Killing the retweet, and other ways to fix Twitter

Twitter is caught between two product visions. Which should it choose?

Imagine you’re building a car, and in order to test people’s enjoyment of it you measure how often they interact with all the buttons. I get in to drive it and I immediately feel frustrated. Gear changes are clunky, acceleration is sluggish, the traffic is a nightmare. I hammer the horn in frustration. BEEEP - aha, a successful piece of user engagement! BEEEP - more user engagement! BEEEEEP - look how much this user is engaging with our product!

This is not unlike the retweet button on Twitter. I see Katie Hopkins, I am angry with Katie Hopkins, I hammer the button and write “Katie Hopkins sucks.” In the real world, I’m expressing my displeasure with Katie Hopkins in the hope that some all-seeing deity will hear my cries and respond. And God is listening: they’re meticulously cataloguing every click of my mouse as a successful interaction with one of Katie Hopkins’ tweets, and updating their algorithms accordingly.

Last week, I talked about the role of Twitter, the public and the wider media ecosystem in ‘creating’ villains through a perverse system of economic incentives that allow people to efficiently monetise rage. This week, I want to focus a bit more on the solution. If you’re in charge of Twitter then how, assuming ‘delete Twitter’ isn’t a viable option, do you break the chain?


The starting point, as with any product, has to be Twitter’s vision. What is Twitter actually for? A trite answer might be ‘making money’ or ‘getting lots of users’, but these aren’t useful visions to build a product around. They don’t help you decide which features to prioritise and which to discard. For this, you need a coherent idea of who your users are, and what they’re trying to achieve.

Twitter’s creation sparked a years-long debate (even among its founders) about what it was for. Some saw it as a microblogging platform, a place for people to broadcast status updates to an audience. Others saw it as a venue for conversation, a global village green where communities of users could congregate and talk to each other.

Whether features make sense or not depends which camp you fall into. A search function works for a broadcast medium, but sounds awful - almost creepy - added to a conversation platform. Mentions and replies are essential features if you’re talking among friends, but unmanageable if you’re broadcasting to millions.

The Quartz ‘museum exhibit’ article I referenced in my last post suggested that this conflict was resolved in favour of conversation, but I disagree. Even a casual glance at the evolution of Twitter’s feature set shows this isn’t true - or if it was, nobody told the designers. None of the original features were discarded, and new ones have been added in support of both visions over the years.

The result was an accumulated jumble of features which worked for and against each vision, but succeeded - for a while at least - if you focused purely on short term user growth. It didn’t matter if users were happy, sad, frustrated, angry, as long as they kept turning up each day and engaging with content. Except that eventually users began to question why they felt so damned unhealthy.

The first step toward fixing Twitter is to resolve, once and for all, the question of what it’s for. To do that, we need to understand where the value is… and perhaps more importantly, where it isn’t.

The value of Twitter is not casual conversation. There are far better places for people to talk to each other, where your comments won’t lead to global notoriety, or get you fired. It’s not even clear that it helps Twitter. The traffic generated from any given conversation is tiny. It’s not particularly attractive to the advertisers who account for over 86% of the company’s revenue. It costs money to support. It’s striking that Twitter squeaked into profit last year even as its user base declined.

Twitter’s real value comes from the same place as YouTube’s, Instagram’s and Spotify’s - quality content creators. These are the people and brands who draw users to the platform. They generate, broadly-speaking, the best content with the most engagement. They create newsworthy content that attracts the attention of the wider media. When you see Twitter in the news, nine times out of ten it was one of these users who put it there.

Losing five million casual monthly users would make little difference to Twitter’s bottom line, or the experience of most users. But losing fifty thousand of the best, most prolific content creators would make the site less relevant to people’s lives. That would impact Twitter’s most important metric - monetizable daily active usage. It would be less valuable.

In other words: Twitter needs to go back to the future, back to its original vision as a place for people to share content. Maybe not ‘microblogging’ as such, but something not a million miles away from it.

There are two related areas where I’d look to move the needle: improving the quality of content, and giving creators more control over their content once published. The following are all hypotheses - I have no idea whether they’d work or not, but they would be interesting things to test with users


People like to blame anonymity for low quality content on Twitter, citing the infamous ‘egg’ accounts with names like ‘John9310137’, but a casual inspection of other content-driven websites shows that many other support anonymity without the same problems. A more likely culprit is effort: Twitter’s features are designed to reduce the time between someone thinking something and then blurting it out to a large, random audience. Combined with the natural desire to be the first to say something, it lubricates conversation the way alcohol lubricates a noisy crowd, lowering inhibitions, encouraging people to react first and think later.

