Oh Twitter…

Experiencing a Twitter take-down

A few days ago, I was scrolling on Twitter when I came across a Twitter thread by Dr. Matthew Sweet (@drmatthewsweet) about Johann Hari’s new book Stolen Focus. I haven’t read Stolen Focus, only the parts about our work. Those parts were completely fine, see full backstory below.

I don’t know if it counts as “viral”, but the thread has 1.9K retweets and almost 5K likes at the time of writing.

As I read the thread, I have to admit that I grew somewhat uneasy.

Disturbingly (from my perspective), the thread uses our paper (with Philipp Lorenz-Spreen, Philipp Hövel, and Bjarke Mønsted) Accelerating Dynamics of Collective Attention as a main example of a problem with Hari’s book. Here’s Dr. Sweet getting into it:

He goes on:

This is where it got even more troublesome.

Sweet doesn’t seem to understand the context of the quote and ends up completely misrepresenting the point we’re making in the paper.

We do say that the “phenomenon lacks a strong empirical foundation”. But the point of that statment (which is from the abstract) is to suggest that the research we present in the paper, provides that very missing foundation.

(I’m also a little bit offended that Sweet takes a potshot at us by stating that the paper “didn’t make a big impact” [1].)

So what’s the criticism of our work?

Ok. The next tweets contain the actual criticism of our paper:

That’s it?!

The argument seems to be that if you’re a “fellow historian”, you won’t like our approach. Perhaps, between the lines, there is also a claim that the datasets we study are not representative enough.

Importantly, from substantive perspective, Sweet doesn’t engage with what we’re trying to do in the paper at all. And there are zero attempts to address the empirical behaviour that we are observing in the paper.

Our careful, peer reviewed work documents systematic changes across time in a range of datasets. Somehow that doesn’t matter.

Collective versus Individual

To be fair, in one subsequent tweet Sweet goes beyond criticising our work and points to a problem he has with the book.

I fully agree that these two phenomena should not be conflated. For the record, we are very careful not to do that in the paper [2]. (And in the excerpt I read from the book, that was not a problem either. )

The point about Chaucer

But regarding the part of the tweet above which mentions Chaucer’s 9-day wonder, it once again is clear that Sweet hasn’t understood what we’re doing in the paper.

The point of our paper is that this “9-day wonder” is a period of time that we show to be shrinking over time. And it looks like it has been shrinking for a long time. So whatever period Chaucer was talking about is much shorter today.

Through careful empirical work, our paper documents precisely that evolution.

So what now?

The big question now is what we, as authors, should do in a situation like this?

On one hand, I am proud of our paper. As explained above, even with my most generous reading, I find Sweet’s criticism of our work to be loud, full of mistakes, and without almost any substance (good old argumentum ad lapidem, cf [1]). And since thousands of people have read his criticism on Twitter, I’d like to publicly call bullshit on the whole thing.

On the other hand, my intuition is that it’s best just to stay out of Twitter fights [3]. Twitter is not the place for nuanced discourse.

For now, I’ve chosen the option of writing this blog post.

I think the post is long enough and navel-gazing enough that no-one will read it. (Sorry, for making so much out of a small thing, folks, but it is my blog after all.)

It’s also an opportunity to add some nuance to the discussion that’s just not possible on Twitter. For now, this is my solution to have a defence of our work on the record, while not getting in Twitter fights

Addendum: Backstory

This whole thing started back in late 2019 I was contacted by Johann Hari who wanted to talk about a paper that I’d recently written in collaboration with Philipp Lorenz-Spreen, Philipp Hövel, and Bjarke Mønsted: Accelerating Dynamics of Collective Attention.

The paper had come out earlier that year, and we’d all been doing a lot of press for it, so I didn’t think much of it. Hari was in Copenhagen, he stopped by my office and we had a long, nice chat about the paper.

Then, in January 2021, I heard from Hari again. He was hard at work on his book, and sent me the text of the part about our work (and the conversation we had) to read through. His email specifically noted “It’s very important to me to make sure every aspect of the book is entirely factually accurate so I wanted to send it to you so you could read it through.”

I read the excerpt, and I found it to be a careful description of our work, with more detail than many other journalists would care to add. There were also a few sections of me speculating about reasons and mechanisms, but I found those sections to be clearly labeled as such. So I sent it back with a couple of small technical notes.

Full disclosure: I was also a little bit flattered because Hari had turned me into a kind of living character in the book – a kind of hero, which was new and exciting.

