analyse how disinformation is framed in 238 speeches by European Commission in the last 10 years.

The figure shows 11 small bar charts arranged in a 3 by 4 grid, each labeled with a topic number from 1 to 11. Every bar chart has the same horizontal axis labeled misconception and the same vertical axis labeled proportion. The vertical scale runs from 0 to 0.5 with tick marks every 0.1. Within each chart, 6 vertical bars represent misconceptions from a shared legend on the right: A social media problem, Faster than reliable information, Impact on people’s behaviors, Many people are misinformed, People are easily misguided, and Pervades the Internet.

151

Alberto Acerbi @acerbialberto.com · 16/09/2026

"On why co-working with large language models does not put an end to thinking, but relocates part of it to the meta-level" medium.com/@stefano.pal...

medium.com

Prompting Is Thinking Too

On why co-working with large language models does not put an end to thinking, but relocates part of it to the meta-level.

030

Alberto Acerbi @acerbialberto.com · 16/09/2026

"Copying whatever the environment happens to show is enough to produce most of the collective structure of this population." arxiv.org/pdf/2609.09150

arxiv.org

010

Alberto Acerbi @acerbialberto.com · 11/09/2026

I just realised I mixed up former/latter in the opening sentence! :) "Whether you hate or love them (most likely the latter), social media..."

020

Reposted by Alberto Acerbi

Rob Sica @robsica.bsky.social · 09/09/2026

🔪"My evolutionary background, especially influenced by the idea of epistemic vigilance, suggested we are not as gullible as generally believed. If anything, we are too stubborn when evaluating new ideas."

machineculture.io

Model Convo: Alberto Acerbi

On Cultural Evolution, Epistemic Vigilance, and Technopanic

031

Alberto Acerbi @acerbialberto.com · 07/09/2026

Italian here :) I think they wanted to say that the computer *arranged* them by semantic groups or something like that...

030

Reposted by Alberto Acerbi

Mason Youngblood @masonyoungblood.bsky.social · 06/09/2026

New tutorial preprint about simulation-based inference! Shows how to fit a cultural evo model to baby name data using approx Bayesian computation, random forests, and deep learning. Wrote for @eslr.bsky.social but wanted to share with others in @culturalevolsoc.bsky.social! :) osf.io/preprints/so...

osf.io

OSF

14023

Alberto Acerbi @acerbialberto.com · 02/09/2026

Nature feature on AI homogenisation. Yours truly quoted at the end being "relatively optimistic". Unfortunately the part were I talk about our recent work did not go through, I guess they preferred the general comment of the "cognitive anthropologist"... www.nature.com/articles/d41...

nature.com

Bland new world: is AI making us all think the same?

Researchers worry that generative AI is homogenizing culture and cognition.

012

Alberto Acerbi @acerbialberto.com · 31/08/2026

Another food-related traditions that turns out to be relatively recent: www.nytimes.com/2026/08/30/w...

Dining late is a relatively modern Spanish custom. During World War II, the dictator General Francisco Franco pushed clocks on the Spanish mainland forward an hour. Spain never reverted, meaning its official time runs ahead of its geographic position.

030

Alberto Acerbi @acerbialberto.com · 29/08/2026

I did not know this: eBay ran field experiments turning off paid search ads on google and there was zero measurable effect on sales. onlinelibrary.wiley.com/doi/abs/10.3...

This begs the question: why do well-known companies spend large amounts of money on what seems to be a rather ineffective marketing channel?

0155

Alberto Acerbi @acerbialberto.com · 29/08/2026

Cultural evolutionists often talks about the possibility of loss of cultural traits. The walkman could be one: obsoletesony.substack.com/p/why-sony-c...

obsoletesony.substack.com

Why Sony Can’t Bring Back Its Classic Walkman Models

Explained by a former Sony engineer

0101

Reposted by Alberto Acerbi

Nicholas A. Christakis @nachristakis.bsky.social · 26/07/2026

We have done many field experiments documenting the policy relevance of social contagion and I wholly agree that “diffusion should be part of behavioural policy making.” This paper is a valuable exploration of mechanisms involved.

148

Alberto Acerbi @acerbialberto.com · 21/08/2026

PPS: There are points based on 19 months, others on 6 or 7, I did what I could with "ugly" data, as explained here (in Italian): mappine.it/p/mappine-10...

mappine.it

mappine #10: i brutti dati sulla sicurezza

001

Alberto Acerbi @acerbialberto.com · 21/08/2026

PS: Data is better than personal opinion and hearsay, but it's not the "Truth". It could be that crimes are being discovered less or that definitions change over time. But it definitely tells us that the common perception isn't obvious.

121

Alberto Acerbi @acerbialberto.com · 21/08/2026

The perception that crime is on the rise is widespread in Italy. But hey, there's data! I plotted data from the Viminale (via mappine.it) from 2012 to today on homicides, thefts, and robberies. They're below.