One solution might be to introduce more ‘friction’ - in other words, make posting a slower, more thoughtful process. This can work well on sites like TikTok or Instagram, where it takes a little more time and effort to craft a photo or video clip to share. Facebook and LinkedIn provide a more psychological friction: the knowledge that friends, family or colleagues may be watching encourages a greater level of self-awareness.

On Twitter, tighter character limits could force people to think more before posting content - the 140 character limit of old spurred a lot of creativity as people tried to express complex thoughts in limited space. Minimum character limits would be an interesting experiment, perhaps eliminating a lot of the spammy one- or two-word responses that basically amount to an upvote or downvote.

A more radical approach would be to change the nature of content on the platform. People have long joked about the idea of having Twitter read your tweets back to you, to check if you really want to send them, but voice tweets are set to become a real feature in a matter of weeks. With a generous, 140-second time limit, these have the potential to disrupt the podcast industry in intriguing ways. They also represent a greater investment on the part of content creators, and so in theory could be strong signals of quality.

But the best way to increase friction might be to eliminate the retweet.

Ditching quote-tweeting is an easy sell. It’s the source of a huge amount of spam on Twitter. It means that hundreds of people can dump the same viral tweet into your timeline with versions of the same pithy comment or ‘hilarious’ joke. It allows people with millions of followers to share out-of-context snippets of conversation or commentary to vast audiences, burying the original tweeter in an avalanche of often-abusive reactions that render the site unusable for days. Quote-tweeting enables the performative hating of content in a way that drives negative behaviour across the site. It is completely unnecessary when replies exist to support the same engagement in a more manageable way.

But the retweet?

Retweets are to fake news what mosquitoes are to malaria. They are so low quality, so spammy, that Twitter already filters most of them out of your timeline to avoid drowning you in duplicates. Their functionality is largely duplicated by the ‘like’ button, which (confusingly) also shares content into your followers’ timelines if they choose to see it. Removing them would force people to think before sharing - even for a few seconds - and would be another step toward calming the platform down.

While we’re giving people more control over their tweets, let’s have optional expiry dates, and the ability to opt out of search and lists. It simply isn’t healthy to have your every casual remark logged, recorded and searchable for posterity. It makes the site a haven for stalkers and obsessive weirdos who like to pore over decade-old tweets searching for evidence of some hypocrisy or wrong-doing, and it encourages people to seek out crappy content just to get worked up about it.

( Esther Choo posted a great thread on lists while I was writing this that’s worth reading in full: “Lists are a way for bad players to do efficient, coordinated harassment. Particularly if you notice an uptick in the volume of trolling, check your lists as one way of interrupting the process.” )

Which leads me to my final suggestion, and something Twitter have been working on recently: giving content creators more control over the replies to their content. This is an approach that works well on other creative platforms, like blogs, or Youtube. Comment sections that are well-moderated, with strong engagement from the author, tend to be happier and more productive places. Twitter on the other hand is a wild hell-site where even the most innocuous tweets can trigger a frenzy of hatred. Dumb comments sit under professional content like dog turds on a pavement. There’s no need for it. It adds nothing to the user experience of people who aren’t giggling idiots. Get rid of it.

Some people will take issue with this, saying that limiting responses will be less democratic, or reduce dissent or scrutiny. That’s nonsense. Twitter is neither democratic now, nor a source of serious scrutiny. It’s survival of the shoutiest. Any useful feedback or criticism is drowned out by the white noise of screamed reactions. Discussions would be better had on other platforms, or on the articles or blog posts being shared - at least then people would have to read them.

Some of these ideas might work, and some might be a disaster, but my bigger point is this: Twitter doesn’t really know what it is. It’s trapped between two competing identities: broadcast medium versus conversation medium. As a result, it has accumulated layers of features that just don’t work together. The solution is to pick one and go for it. Build for that vision, and ruthlessly cut the stuff that contradicts it. There are signs that this is happening, that Twitter is finally going all-in on quality content and the people who create it. Time will tell if it’s the right move.


Thanks for reading. Let me know how you’d fix Twitter in the comments, and stay tuned for next week’s post by clicking the big… orange? gold? button below!