When you choose to publish with PLOS, your research makes an impact. Make your work accessible to all, without restrictions, and accelerate scientific discovery with options like preprints and published peer review that make your work more Open.

PLOS BLOGS Absolutely Maybe

Mapping the Mastodon Migration: Is It a One-Way Trip or an Each-Way Bet for Science Twitter?

Cartoon of a sad-looking bird out on a limb.

Like many people, I opened a Mastodon account on the weekend the Musk era began on Twitter. This post picks up from my first Twitter migration post, a few days after that. I’d been looking into ways of diversifying my social media activity for a few months. I had my fingers tightly crossed that Mastodon’s open source platform would be my kind of place: Not a Twitter alternative, but something with its better features, and without its most pernicious ones.

A lot has happened since November 3! I don’t know how much of science Twitter has taken the off ramp without joining the Mastodon migration. Just because people aren’t shutting down their Twitter accounts, doesn’t mean they’re still there. For those who are engaged, I think we’re along a spectrum – from diehard Team Twitter at one end, to all-in Team Mastodon at the other. And it’s very fluid. There are people who have only one or the other account, but a lot of the Twitter emigrés to Mastodon seem to have both. Some prefer Twitter, and are hoping for Twitter to survive its current turmoil, and carry on. Others prefer Mastodon in theory, at least, but haven’t really taken to it yet, or have had bad experiences there – or even terrible ones. And then there are major fans, enthusiastically building infrastructure, new communication ways, and networks.

Spectrum with a color gradient, Team all-inTwitter at one end with an off ramp. Team all-in Mastodon at the other end. Straddling in the middle.

For context, here’s a great 2-minute recap of the wild fortnight since the takeover – although it ends before the $8 verified parody accounts caused a colossal crisis, and the latest rounds of awfulness to Twitter’s staff and contractors:

The chart below mapped one aspect of the Mastodon migration against some of the events of the first week after Twitter’s takeover:

The source for the Mastodon data in that chart is here. It’s a bot posting every hour. I don’t think people creating new accounts is a metric that gives us a good idea of activity. [Update on November 16] Nemo_bis pointed out that this bot is counting instances that have opted in, not Mastodon-wide – while @fediverse@mastodont.cat “travels the graph more (but fluctuates more for the same reason).”

I don’t know where the flat line in that chart for Twitter comes from. Musk keeps tweeting that new users are flocking in and just about melting the servers. While that’s data-free hyperbole, I wouldn’t be surprised if it’s at least somewhat true. In my circles, only a tiny proportion of people are shutting down their Twitter accounts, and it makes sense that there’d be a flood of newcomers. It’s what many feared would happen: It’s the corollary, isn’t it?, of the flood in hate speech that followed the takeover. The nature of who’s active on Twitter may be changing radically.

Before we look at some more data, there some things to keep in mind about its limits. We don’t know what proportion of users are bots or people who have multiple accounts. That’s true for both Twitter and Mastodon. I dug into the best data I could find for Mastodon growth. Remember the proportions here, too: Twitter may have around 300 million active users per month.

Twitter usage was reportedly in decline before the Musk takeover. It’s usage by humans that matters, not the number of accounts. My own use of Twitter had been reducing, and I think that was widespread among people who don’t thrive on arguing and outrage, and are doomscrolled- and Covid-warred-out. I believe there’s a lot of quiet (semi-)quitting going on.

The best metrics I’ve seen for Mastodon growth are the number of active users per month, and the number of instances. Instances are the communities people join: Mastodon is what those instances are called as a collective. It’s part of the Fediverse – a federated universe of open source systems. When the Mastodon instances are connected to each other, you can navigate them all at once as though they are a single system. But instances are run separately, and can disconnect from each other. For example, an instance that’s a gateway to hate speech can be blocked off from other instances.

People’s experience of Mastodon is mediated by the instances they have joined. Mastodon’s help page says more effective methods for blocking than it already has are in development. And that’s clearly urgent. I’m going to post about the issue of moderation separately, because it’s such a critical issue for the potential of Mastodon, especially for the people most vulnerable to hate speech and harassment. How Mastodon shapes up on this will be determinative in many ways.