131

Alberto Acerbi @acerbialberto.com · 19/08/2026

However, each model has its own signature, so what we get is a fragmentation into different, internally homogenised, "cultures". This seems something to think about! Link: osf.io/preprints/ps...

osf.io

OSF

011

Alberto Acerbi @acerbialberto.com · 19/08/2026

Main result (for me): the frequency-dependent bias we estimate for LLMs is much stronger than anything reported for human transmission chains, in some cases an order of magnitude higher, i.e. we clearly detect an "homogenising" effect.

101

Alberto Acerbi @acerbialberto.com · 19/08/2026

We ask LLMs to select/mutate randomly, for efficiency, attractive, and original content, for a total of 16 combinations. We then apply ABC model fitting for frequency and content bias (as used on human and archaeological data).

100

Alberto Acerbi @acerbialberto.com · 19/08/2026

New preprint with @carigol.bsky.social and @jstubbersfield.bsky.social on AI/cultural evolution. Three LLMs models (Mistral 7B, Qwen2.5 7B, GPT-3.5) select and mutate short health-advice statements over 100 steps of transmission-chain.

Figure9:Correlation between measured cosine diversity and estimate of frequency-dependent bias(J) and content-dependent bias

1103

Alberto Acerbi @acerbialberto.com · 16/08/2026

"When a man is tired of social media, he is tired of life"? Hopefully not, because I am enjoying (any) social media less and less.

030

Alberto Acerbi @acerbialberto.com · 12/08/2026

If you live in the mountains (like I do) this website show you where it will be possible to observe the eclipse tonight eclipsemap.xyz#11/46.0664/1...

eclipsemap.xyz

Solar Eclipse at sunset · August 12, 2026

Dove vedere l'eclissi? · ¿Dónde ver el eclipse? · Wo die Finsternis sehen? · Où voir l'éclipse?

111

Alberto Acerbi @acerbialberto.com · 12/08/2026

This pattern often noticed for cars seems to hold also for furniture... from: europeancorrespondent.com/en/r/how-col...

Bar chart titled 'Furniture lost its colour: Recently purchased IKEA sofas are more likely to be grey than older ones.' It shows the most common sofa colours per year in IKEA catalogues from 1960 to 2021 as a sequence of coloured vertical stripes. Three callouts highlight the trend: green and blue dominate the 1960s; patterned sofas (stripes, dots, checks) take over in the 1990s, shown as dense black-and-tan patterned bars; and 2011 has the most neutral colours ever. Moving left to right, the bars shift from saturated greens, blues and reds in the 1960s-80s, through heavily patterned dark tones in the 1990s, to blacks and whites in the early 2000s, ending in solid grey and taupe shades through the 2010s to 2021. Source: IKEA catalogues 1960-2021, manual count. Credit: The European Correspondent, Toon Vos.

22011

Alberto Acerbi @acerbialberto.com · 07/08/2026

Another research in line with "The cultural ecology of social media" "Empirical research...has treated platforms...as distributive systems that expose users to particular messages. The mediatization literature, however, suggests shifting attention upstream: from circulation to production."

142

Reposted by Alberto Acerbi

koenfucius @koenfucius.bsky.social · 26/07/2026

Behavioural policy relies on social transmission—but how do different social learning mechanisms perform? Paper by @gaveltri.bsky.social & @acerbialberto.com examines four canonical social learning rules, and argues diffusion should be part of behavioural policy making: buff.ly/cbyAnuD

084

Alberto Acerbi @acerbialberto.com · 17/07/2026

I never mentioned this, most of my followers are academics: Technopanic was not written for you! I wrote it so that people outside academia could engage with the evidence, rather than having it filtered through news headlines. Pass it on to your friends! mybook.to/technopanic

mybook.to

Technopanic: How Justified Are Our Fears About Digital Media?

Read it now

171

Alberto Acerbi @acerbialberto.com · 17/07/2026

You are completely right! I mention it en passant a few sentences later, point is the book is aimed to non specialists and I want to first make the “classic” point that correlation is not causation in general.

110

Alberto Acerbi @acerbialberto.com · 29/06/2026

I am definitely applying! But, seriously, you should.

152

Alberto Acerbi @acerbialberto.com · 23/06/2026

I would say that no-one would claim that we "see no results of cultural discrimination/ways of thinking what the STEM are like with respect to gaps", I just think that the only-culture positions are less convincing - to me.

110

Alberto Acerbi @acerbialberto.com · 22/06/2026

Bravi cultural analytics students! Here a Pudding-style visualisation made (voluntarily!) for a final project. Italian political parties use a populistic language when they are at opposition, but not when at government. website2-tawny-nine.vercel.app

131

Alberto Acerbi @acerbialberto.com · 22/06/2026

...discussion, which is good!

100

Alberto Acerbi @acerbialberto.com · 22/06/2026

Thank you! It is not my research area, so I can't really judge on most of the details. My feeling was though that is a nuanced position, more reasonable (to me) than others that I see e.g. in sociology, so I am surprised this would be controversial. The points you make are indeed matter of...

110