Reflecting on Twitter, White Flight, & “Quote Tweet” Tensions at Mastodon

Why can’t we “quote boost” at Mastodon, the way we can “quote tweet” at Twitter? That might sound like an innocuous question, but it’s a volatile one. Quote tweets let you tag someone else’s tweet with your own addition above it, as a new tweet from your account. Both the original and your add-on are in full display.

It’s used in many ways. For example, you can add a hashtag that the original poster missed – as I did here to create an example for this post. The account that posted the original tweet gets notified. After a quote tweet, some people retweet it, and others retweet the original tweet.

Super-simplified, here are the main strands of discussions about quote boosting….

• We miss it: It was so useful and easy, and we did it a lot – Mastodon is so clunky without it…
• …Ah, but see – the friction is a feature. Extra clicks for you to be able to quote a post, and then another click for someone to read that quoted post, slow things down: That’s part of Mastodon’s overall anti-virality design, to avoid Twitter’s toxicity…
• …The quote tweet in particular is very late-stage Twitter – the enemy of thoughtful conversation. It encourages talking about people behind their backs, instead of replying to them…
• …It’s also a cyberbullying enabler – a tool for pile-ons…
• …But the quote tweet is integral to Black Twitter culture and activism, and the absence of this affordance marks Mastodon as designed by, and for, whites: Even though it’s not intended to be exclusionary, that’s its impact.

For context, let’s go for a walk through Twitter’s past, and the evolution of quote tweets as an affordance for digital action, as well as Twitter activism and Black Twitter.

UPDATE January 12, 2023: I had placed a few study results in the timeline at the time of the usage data analyzed. However, since then I have done a much deeper dive into the literature, and have written another post digging into the results of over 30 studies of quote tweets. That strengthens the conclusions at the end of this post.

2004

Tad Hirsch from MIT developed TXTMob, an open source text messaging service based on bulletin board software, for protestors at the 2004 Republican National Convention: It was the main precursor/inspiration for Twitter.

2006

Twitter was launched. It was called Twtter (inspired by Flickr), and it was, in essence, a free text messaging service.

Tarana Burke started the MeToo campaign on MySpace.

2007

Users developed the retweet practice, initially called an echo. It was manual: You cut and pasted someone’s tweet, and started it with RT, followed by @username. Sometimes the original tweet would need to be crunched to fit the 140-character limit – by changing “would” to “wld”, for example: People started calling those MTs (modified tweets). For both RTs and MTs, if you had enough character spaces left, you could add something of your own. Somewhere along the line, QT for quote tweet was also used to indicate using someone else’s tweet in yours.

The first RT is believed to be a tweet about social media being social narcissism.

Users developed the hashtag that year as well – you had to use a search engine outside Twitter to find other tweets using it. The first proposal for hashtagging was by Chris Messina in August – he was inspired by its use as a label in IRC (Internet Relay Chat). It started taking off when Messina urged a friend to use #SanDiegoFire for his frequent updates on local fires.

Anil Dash blogged about jokes in an African-American tradition taking off on Twitter, as part of the social media participation in the US presidential election: It was the year Barack Obama first ran for, and won, the presidency, with a strong social media campaigning element.

André Brock later pointed to Dash’s post as the start of discussion about the phenomenon later known as Black Twitter. Culturally specific hashtags drove the development of a vibrant and distinct community, which would power up the daily hashtag leaderboard: “Black Twitter hashtag domination of the Trending Topics,” he wrote, “allowed outsiders to view Black discourse that was (and still is) unconcerned with the mainstream gaze.”

2009

The RT reportedly surged during what was called “The Twitter Revolution” in Iran. (Note: That was from Wikipedia, but without a citation to support the statement.) Iranians used proxy servers to bypass government blocking of social media, but organizing on the ground was mostly done by text message, email, and blogging – Twitter was how the world outside Iran followed what was going on.

Twitter began rolling out their retweet feature. Fake RTs had been an issue for Twitter.

Twitter began hyperlinking hashtags in July.

