Two days after the riot at the U.S. Capitol, Twitter’s most senior policy executive faced her 5,200 colleagues on a video conference and made an impassioned appeal.

Vijaya Gadde‘s voice was breaking as she implored her colleagues to have patience while her team deliberated over what they knew was the most important decision in the social media service’s 15-year history.

Some Twitter employees left the meeting on Jan. 8 not knowing what to think. They were worried for their colleagues’ safety — some had already received security threats. But they were also angry that Gadde’s team had let President Trump’s account back onto the service after a 12-hour ban for appearing to encourage the Capitol rioters on the day of the failed insurrection. He had already tweeted again, telling followers they were patriots who would not be disrespected.

Less than three hours after Gadde’s emergency meeting, Twitter banned Trump forever. In an instant, the megaphone of the leader of the free world was wiped out, along with his following of 88 million he had built throughout his presidency — some of whom amplified his every word. And it brought to an end an era of free speech online that Twitter — which a senior executive once referred to as “the free speech wing of the free speech party” — had itself helped create.

Twitter on Jan. 8 banned President Trump from its site, a punishment for his role in inciting violence at the U.S. Capitol. (The Washington Post)

A dozen current and former employees and close observers of the company reconstructed the critical decision, marked by tearful meetings, bitter internal arguments and the culmination of years of debate within the company.

Twitter declined to comment and pointed to its lengthy company blog post on the topic.

Billionaire Twitter CEO and founder Jack Dorsey wasn’t celebrating the moment. He lamented what the decision meant for the vision of an open Internet that he and his Silicon Valley peers shared when they sought to build a new type of microblogging service 15 years ago. While necessary for public safety, he said it would be “destructive to the noble purpose and ideals of the open Internet” over the long term in a tweetstorm this week, his first comments since the ban.

Dorsey, Gadde and others felt they didn’t have much choice. Earlier that week, a mob of Trump supporters, following the president’s calls on Twitter and in speeches, had attempted a takeover of Congress, leaving one Capitol Police officer and four others dead.

Twitter’s decision would reverberate internationally, prompting concern from foreign leaders and reshaping how a global public viewed the might of social media platforms. It triggered more aggressive calls for regulation of the loosely regulated tech industry. It would fundamentally alter some of Silicon Valley’s most cherished stances on free speech and the big tent of expression its leaders long described as core ideals.

And it would fracture the public square even further, as Trump and some of his supporters were also banned from Facebook, Twitch, Snap, Spotify and other services. The growing right-wing social media site Parler, which many Trump supporters had flocked to, was itself taken offline by its service providers and removed from the Google and Apple app stores.

Twitter then purged 70,000 more accounts last weekend for affiliations with the conspiracy theory QAnon. Leading Republicans lost thousands of followers, while conservatives continued to decry the unchecked power of tech companies. And already, researchers reported evidence that the actions caused a significant drop in the volume of misinformation online.

Until recently, the company was so reluctant to police content that it only put into place a formal misinformation policy last year. Its corporate slogan — in a 2016 rebrand — was to be the place to hear “what people are talking about right now,” particularly among agenda-setting influencers and political figures. The use of Twitter by public figures to break news allowed the company to punch above its weight against giants like Facebook and YouTube.

Public figures were so important to Twitter that, like Facebook, the company long gave them a newsworthiness exemption. They were allowed to say things that would be considered to be policy violations if uttered by everyday users, on the grounds that what these people said was in the public interest. Twitter, and Dorsey in particular, long maintained that even the worst lies should be corrected with more speech rather than removing them. It was Twitter’s responsibility to show “how global leaders think and treat the people around them,” he said.

But Twitter came under more and more pressure as the Trump presidency wore on. It was clear that he would use the platform to harass citizens, including a former adviser who was a Black woman, whom he compared to a dog. He also criticized a Gold Star family. He also shared falsehoods and conspiracy theories — a communication style that would help him build the most avid audience in the history of the platform, say researchers.

Twitter’s largely liberal employee base faced growing criticism, and workers complained that the first question they were asked when they told someone they worked at the social media service was, what about Trump’s account? His account was even briefly deactivated once by a rogue Twitter employee in 2017.

By 2018, Dorsey and Gadde, whose title is legal, policy and trust and safety lead, knew they had to rethink their approach to powerful people’s megaphones. Executives began to devise new policies and product features that would enable the company to place a specific label to cover up a tweet. The warning label — launched in 2019 — was a middle ground. It would tell users that the tweet broke Twitter’s rules but was being allowed to stay up because it was newsworthy. People who wanted to read the content could click to see what was behind the label.

The label was first applied to Trump’s comments in May, when he appeared to suggest that racial justice protesters could be shot — “when the looting starts, the shooting starts.” It was flagged for breaking rules on “glorification of violence.” Two other tweets that contained falsehoods about mail-in ballots were labeled as well.

Even before the election, Twitter took measures to guard against his attempts at misinformation. The efforts were part of a policy to thwart election misinformation, which included labeling and blocking many tweets.

During the week after the election, Twitter covered up so many tweets by Trump and his high-profile supporters that it was difficult, at times, to find a tweet that wasn’t blocked. The company also disabled retweeting and other features that are used to spread misinformation in real time.