Back to the number of instances. That’s critical to Mastodon’s design: Responsibility for moderation and keeping the service running properly is distributed across the instances. So increasing the number of instances scales up Mastodon’s capacity.

There were well under 4,000 instances when Musk took over Twitter, and by November 13, the total wasn’t far off 6,000. There is fluctuation: Instances are shutting down as well as establishing. (You can migrate to another instance if yours shuts down, or you’re unhappy with it.) Active monthly users increased from below 400,000 when Musk took over Twitter, to over 1.5 million on the day I posted (November 13).

[Update November 16] Thomas Lumley pointed out that the growth in active users includes new arrivals, as well as a substantial contribution from people who already had accounts, but had been dormant. On November 12, Mastodon developer Eugen Rochko posted that there were a million more people using Mastodon then, than did before the takeover. On November 16, the number of active users during the previous month had grown to over 1.75 million.

[Update November 20 – still November 19 in many parts of the world] On Thursday, November 17 Musk issued his commit-to-“hardcore”-or-go-by-tomorrow ultimatum to the remaining half of Twitter’s employees; November 18, apparently close to half the remaining staff chose to go (it’s not known what proportion of the remainder had no choice, because of visas etc); November 19, Musk polls Twitter on whether to restore Trump’s account. Mastodon active users jumped from over 1.8 million on November 18, to over 2.2 million in 2 days. An updated chart from my Mastodon post shows the growth in the last 3 weeks:

The user part of the chart below, with additional data points. Latest data points are in the text of the post. The full chart up to a week earlier is in bullet point form below this post.

Here’s my original chart, showing the growth in active users against key Twitter events that could trigger Mastodon migrations, up to 13 November:

Chart of growth in monthly active users at Mastodon mapped against key events. It's included in bullet points below this best.
* The data on usage comes from a few data points that may not reflect the specific date that the level of usage was reached. The data for this chart and the number of instances are included below this post, including the sources used. Note: Twitter has particular obligations to the Federal Trade Commission (FTC), pursuant to privacy breaches in 2011.

There’s a non-zero chance that because of the wipeout of so many personnel, Twitter begins collapsing technically, gets hacked, and/or has a major privacy breach. It’s theoretically possible for it to be taken offline for legal or national security reasons. The financial situation is perilous too.

When major companies began baulking at advertising on Twitter, that was already a bellwether. If there’s too much reputational damage for them to risk being on the platform, then what about everyone else? And that was before things got really bad. It only took a couple of weeks. With each major new crisis at Twitter, we could see the risk of reputational damage and harassment/trolling increase, while the rate of desirable engagement decreases.

Cartoon of a female senior scientist giving a conference presentation, with a graph showing a rise and precipitous drop. She says "We call it the Twitter Law of Plunging Returns".

What if Twitter somehow does recover from all this? And could the science Mastodon migration lead to thriving a new social ecosystem? Mapping the Mastodon migration from the outside at the moment is tough. I don’t think we have a good handle on the size of science Twitter, either. Take for example 2 studies from 2017, when Twitter was already very well-established. I dug into the problems of one of them in this post: The estimate there was over 110,000 scientists on Twitter, substantially social scientists. Here’s another, based on trying to see how many authors of articles in the Web of Science could be matched reasonably well to Twitter accounts. The estimate in that one was over 330,000, with a minority in natural sciences.

What about Mastodon? One way to get an estimate is counting up the users in the Mastodon instances created for scientists. I think that you get a major under-estimate that way. It looks to me like most scientists are following the main Mastodon migration into the very big instances. Some who are vulnerable to attack have sought out instances where they have confidence in the strength of moderation, including identity-based instances. Others are going with regional instances, such as aus.social and mastodon.nz in my part of the world.

Accepting it’s an under-estimate, how many are in the science-specialty instances? I started by picking relevant ones from Fediscience’s list, leaning to life sciences and science communication. Then I added ones that were missing, but I see often – like the ones for the biological ecology/evolution and genomic communities. I found 12 instances, though there’s likely more – especially in non-English-speaking academic networks, and more in the social sciences and general academia. Those 12 have over 33,000 users (see my Google spreadsheet). That’s….a lot, in such a short time.

