How Twitter Shaped Internet Culture
The most consequential features of Twitter were not designed by Twitter. The hashtag, the retweet and the reply were all improvised by users inside a text box that offered none of them, and the company adopted them afterwards. That is the single most interesting fact about the platform, and it explains most of what followed.
What is in this guide
A limit borrowed from text messages
Twitter's defining constraint was an accident of a different technology. An SMS message allowed 160 characters, so the service capped posts at 140 and kept 20 in reserve for the sender's name. That meant a tweet could travel over the mobile phone network of 2006 without being split in two.
A limit designed for compatibility turned out to be a creative rule. It forced compression, which favoured jokes, headlines, announcements and insults, and disadvantaged argument, nuance and explanation. Nearly every cultural effect attributed to Twitter over the following fifteen years traces back to that trade. When the limit doubled to 280 characters in November 2017, the character of the site changed less than expected, because by then the compressed style was a norm rather than a constraint.
The features users invented
Three conventions now considered basic to social media were built by people using the product, not by the people making it.
- The hashtag. In August 2007, Chris Messina suggested using the hash symbol to group messages about a topic, borrowing the convention from internet relay chat channels. The company was initially unconvinced. Hashtags became clickable later, and are now standard across essentially every social platform.
- The reply. Users began prefixing messages with the at symbol and a username to address somebody in public. The service later turned that pattern into a formal reply structure, threading conversations that had previously been a stream of unconnected posts.
- The retweet. Before there was a button, people manually copied a message and prefixed it with the letters RT and the original author's name. The official retweet button arrived in 2009, and with it the mechanic that made the platform's virality both possible and, later, a serious problem.
The pattern of users inventing the grammar of a platform and the company formalising it afterwards repeats constantly: subscriber counts, stories, reaction emoji, thread numbering. If you want to know where a platform is going, watch what its users are already doing awkwardly by hand.
How it changed breaking news
For roughly a decade, Twitter was the fastest public information system in the world, and it got there by accident. Anyone standing near an event could publish to a global audience in seconds, without an editor, a broadcast van or a byline.
The turning point most often cited is January 2009, when a ferry passenger photographed US Airways flight 1549 in the Hudson River and posted the image before any news organisation reached the scene. Similar patterns followed for earthquakes, protests, elections and disasters: first reports arrived from participants, and professional coverage caught up afterwards.
The upside was speed and access, particularly during events where official accounts were unavailable or untrustworthy. The downsides were structural. Unverified claims spread at the same speed as verified ones. Corrections travelled far less widely than the errors they corrected. And the phrase social media revolution, applied enthusiastically to protest movements from 2009 onward, tended to overstate the role of the tools and understate the organising that made those movements possible, a criticism made at the time by researchers and by many participants themselves.
How it changed how companies speak
Before social media, a company's public voice was press releases and advertising, both slow and both formal. Twitter created a space where an organisation was expected to reply, in public, immediately, in a normal human register.
The famous demonstration came during the 2013 Super Bowl, when a power failure blacked out the stadium and Oreo posted a spare, well timed image about dunking in the dark within minutes. It was retweeted tens of thousands of times, cost approximately nothing, and launched a decade of brands attempting real time marketing with wildly variable results.
The format also produced genuinely new voices. Public institutions found that a first person account could carry science to an audience that would never read a press release, and space agency rover accounts writing in the first person are the clearest example. Meanwhile a generation of brand accounts discovered that being funny is difficult, that being funny while selling fast food is harder still, and that the failure mode is memorable in the wrong way.
What it did to the English language
Platforms leave vocabulary behind. Some of these words existed before and were repurposed; all of them reached general use through this one service.
| Term | Meaning |
|---|---|
| Hashtag | A topic label, now used in speech and in print far outside social media. |
| Subtweet | Criticising somebody publicly without naming them, so they may or may not notice. |
| Ratio | A post that attracts far more replies than approvals, taken as a sign of overwhelming disagreement. |
| Thread | A numbered sequence of posts used to defeat the character limit, which became a genre of writing in its own right. |
| Quote tweet | Republishing somebody's post with your own comment attached, which turned commentary into a broadcast format. |
| Main character | The person the entire platform has decided to discuss on a given day, almost never voluntarily. |
| Doomscrolling | Compulsively reading bad news, a term that reached wide use during 2020. |
Comedy changed shape too. The 140 character limit rewarded the setup and punchline in one breath, which produced a distinctly compressed joke form, and a generation of comedy writers who learned the craft in public, for free, in front of a real time audience.
The costs
An honest account of the platform's influence has to include what it did badly, because those effects were as durable as the good ones.
- Speed beat accuracy. The same mechanics that made real time reporting possible made rumour, misidentification and fabricated screenshots spread just as fast, and misidentification after major incidents caused real harm to real people.
- Outrage was structurally advantaged. Engagement based ranking systematically favours content that provokes a strong reaction, which is not the same as content that is true, useful or representative.
- Context collapsed. A remark written for a few hundred followers can reach millions of strangers who lack every piece of context that made it reasonable, a dynamic that turned individual bad posts into disproportionate public events.
- Harassment scaled easily. The features that let a good message find an audience of millions worked identically for coordinated abuse.
- The archive is fragile. Deleted accounts, policy changes, restricted access for researchers and the 2023 rebrand have made large parts of the record hard to cite, which is exactly why we verify tweets against original sources or contemporaneous reporting rather than screenshots.
A short timeline
| Year | What happened |
|---|---|
| 2006 | First tweet posted in March; the service opens to the public later that year. |
| 2007 | Breakthrough at the South by Southwest festival; the hashtag is proposed in August. |
| 2009 | The Hudson River photograph; the official retweet button arrives; verified accounts are introduced. |
| 2013 | Real time brand marketing goes mainstream after the Super Bowl blackout. |
| 2014 | A group selfie at the Academy Awards becomes the most shared post on the platform to that point. |
| 2017 | The character limit doubles to 280 in November. |
| 2022 | The company is acquired in October, beginning a rapid series of product and policy changes. |
| 2023 | The service is rebranded as X; competing short post platforms launch and gain millions of users. |
How we choose tweets for the game
The Who Tweeted deck contains only tweets we can trace to the original account or to reporting that quotes them directly, with the posting date recorded. We do not use screenshots as evidence, we do not invent plausible tweets to pad the deck, and when we could not verify enough material to reach a round number we stopped at the number we could verify. That decision is described in how we research and verify every question.
Frequently asked questions
What was the first tweet?
Jack Dorsey posted a short message about setting up his account on 21 March 2006, using the service's original spelling of its name. The platform opened to the public later that year and broke through with technology audiences at the South by Southwest festival in 2007.
Who invented the hashtag?
Chris Messina, a user rather than an employee, proposed using the hash symbol to group messages in August 2007. Twitter did not build support for it at the time and only later made hashtags clickable. The same pattern applies to the reply convention and the retweet, both user inventions that the company adopted afterwards.
Why was Twitter limited to 140 characters?
The limit came from SMS. A text message allowed 160 characters, and the service reserved 20 for the username so a tweet could be sent and received by phone without splitting. The limit doubled to 280 characters in November 2017.
Is Twitter the same thing as X?
It is the same service. The company was acquired in October 2022 and rebranded as X in July 2023. Tweets posted before the rebrand are still generally referred to as tweets, which is the convention we follow in the game.
How do you verify that a tweet in the game is real?
Every tweet in the deck has to be traceable to the original account or to contemporaneous reporting that quotes it directly. Screenshots are not accepted as evidence, because fabricated screenshots are trivial to produce and circulate widely.