Social Media Regeneration and Divisiveness

We're going in different directions, and we're starting to see some consequences.

You know how people used to say reformed smokers were the worst? The most annoying non-smokers of all? I think I might be that now, but for Twitter. I’m in my social media happy place on Mastodon, along with many other Twitter-quitters. And this week, when I saw who is still tweeting away as though nothing’s changed, I was quite taken aback.

Dave Karpf wrote that he’s on “Twitter Tumbleweed Watch”. He’s convinced his part of Twitter “is becoming a ghost town”. That’s based on how few people generally see his tweets now, and thus, how little engagement there is with them – in stark contrast to how much engagement is going on at Mastodon. That matches my experience, and there are lots of recovering tweeters saying the same, though some say the reverse. When I dug around a bit in Covid Twitter, the picture got complicated quickly.

The divisive, sensationalizing people in Covid Twitter are still getting a lot of engagement, whatever side of the divides they’re on. Hmm. The people who like helping Twitter’s bellows pump air onto every ember of anxiety or outrage are still there in droves. Maybe the people who hated that are more likely to have left, and Twitter’s getting ever more divisive? That said, when I looked up some of the non-sensationalizing people from Covid Twitter with huge accounts, they were still getting a lot of engagement, too. Hmmmmm.

The Covid-tweeting people I was looking at had blue checks. Dave Karpf and I, on the other hand, do not. That figures. Twitter is increasingly becoming pay-to-play. Some of the people who are saying the “audience” is disappearing at Twitter do have blue checks, though. (Here’s one that popped up while I was writing this.) So there’s a combination of factors feeding into change (or the lack of it) – how much of a specific community has drifted away from Twitter, how many of them have regenerated an active community on Mastodon, and Twitter’s evolving priorities for tweet visibility.

We might see how big a factor the blue check is in all this soon. Musk recently tweeted all the blue checks will be taken away from everyone who doesn’t pay for them in a few months – and with that, the current high algorithm status of their tweets. If he really does it, and tumbleweeds roll into current power users’ accounts, it could push many Mastodon’s way. Especially with the pull of the recent arrival of Ivory – Tweetbot, but for Mastodon, with the developers now free of Twitter’s constraints on innovation, with new possibilities and constraints that are socially-oriented. As Dwight Silverman wrote, once Musk cut off the independent app developers, many turned to Mastodon, where “new life blooms….[A] new social media world order is being born.”

Meanwhile, Twitter is getting nastier and thus, surely more divisive, even if all users aren’t aware of it. It reminds me of the early days of Gamergate and Breitbart, when so many people didn’t seem alarmed at the rise of the “alt right.”

Some research is trickling out now, and it’s an ugly sight: We’re watching influential social media rapidly, and deliberately, tilting antisocial. We don’t know how much damage that’s going to do, to individuals within it, and to societies, if it keeps going down this path. The stream of studies measuring the impact has begun – and there are sure to be studies, too, of activism and other measures to counter and limit the harm this new Twitter has the potential to do.

I’ve started a tracking post to gather studies and data – I’ve found 5 studies you can dig into there to start:

• 3 studies show an explosion of hate speech in the immediate aftermath of the takeover. For example, the use of some hate speech went from around 80 an hour to well over 4,500 an hour. Spikes in hate speech online – not just at Twitter – may coincide with spikes in hate crimes in real life. The authors of one of the studies concluded their results were consistent with what a “less stringently moderated Twitter would look like.”

• Twitter accounts associated with far right networks were among those grabbing blue checks in the first (ill-fated) round of sales in November 2022. A sample of 961 of them had a 70% increase in retweets after Musk’s takeover, whereas a sample of 933 that weren’t associated with far right networks had a 28% increase. (No data on overall engagement rates at Twitter.)

• A large survey run periodically, sampling well over 20,000 Americans each time, suggests that there could have been an overall drop of about 3 percentage points in people saying they’re using Twitter, between October 2022 and December/January – which could be 10% of individual users. If it’s indicative, then the balance of Twitter users shifted towards the political right. Democrats had been around 37-38% at 3 earlier time points in 2022, but 33% after the Musk takeover. There was no statistically significant difference in the number of independents and Republicans in the post-takeover survey compared to the previous one.

With Trump soon to return, as FrameLab points out, it’s going to get worse: “Ultimately, it will prove impossible to beat Musk by remaining on the platform he controls. He and his goons control the most powerful aspects of the site.” FrameLab’s post is a good read if you’re on Twitter and wondering about what to do.

Over at Mastodon, on the other hand, there was a major surge in activity after Musk booted a bunch of journalists off Twitter, and it’s been very lively since. A critical mass of some communities have relocated – like software developers and related academics. A lot like the early days of the internet, and who was first on the “information superhighway.”

Most communities haven’t quit Twitter and re-formed on Mastodon, though. Even if a lot of people from particular interests opened Mastodon accounts, comparatively few are actually using them. There are some terrific people, though, across all my areas of interest. And I’ve picked up some new interests. I thought I’d miss Twitter a lot, but I don’t. (If you’re interested in dipping your toes in the water, here are my shortcuts.)

Speaking of where people’s attention is going, I had a reminder recently of how many people use Wikipedia – and something hit home that, just as with social media, comparatively few of us would be classified as “power users” in terms of producing Wikipedia’s content.

A Google Doodle of the day featuring a woman scientist, Marie Tharp, inspired Cassidy Villeneuve to write a cool blog post about Tharp, and about her Wikipedia page, too. It’s a fascinating story. On the day of the Google Doodle, she writes, more than 100,000 read Tharp’s English Wikipedia page.

That got me wondering how many page views the 38 Wikipedia bio pages I created have had. Some were just small exercises at Wikipedia edit-a-thons where I was helping people learn to edit Wikipedia. And I’ve done a lot more work at many bios I didn’t create. It’s quick and easy, though, to check out the ones I created. They’ve now had over half a million views on English Wikipedia alone. Most are of people of color (30 of them). Around half had more than 10,000 views each, and those had all been translated into at least one language as well – one into 16 languages.

Given that I am generally picking out lesser-known scientists, mostly women of color, this made me feel great! An election going on around Wikipedia governance, though, surprised me. When another Wikipedian pointed out how few of us met the eligibility criteria to vote, I found it hard to believe, so I checked it out. Turns out, I’m in the top 0.1% of editors on English Wikipedia – there are only 45,000 of us in that tier. I didn’t even think I was a particularly prolific editor! Hmm, again. I think I should do more to get people to see how gratifying contributing to Wikipedia can be.

It’s January, so it was time again for my annual round-up of peer review research – and it was a bumper year for randomized trials of peer review at journals. That’s below my sign-off, along with some more metascience, an interesting look at eye charts to test reading – plus the Great Pyramid.

I hope you have a good week!

Hilda

• In 2019, I wrote a 2-part series at my PLOS blog, Absolutely Maybe, on what we’d learned from research on peer review at journals to that point. And each year since, I’ve written an annual roundup of my pick of research that advanced our knowledge in the previous year. This week, I posted the research roundup for 2022. I think it was an all-time record for the number of randomized trials of editorial peer review in a year – 5 in all. We usually can’t even count on a single one appearing. So there’s a bit to digest. And you can see all the previous posts with this tag.

• And here again is the post I’ve started for research on Twitter in the Musk era. With links to my other relevant posts.

• More evidence of poor research reporting, unfortunately. A study found that about half of journal publications of trial protocols didn't report on monitoring plans - even though that's required by the SPIRIT Statement, and even though most of the journals involved had endorsed that Statement. A measly 8% of the protocols reported all 3 aspects of monitoring plans (scope, frequency, organization). This study was in 811 protocol papers from selected journals in the first months of 2020. (On Mastodon.)

• The eye test we probably all think of is a chart on the wall with rows of letters, called a Snellen chart. That tests our visual acuity – how precisely we can pick out tiny details. There are tests of reading, too, with short paragraphs in different font sizes, for example. A new review dug into them. The most commonly used reading tests, it turns out, aren’t validated and they vary a lot. Digital ones are set to take over, though, measuring reading time and recording reading aloud to enable more accurate assessment of errors.

• And on the subject of vision, new rule! Text-to-speech software can confuse people when “m” is used for “million” (as in, 4m) – because it’s the same as metres. Thus, the better abbreviation for million is “mn” (and “bn” for billion, “tn” for trillion). (Via Emma Dingle on Mastodon.)

• Finally, something completely different, from Saloni Dattoni’s latest newsletter. It’s a collection of discussions about interesting research results, and there’s also this gem. You can go on a virtual tour inside the Great Pyramid at Giza. (It’s a project from a university in Boston.)