I don’t know how many of those people are active, but science Mastodon looks like it’s booming and thriving to me. I saw someone post that one day it suddenly felt like a science party, and I think that’s a good description. Within about a 48-hour period, a critical mass of scientists had gotten their bearings and Mastodon brio, and were finding each other. Even if Twitter pulls itself out of what looks like a possible death spiral, I can’t see us all ever going back. As Casey Newton wrote, Musk’s takeover has been “brutish and poorly planned.” This level of smashing and burning doesn’t just lead to damage. It’s destructive – not just for a tech company, but for the communities, activist systems, and livelihoods being shattered around it.

Meanwhile, over at the elephant-y place, people have formed new communities with lightning speed, collaborating on new ideas for how to connect and build. Science journalists are there, and the number of journals is slowly growing, though I haven’t seen a list. Some universities have set up instances.

For me, after just a couple of weeks, my Mastodon feed is so much more interesting that I’ve stopped keeping up with my Twitter one. The information and conversations are great, with familiar and new people. It’s broadening my mind at a rapid clip, and it’s far more fun. It has energized me enough to face the daunting challenges that lie ahead. As each day goes by, the owner’s attacks on journalism and the people who work for him push me further away from the birdsite. And the open source, non-profit elephant pulls me closer.

~~~

You can subscribe to my (free) newsletter to keep up with my work.

Want to read more about all this? From me, there’s my November 3 post and Casey Newton’s post on the Twitter meltdown, both of which I’ve linked to above. And here’s half a dozen more:

  • Sherrilyn Ifill on the role of Twitter in Black Lives Matter, Musk’s takeover, and a moment of truth for big business: On Twitter’s End.
  • From Team ElleBlack Twitter’s Mass Exodus: The Community Will Find Another Way.
  • Home Invasion: Hugh Rundle on grief about how the Twitter tsunami ends a phase for Mastodon’s communities – “All of us lost something this week. It’s OK to mourn it.”
  • On grief for the dying of the revolutionary internet – though I think this might usher in something far better: Paul Ford, at WIRED.
  • A detailed reporting of Musk’s takeover of Twitter – a New York Times report of what happened within the company, by Kate Conger, Mike Isaac, Ryan Mac, and Tiffany Hsu.
  • Clive Thompson on Mastodon’s antiviral design, and why some friction that slows things down could avoid some of the harm other social media has caused.

Disclosures: My Twitter followers peaked a few hundred above 32,000, and were down to just under 31,800 as of writing. I joined Mastodon on October 31. I am not aware of PLOS policy or plans, though PLOS Biology joined Mastodon on November 8.

Update November 16: Two paragraphs on Mastodon data were updated (both noted inline). My thanks to @Nemobis@mamot.fr and Thomas Lumley (@tslumley@wandering.shop) for the extra information.

Update November 20 (Australian morning): Updated with the next big jump in users.

Images:

Chart on Mastodon growth* in bullet points:

  • October 27: Mastodon had 328,293 users in the previous month, and 3,676 instances.
  • October 27: Musk acquired Twitter.
  • October 31: Paid blue check plan announced.
  • November 4: Half of Twitter’s workforce fired.
  • November 8: Mastodon had 1,036,362 users in the previous month, and 3,612 instances.
  • November 9: $8 blue check rollout.
  • November 10: Mastodon had 1,146,216 users in the previous month, and 4,710 instances.
  • November 10: Twitter’s security and privacy executives resign; FTC expresses concern.
  • November 11: $8 blue checks halted; Musk reportedly tells staff bankruptcy not off the table.
  • November 12: Mastodon had 1,460,464 users in the previous month, and 5,874 instances.
  • November 13: Mastodon had 1,582,751 users in the previous month, and 5,779 instances.

* The data on usage comes from a few data points that may not reflect the specific date that the level of usage was reached. The data for this chart and the number of instances are included below this post, including the sources used. Note: Twitter has particular obligations to the Federal Trade Commission (FTC), pursuant to privacy breaches in 2011.

Sources for Mastodon growth and timeline:

Leave a Reply

Your email address will not be published. Required fields are marked *


Add your ORCID here. (e.g. 0000-0002-7299-680X)

Related Posts
Back to top