After the election, Twitter let up a bit on the pace of its labeling, but a violent Trump-led movement to fight the election results was just getting started, including on Twitter.

Trump used Twitter to call for a rally at the Capitol. “Big protest in D.C. on January 6,” he tweeted in late December. “Be there, will be wild!”

That Wednesday, thousands of protesters showed up to answer Trump’s online calls, including some who were armed and prepared for violence. As rioters stormed the Capitol, causing property damage, Trump encouraged their actions with tweets.

“These are the things and events that happen when a sacred landslide election victory is so unceremoniously & viciously stripped away from great patriots who have been badly & unfairly treated for so long. Go home with love & in peace. Remember this day forever!”

Twitter suspended him for 12 hours that evening and made him delete the offending tweets, while Facebook suspended him indefinitely the following morning. A blog post from CEO Mark Zuckerberg said that despite any news value the comments might have, the risk of violence was too great to keep them up.

“We believe the risks of allowing the President to continue to use our service during this period are simply too great,” Zuckerberg wrote. “Therefore, we are extending the block we have placed on his Facebook and Instagram accounts indefinitely and for at least the next two weeks until the peaceful transition of power is complete.”

Following the deadly Capitol riot, Twitter employees engaged in vigorous debates on Slack, the preferred communication channel for its thousands of people working remotely.

“I’ve been part of the ‘he’s the president, we can’t deactivate him’ crowd for 4 years now but even I have to say, I feel complicit allowing this to happen and I would like to see him deactivated immediately,” one employee wrote on a Slack channel, which was first reported by NBC News.

The outrage boiled over into an open letter to Dorsey and Gadde, drafted Wednesday. The letter argued that Twitter should suspend Trump’s account permanently and conduct an investigation into the past several years of corporate actions that led to Twitter’s role in the riot.

Others argued that Twitter needed to build policies for the long term and that any action taken would be seen as a politically vindictive move against Trump. A company-wide email from Dorsey the morning of Jan. 7 following the 12-hour suspension — which had lifted a few hours earlier — seemed to reiterate that stance. He argued that Twitter needed to build rules that can endure beyond any given moment in time.

Trump still hadn’t tweeted again since the suspension, but a chorus of advocacy organizations, Twitter allies and even former executives began to call for a permanent ban.

University of Virginia Law professor Danielle Citron, a longtime Twitter adviser and free-speech expert, said the company should have suspended Trump “the minute” he tweeted about looting and shooting. “If they had properly applied their rules, he would have been gone.”

Twitter employees were starting to face a backlash. Some made moves to lower their social media profiles — removing Twitter from their bios — and began forwarding questionable posts that might result in an attack to security.

Then, on Thursday evening, after more than 25 hours, Trump tweeted again. First a conciliatory video.

On Friday came the familiar attack lines, vowing his “Patriot” supporters would not be “disrespected” or “treated unfairly,” and a final tweet saying he would not attend President-elect Joe Biden’s inauguration.

The letter, which was shared on Slack internally Thursday in a channel called #iamjackama, gained more than 350 signatures by early Friday.

Dorsey and Gadde held the emergency town hall on Friday over Google Hangouts. Dorsey — in French Polynesia at the time — opened the meeting and turned the floor over to Gadde.

In her remarks, a visibly emotional Gadde explained how hard the teams making the decision were working and that some were doing so while facing serious security threats. She explained that the company was still conducting an investigation process over the impact of the tweets.

Some employees were skeptical. “People are fed up. They were like, ’You’re crying, but I’m crying too, so now what?’” an employee said.

Others couldn’t tell whether the executives were defending the decision to leave the account up or foreshadowing a stronger action that would soon be taken.

Behind the scenes, Gadde’s team was closely examining the implications and online reaction among Trump’s most fervent followers to each of Trump’s tweets. The way they were interpreted was key to the company’s reasoning. The initial video Thursday left some feeling betrayed, Twitter researchers learned, while the second “Patriots” tweet and the third tweet boycotting the inauguration were being read as a call to arms and a sign to continue the fight.

Less than two hours later, and 48 hours after Trump’s initial suspension, Gadde’s team suspended the president, with Dorsey’s sign-off.

But Dorsey was troubled by the action and wrestled with the notion of power. That disappointment spilled into the open on Wednesday night, when Dorsey said, in a characteristically philosophical tweetstorm, that he felt the suspension reflected a failure on his company’s part to maintain a space for civil discourse.

For years, he and Twitter had benefited immensely from being at the center of conversations, the public square where news happened. Now, a dangerous precedent was set, he said, in that people were retreating from the public square even further. Dialogues across different viewpoints would be even less possible, and technology companies were becoming even more powerful than governments. That the company had to ban Trump, he felt, was a failure of his service to maintain civil discourse, perhaps existentially.

Dorsey said he didn’t believe the answer to the problem was stronger government. Instead, he championed a nascent vision of a future Internet that would be self-governing and inspired by the decentralized model of his other passion, bitcoin.

It wasn’t clear from the tweets how that decentralized vision would protect the public from the next Donald Trump.