One year after doubling tweet length, Twitter finds the average tweet is now 33 characters, one fewer than before the change; 5% are longer than 190 characters
Context & Ripple Effects
The 280-character experiment followed a fast arc: a September 2017 test on some users (the initial expansion from 140), a worldwide rollout two months later in every language except Chinese, Japanese, and Korean (the global rollout), then a BuzzFeed study in December finding long tweets earned nearly twice the retweets and likes. By February, Jack Dorsey was already defending the change, insisting the average tweet length had stayed flat while engagement rose.
This Axios data point closes the loop on Dorsey's claim with hard numbers: a year in, the average tweet is 33 characters — actually one shorter than before the doubling — and only 5% of tweets run past 190 characters. The extra room went almost entirely unused by the median user, which matters because it shows the change was never about making everyone write more.
First-order effects
- Twitter's own metrics vindicate Dorsey's February defense: median behavior barely moved, so the 280-character limit cost nothing in feed density while handing heavy users more room.
- The 5% of tweets exceeding 190 characters identifies who actually consumed the new capacity — a small, high-output segment whose longer posts were already earning outsized retweets and likes per the December study.
Second-order effects
- With engagement gains concentrated in that long-form minority, Twitter's incentive shifts toward serving power users with even larger limits rather than optimizing for the silent majority posting 33 characters.
- Competing platforms reading the same data see that raising a character ceiling is a low-cost engagement lever — the constraint itself becomes product surface worth tuning.
Third-order effects
- If the pattern holds, character limits stop being a fixed identity constraint and become a tunable, eventually tiered feature — the logical endpoint being paid access to much longer posts, which is where Twitter's later 4,000-character subscriber offering landed.
- Platform design increasingly follows measured engagement distributions rather than stated product philosophy: a change justified by a 5% cohort's behavior signals that vocal power users, not median users, drive format decisions.
The trend: Social platforms are converting once-fixed format constraints like character limits into adjustable, engagement-optimized features — and ultimately paid tiers — calibrated to their most active users.