That year, danah boyd pointed out that there was a massive white digital flight from MySpace to Facebook, though African-Americans were still very much present there: “[W]hen people are structurally divided, they do not share space with one another and they do not communicate with one another,” she said. “This can and does breed intolerance… Although most of you call these sites ‘social networking sites,’ there’s almost no networking going on. People use these sites to connect to the people they know… People connect to people who think like them and they think like the people with whom they are connected. The digital publics that unfold highlight and reinforce structural divisions.” 

A survey of Twitter use by Pew Research a couple of months later included 19% average Twitter use in the US, with 19% for Caucasians and 26% for non-Hispanic African-Americans. There were only 1,698 internet users in that survey, 19% of whom used Twitter – so just over 320 people, which is a very small group of people. The authors did not tag the racial variation as significantly different.

The phenomenon of Black Twitter became a topic of public conversation in white America, for example, in Choire Sicha’s post on hilarious trending topics: Twitter’s trending topics bar, he argued, was able to break down the characteristic “stupid little bubble” we create for ourselves in our online social networks.

2010

The Arab spring started – popular uprisings that began in Tunisia and spread. It was fueled, in part, across social media – especially Facebook. By the middle of 2012, the Arab spring had been suppressed.

2011

Twitter reached its first 100 million users.

2012

Twitter reached 200 million users.

Pew Research ran their survey again at the end of the year, this time with 1,802 internet users, 16% of whom used Twitter. So again, a small group to break down by race. The rate of Twitter use by non-Hispanic African-Americans was 26% again; for non-Hispanic Whites, it was 14%.

2013

The Black Lives Matter movement began after the acquittal of Trayvon Martin’s killer. Candice LaShara Edrington and Nicole Lee write that it originated on Facebook with Alicia Garza and Patrice Cullors. With Ayọ Tometi, they migrated BLM to Twitter in July.

Feminista Jones wrote in Salon that while Twitter was a source of news and entertainment to others, for people of color, Twitter had “become one of the most important tools of modern sociopolitical activism…”

African-Americans have historically relied on “alternative” communication styles and underground means to connect and build networks. Centuries ago, newly captured African slaves were separated from those who spoke their native languages to discourage organized attempts at escape. For them, finding universal means of communication, like “spirituals” or “work songs,” became essential to their survival and that of future generations. We have often used grass-roots communication to organize and mobilize efforts to achieve freedom and equality.

Feminista Jones, 2013

2014

The hashtag #BlackLivesMatter helped propel a social movement after the killing of Michael Brown in Ferguson in 2014 (Edrington and Lee 2018).

The Gamergate campaign of online harassment on Twitter took flight, organized from 4Chan and reddit. It went on for about 2 years. It was the ability to create accounts in multiple identities that enabled it: You just couldn’t block the swarms of them fast enough to get out from under them. (Disclosure: That happened to me during a Gamergate pile-on in 2015.)

The Iranian #MyStealthyFreedom campaign protesting against mandatory hijab began on Facebook, and then Twitter.

2015 ⭐ Quote tweet introduced

The affordance for quote tweeting was introduced in April. From then, if you hit on the retweet symbol, you had 2 options: Continue to retweet, or start your own new tweet embedding that tweet – with a fresh character limit.

A former senior Twitter employee, Jon Bell, has posted on Mastodon that it did not increase abuse. But he didn’t explain what he meant by “abuse”. For example, were they only looking for the use of a particular set of terms of abuse, or a wide variety of possible negatives uses, or use as an attack vector? There was no reference to a publicly available source of data, and we don’t know what timeframe was meant: Was this for just a short time after its introduction, before usage patterns developed?

The introduction of the quote tweet happened around the time the US presidential campaign was getting into swing, driving division and aggression: Introducing the affordance into another context might work differently.

2016

The divisive US presidential election ended in Trump’s win in November.

2017

Tweet character limit increased from 140 to 280 in November.

Alyssa Milano kicked off a massive Twitter surge of #MeToo.