Addendum: I’m not alone in thinking that @drmatthewsweet is being disingenuous

After writing the above, I came across a Twitter thread by Erik Hoel making very similar points. I’ve unrolled it below for the reader’s convenience

Originally tweeted by Erik Hoel (@erikphoel) on January 7, 2022.

1/ How to grind an axe with a pop-sci book. You might have seen the viral thread of @DrMatthewSweet implying that the author @johannhari101 cannot “be trusted to handle data.”

2/ This is a great way to get some attention for yourself: pick a book, then find a way to say the author is misusing sources (one of the gravest authorial sins) when really you simply disagree with the original sources (a commonplace, boring objection that gets you no likes)

3/ @DrMatthewSweet does precisely this switcheroo in his recent viral thread. Let’s first look at what he says is the most “serious claim,” the strongest evidence of @johannhari101‘s inability to “handle data” (the second objection)

4/ @DrMatthewSweet implies the authors of the paper @johannhari101 cites *concede* themselves that the hypothesis of shrinking cultural attention spans “lacks a strong empirical foundation.” The implication is that Hari misuses the citation.

5/ Except the authors don’t concede anything like what @DrMatthewSweet is saying! That’s a quote from the Introduction of the paper, when the authors talk about *previous research* and say “but so far” – implying their very paper *is* the empirical research that’s needed.

6/ Does @DrMatthewSweet quote that? No. So in accusing another author of cherry-picking, he cherry picks, and not just some data, but a wild out-of-context quote that makes it sound like the conclusion of the study. It’s not! It’s the *previous* research. Here’s the abstract:

7/ It’s obvious that the paper says what @johannhari101 says it says. That, of course, doesn’t mean it’s correct! It could be wrong. I haven’t read @johannhari101‘s book, I don’t know him personally, and I’m pretty skeptical of pop-sci books lately:


8/ But this was @DrMatthewSweet‘s “most serious claim” and it’s really just that he doesn’t agree with the paper, NOT that @johannhari101 misused it. To make make this point, @DrMatthewSweet took a pull quote totally out-of-context. The Discussion is clear


9/ The other claims are much smaller, like that @johannhari101 cited a non-peer-reviewed study to make a point, and that he doesn’t know that Facebook let’s you find nearby friends. Frankly, they seem like smaller sins than @DrMatthewSweet‘s misrepresentation of that pull quote.

10/ So it looks like rather than @johannhari101 misusing sources, @DrMatthewSweet directly does so in his thread. Overall, it’s just a substitution of a boring disagreement (“I don’t like this book & the research behind it”) with the exciting implication of authorial misconduct.

Originally tweeted by Erik Hoel (@erikphoel) on January 7, 2022.


[1] For fun, I clicked on the paper’s “metrics” page – which provides a measure of online impact. Here’s what it states:

I don’t know what you have to do to “make a big impact” in Sweet’s world, but I am personally of the opinion that being in the 99th percentile of all tracked articles of a similar age in all journals is pretty decent.

And while we’re nitpicking and speculating wildly. Half the author team is from Germany, so it’s also factually incorrect to state that the researchers are “in Denmark”. Probably it’s my paranoia here, but how is it relevant where we’re from!? Am I detecting a hint that research from Denmark should be taken less seriously?

[2] As detailed in the “Backstory” addendum, I got to read the part of the book about our work, and there Hari is careful to talk about collective attention and not individual attention.

[3] Also, it turns out that Hari is a controversial figure in the UK. While my personal interactions with him have been great, a Twitter search reveals that a lot of people have an axe to grind with him. So I also worry that if I jump in to defend our work that somehow it will be confused with a defence of the book as a whole (which I still haven’t read in full and thus don’t have an opinion about), and which might direct the intense vitriol of the Hari-haters my way … something that I am also not really interested in.

Categorized as Misc

By Sune Lehmann

I’m a Professor at the Department of Applied Mathematics and Computer Science, at the Technical University of Denmark. I'm also an Adjunct Professor at the department of Sociology, University of Copenhagen


  1. “I think the post is long enough and navel-gazing enough that no-one will read it.”

    Wrong! Great post – I say less Twitter, more writing. What are you researching next??

  2. “I think the post is long enough and navel-gazing enough that no-one will read it.“

    Wrong! Loved the post – I say less Twitter more long form writing. What are you researching next? What’s new in networks?!

Leave a comment

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: