Twitter is testing “humanization prompts” which will show users' shared interest and mutual followers when replying to another user's tweet
Jack Morse / Mashable : Tweets: @twittercomms , @carnage4life , @bigblackjacobin , @tonyromm , @tha_rami , @hunterwalk , @green_footballs , @jmorse_ , @lolgop , @yashar , @brandyzadrozny , and @jolingkent Tweets: @twittercomms : “In the heat of the moment, people can forget there's another human behind a Twitter account. By showing what we have in common, we hope to remind people of what connects us as a starting point.” 🤝 https://twitter.com/... Dare Obasanjo / @carnage4life : Twitter is trying to reduce toxic replies by showing your mutuals & shared topics. I don't see how shared topics helps (we both like memes, so what?) but interesting if having mutual follows or followers makes a difference. Lots of bold experiments now. https://mashable.com/... https://twitter.com/... Edward Ongweso Jr / @bigblackjacobin : I thought y'all were doing a bit when you kept talking about “humanization prompts” https://twitter.com/... Tony Romm / @tonyromm : twitter's most popular interest tag is about to be “thrist” https://twitter.com/... Rami Ismail / @tha_rami : Really excited to finally learn both me and the person wishing me a violent death because I said accessibility considerations for Dark Souls both like “video games” https://mashable.com/... @hunterwalk : Yes! Facebook should be doing this as well. By helping people recognize their simarities, we can build empathy & shared connection, even when we might disagree. This won't stop asshole trolls but helps more of us stay human & kind online https://twitter.com/... Charles Johnson / @green_footballs : Ever get the feeling you're an unwiling guinea pig in a massive social engineering experiment run by idiots? https://twitter.com/... Jack Morse / @jmorse_ : This... is actually kind of an interesting experiment? https://twitter.com/... @lolgop : Looking forward to Twitter reminding me that both @marcorubio and I spent 2016 calling Donald Trump a hellacious con man. https://twitter.com/... Yashar Ali / @yashar : I love the things in common prompt in the screenshot below. I imagine one day I'll get a prompt and it'll say “You and Maria both follow elephant accounts.” or “You and Josh both follow Liam Hemsworth and Sean Cody.” https://twitter.com/... Brandy Zadrozny / @brandyzadrozny : I really like how Twitter tries things that are ... kinda weird? ... Like a lot doesn't work but they are so much more willing to test out things in furtherance of making the platform a nicer place to be. https://twitter.com/... Jo Ling Kent / @jolingkent : Love this idea - makes all of us a little more human... maybe even the trolls and bots too 😜 https://twitter.com/...
Context & Ripple Effects
This test is the latest step in Twitter's multi-year 'conversation health' arc: after testing hidden replies and tripling abuse-account suspensions in 2019, it moved to giving authors structural control over their threads with the reply-limiting feature rolled out through 2020. Humanization prompts attack the same problem from the other side — instead of restricting who can reply, they try to change how the replier behaves by surfacing mutual followers and shared interests at the moment of typing.
The timing matters: on the same day Twitter also announced it can infer which tweets users find funny from likes and retweets (personalized humor recommendations), showing the company pairing behavioral nudges with behavioral inference across the product.
First-order effects
- Users composing replies will see shared interests and mutual followers with the tweet's author before posting, inserting a social-context checkpoint into exactly the moment Twitter says people 'forget there's another human behind an account.'
- Tweet authors gain a passive de-escalation layer that works alongside the reply controls they already got this year — one shapes who enters the thread, the other shapes how they behave once inside.
Second-order effects
- Rivals face pressure to adopt the same nudge-first playbook: Facebook already followed Twitter's lead on pre-share prompts, testing its read-before-sharing nudge about a year after Twitter debuted a similar one.
- If prompts measurably reduce hostile replies, Twitter can lean less on reactive enforcement — the user-report-and-suspension pipeline that drove its 2019 abuse numbers — and shift moderation cost from trust-and-safety headcount to interface design.
Third-order effects
- The pattern points toward moderation migrating from punishment after the fact to friction built into the compose box itself, with platforms competing on conversational design rather than content removal alone.
- It also normalizes using inferred interest graphs (mutuals, shared topics, even detected humor preferences) as raw material for shaping interactions — a design direction whose effectiveness, and privacy trade-offs, remain genuinely unproven.
The trend: Social platforms are shifting moderation upstream, from suspending abusive accounts after reports to engineering empathy cues into the reply box before conflict starts.