2018

The number of retweets began to outnumber original messages, according to an analysis of over 118 billion tweets from 2009 to 2020, in 150 languages, by Thayer Alshaabi and colleagues. They counted quote tweets as original messages, which makes it hard to know the contribution of the practice.

Layla Hashemi analyzed over 15 million tweets in 2018 and 2019 on #MyStealthyFreedom and related hashtags in English and Farsi/Persian, along with interviews. She estimated about 75% of the Persian tweets came from within Iran. Hashemi pointed out that quote tweets increased the chances of retweets.

Pew Research ran its survey again in January. This time, they reported no real difference between the rate of use of Twitter by Black and White Americans (26% vs 24%).

2019

In May, quote tweets were rebranded, “Retweet with Comment”, and you could attach media. I couldn’t find an explanation of why.

2020

There was an unprecedented roar of anger and grief across Twitter in the US, when the video of George Floyd’s murder went viral in May, helping fuel massive protests. On Twitter, it was driven by retweets – shown by a study where again, quote tweets were counted as original messages.

Pew Research conducted in June found that 72% of Black American social media users reported that they had used their account for some sort of political activism in the previous month. Both Black and Hispanic users were more likely to say social media was personally important to them for political activities than White users – there was a gap of 20+ percentage points between White people and people of color on relevant questions.

The biggest study I could find on use of quote tweets analyzed 800,000 Americans who tweeted in the first 9 months of 2020, matching voter records to named Twitter accounts. Those researchers didn’t report an important variation in the rate of use between African-Americans and Caucasians. African-Americans were a bit more likely to use quote tweets than Caucasians, though. African-Americans tweeted 9.1% of all tweets, but 10.0% of quote tweets. Caucasians tweeted 83.7% of all tweets, and a similar proportion of retweets and quote tweets. However, their share rose when it came to replies: 86.6%. Whereas, African-Americans’ use of replies dropped to 7.5%. I think that’s a very telling marker of a different experience of Twitter.

In September, Retweet with Comment was rebranded – back to quote tweet.

In October, Twitter tested prompting users to quote tweet when they hit the retweet symbol. In December, they reverted to the original format. Here’s the detail they provided on this:

We encouraged people to add their own commentary when amplifying content by prompting Quote Tweets instead of Retweets. This change introduced some friction, and gave people an extra moment to consider why and what they were adding to the conversation. Since making this change, we observed a 23% decrease in Retweets and a 26% increase in Quote Tweets, but on a net basis the overall number of Retweets and Quote Tweets combined decreased by 20%. In short, this change slowed the spread of misleading information by virtue of an overall reduction in the amount of sharing on the service. We are taking more time to study and fully understand the impact of this change and are leaving it in-place for now.

Updated on December 16, 2020: We’ll no longer be prompting Quote Tweets, and are re-enabling the standard Retweet behavior. We hoped this change would encourage thoughtful amplification and also increase the likelihood that people would add their own thoughts, reactions and perspectives to the conversation. However, we observed that prompting Quote Tweets didn’t appear to increase context: 45% of additional Quote Tweets included just a single word and 70% contained less than 25 characters.

Twitter blog

2021

Pew’s January-February survey of social media use reported 29% use of Twitter by Black Americans, and 22% by White.

Pew Research also concluded that the severity of online harassment had increased since 2017, singling out targeting of people based on their gender, race, and sexual orientation.

2022

Pew Research conducted at the end of April found that 71% of Black Americans and 61% of English-speaking Hispanic and Asian Americans agreed that calling people out for potentially offensive social media content holds them accountable: That was only 44% for White Americans. The opposite position was saying it punishes people who don’t deserve it: That position was held by 26% of Black Americans vs 51% of White Americans, and 38% of women vs 52% of men.

And finally, an example from China that illustrates the changed scenario at Twitter for activism. A September headline in the South China Morning Post: “Chinese social media users are flocking to the decentralised Mastodon platform to find community amid crackdown at home.” It’s a small flock, for China: They were pointing to about 50,000 people. The platform, they said, is stopping people being harassed by users loyal to the Chinese Communist Party.

This month, Twitter’s catastrophically reduced human rights and security personnel struggled for hours to contain a concerted tsunami of porn links from Chinese accounts spamming searches and hashtags that could otherwise surface news of Covid lockdown protests in Chinese cities. The campaign, according to a report in The Washington Post, was spotted from outside Twitter: An ex-employee said that all the personnel for Chinese influence operations were among those who had resigned from Twitter recently.

Which brings us to the current debate around quote tweets at Mastodon…

A note first on Mastodon. It’s a decentralized social network that began in 2016, but this type of system has been around since at least 2007. Because it’s decentralized, Mastodon has grown into thousands of “instances” – like thousands of big and little “Twitters” flying in close formation on shared open source software. Instances can block other instances – for example, instances that allow hate speech, or moderate inadequately, can be blocked not just by individuals, but by whole instances and even all of Mastodon. So the experience of being on Mastodon depends somewhat on your instance.

Mastodon’s originator has the last word, and in 2018, he posted that he was categorically against quote boosts. I don’t know enough about Mastodon’s history to know why it isn’t standard, but other Twitter features that drive problematic behavior are. In the table below I compare relevant Twitter and Mastodon features based on those highlighted in a review of the literature by William Brady and colleagues in 2020. Those authors argued that speed of interactions might encourage faster, more emotional, reactions, as well as other factors.

Features that do, or might, drive negative behavior

(as suggested by Brady et al 2020, plus *)

* Twitter reports this feature successfully slows down reflexive retweeting

Feature	Twitter	Mastodon
Notifying each engagement in real time	Can be switched off.	Can be switched off.
Showing number of engagements on posts	Prominent.	Behind a click, but presumably, instances blocking other instances affects the visible tally; numbers of engagements shown in trending posts (the Explore button – see below).
Showing number of followers	Yes.	Yes (on profile only, not hover).
Profit-driven algorithm pushing engagement	Yes.	No.
Trending topics	Yes.	Yes (the Explore button), but it’s not displayed on all instances & apps.
Speed	Heightened (except *).	Some deliberate points of friction.
* “Have you read [linked article]?” prompt before retweeting	Yes.	No.
Re-blog	Yes (called retweet).	Yes (called boost).
“Quote tweet”	Yes.	Not standard. Instances vary in how much of a linked post they show, from no preview to a display that looks much like a quote tweet. “Copy link to post” is one of the options in the menu under 3 dots on a post (to the right of the bookmark symbol). If the instance of the linked post is blocked by your instance, you wouldn’t see what the quoter is talking about, unless they had a screenshot. A screenshot would be inaccessible to people using screen readers if it’s not reproduced in full in the alt text.
Notification of “quote tweets”	Yes.	No, because it’s just a link – addition of your full @username is required for a notification.

Without knowing what lay behind these decisions, the result feels a bit arbitrary – especially the situation with quoting posts. The challenges of moderation in volunteer-run services clearly makes prevention of potentially abusive behavior like pile-ons quite critical. But as Gamergate showed, that’s a risk even without quote tweets.

Here’s what some of the current inconsistency looks like on my instance, which is one of the large ones run by Mastodon’s founder (Mastodon.online). This is a pair of test posts, since deleted, from my Mastodon account. The top one shows what it’s like when I link to one of my Mastodon posts – it’s restricted to the link itself, which is less than you usually get when you link to something. In the second one, which is a link to one of my tweets, you get the normal linking experience – the link itself, and a little preview. (Using the “Copy link” facility is similar to a link to the tweet.)

Given that Mastodon posts can be much longer than tweets, a quote boost couldn’t be identical to a quote tweet. You can click here to see an example of what a fuller display could look like at a Japanese Mastodon instance that has this enabled. The quoted Mastodon post is truncated, and the link that’s part of the original Mastodon post is shown with its own preview.

So where have I landed after all this? I’m not convinced that the fears about developing quote boosting in some form are justified, given the different context. Fortunately, though, our only options in the face of differences of opinion aren’t to pull up the drawbridge or let it rip, based solely on opinions.

From what I’ve read in github discussions on potentially developing a quote boost affordance, though, it doesn’t look as though data, formal experiments, and wide community participation have played a major role. I must have missed critical artefacts from Mastodon’s history, so I could be wildly off the mark. However, given the previous size and resource base for Mastodon, if my impression is right, it would be understandable. But the future could be different.

The Pew Research I pointed to above suggests there are major schisms along race and gender lines on what’s a social media pile-on, and what’s calling out that’s necessary for accountability and social progress. And when it comes to race, there have been great experiences at Mastodon, but awful ones too. It adds to a history of problems for people of color at Mastodon, discussed recently in a good short intro at Fast Company.

Johnathan Flowers points out there, that “To the extent that a space’s social organization holds whiteness center, holds white comfort center, holds the norms, values, ideals of whiteness center, that space remains a white supremacist space.” Marcia X said that it’s not only direct abuse, “people are having to defend their positions on racism or sexism (for example) and deal with constant back and forths on basics. That may not be classified as abuse, but it quickly spirals into whiteness dominating conversations.”

That experience – and there’s a lot of it – harks back to André Brock’s 2012 analysis of Twitter: “This public space is constructed by intense monitoring of non-White speakers.” He emphasized a quote from Rawls: “display of moral behavior by members of one group may well look like deviant behavior to members of the other.” (Brock has since authored a book, Distributed Blackness: African American Cybercultures.)

We have to consider the power imbalances here with the serious urgency they deserve. As Flowers articulated in another interview, encountering a space like this is “like going into a predominantly white work environment and recognizing that there are some elements of you that you have to leave at the door. That feeling, that sense of whiteness as a felt space is off-putting to many users.”

The principles from the Design Justice Network are one of the good places to start to move the quote boost question forward. Those principles begin with the need to seek liberation from oppressive systems, and include prioritizing “impact on the community over the intentions of the designer.” To me as a scientist, a process guided by those principles would incorporate subjecting theory to the test, as well as approaching an issue collaboratively.

However this unfolds, as a white person myself, the issue of white behavior looms large. The odds faced by people of color of “whiteness dominating conversations” are bad enough when we dominate these spaces just by sheer numbers. Add to that our greater propensity to insert ourselves into conversations, as shown earlier, and it’s a ghastly picture. If we keep choosing insensitivity and our own comfort, we’re choosing to be part of the problem. We can choose another path.

This is the third post I have written on the Twitter/Mastodon migration. Future posts on this subject at this blog will be tagged Mastodon:

1. On November 3, 2022, at Living With Evidence: Shuffling Communities and Twitter Migration.
2. On November 13, here on Absolutely Maybe: Mapping the Mastodon Migration: Is It a One-Way Trip or an Each-Way Bet for Science Twitter?
3. [This post on December 1.]
4. On December 2, at Living With Evidence: The Relief of Leaving Twitter.

If you know of more studies on the use and/or effects of quote tweets in particular, I’d be grateful if you could let me know either on Mastodon – address below – or in a comment to this post. I moderate all comments on this blog, so you can let me know if your comment is for my eyes only and I won’t release it.

~~~~

I’m @hildabast@mastodon.online at Mastodon. You can keep up with my work at my newsletter, Living With Evidence.

Disclosure: I joined Twitter in October 2010, and Mastodon in October 2022. At the time of writing this post, I had just under 3,000 followers at Mastodon (growing), and over 31,000 at Twitter (dropping).

Images:

• The drawbridge image – “Drawbridge as first line of defense against attackers” – is by Pearson Scott Forseman, public domain, via Wikimedia Commons.
• The cartoon is my own (CC BY-NC-ND license). (More cartoons at Statistically Funny.)

Update December 29, 2022: Link to Jon Bell’s thread was broken – replaced with archived link.