Twitter is part of everyday life for millions of people, allowing them to network with friends, follow the news, and speak their minds. But as Twitter has grown, the company has been forced to exclude harmful accounts. As a result, allegations of censorship have emerged.

So how does Twitter decide which accounts to ban, and which to approve? As anyone who has regularly used the site will testify, the company’s practices can sometimes seem a little murky, but there are some things we know for sure about how censorship on Twitter works.

1. Shadow banning suspect accounts

In 2018, Twitter made an announcement which seemed to confirm what critics had been alleging for years: the social network would start restricting the visibility of certain “suspect” accounts.

Officially, the company stated that the proliferation of trolls had sparked this “shadow banning” strategy, often associated with repressive regimes.

This marked a significant shift from what had come before. Previously, Twitter had banned individual accounts based on the content of their tweets. Now, the company would look at patterns of tweeting. So, for instance, if a user regularly tweeted to strangers, it might be a sign of harassment or trolling.

By determining which accounts qualified as suspect, Twitter could relegate their content to obscure regions of peoples’ feeds, effectively neutralising their activity.

All of that sounds unobjectionable. Nobody really agrees with armies of Russian bots swarming across Twitter, or trolls making life miserable for individuals. But the move has led to criticism of Twitter censoring conservatives.

In fact, this criticism came from the White House itself, with Donald Trump citing “Many complaints” (ironically, in a tweet).

However, the social network responded to allegations of Twitter censoring conservatives, explaining that “[its] behavioural ranking doesn’t make judgments based on political views or the substance of tweets.” Instead, any aggrieved users had their behaviour to blame for slipping down feeds and search results.

2. Evidence of Twitter censoring hashtags

While conservative activists in the USA have vocally accused Twitter of silencing their voices, others have pointed to the hashtags used by the social network. For these critics, the spread of hashtags (marked by a # symbol) across Twitter is far from neutral.

In theory, hashtags are supposed to spread based on their popularity and the speed at which they “trend.” But there’s more to it than that, and this is where confusion can arise.

When a topic trends on Twitter, it’s not just due to gross popularity. Trending topics also have to be popular with “new audiences”, and be new themselves. Twitter aims to keep content fresh, and also prioritise growing their user base. Having a static user base which tweets fanatically is less useful for their business model than attracting a stream of new members.

But from the perspective of a regular user, this can seem obscure. If you see thousands of people tweeting about an event, you would expect it to trend. But if those discussing it are the usual suspects, the algorithms used by Twitter may not pick it up.

This may look like Twitter censoring hashtags, but is actually just a result of the way the company handles information.

This hasn’t prevented many people accusing the network of operating a Twitter censorship policy based on hashtags. For example, the conservative site Breitbart accused Twitter of censoring hashtags critical of Hilary Clinton during the 2016 election.

You can understand why people are angry about the company’s hashtag policy, but it’s not technically censorship. Unless we see the construction of algorithms as a form of censorship. However, Twitter would contest this, arguing that their algorithms are entirely neutral – which all the evidence suggests is the case.

3. Outright banning of certain accounts

In some cases, users can be removed from Twitter entirely, but does this amount to evidence of censorship on Twitter?

To be banned, users need to meet certain criteria, and Twitter is fairly open about what these comprise:

  1. Spamming – sending bulk posts to certain targets, or flooding Twitter’s servers with hundreds of marketing messages every day.
  2. Hacking – anyone who seeks access to Twitter’s systems is automatically out.
  3. Fake accounts – While users can impersonate and parody others, copying bios and avatars is frowned upon.
  4. Sending or linking to malware – no comment required.
  5. Revealing private information about other users (doxxing)
  6. Abuse and hateful content – this is often the most controversial aspect of Twitter’s banning policy. What constitutes a threat or insult can be subjective. But as Twitter explains, most speech is protected. Promoting violence against other users, racial abuse, or targeted harassment are all prohibited, as is encouraging others to commit suicide.

The main mechanism for triggering these rules is via complaints from other users, but these complaints aren’t always upheld. And if they are, in most instances users can appeal.

Users temporarily removed from the platform may have to supply a phone number to access their accounts – which is surely a massive privacy issue. But it’s not what we usually mean by censorship.

Still, this hasn’t prevented many people accusing the Twitter censorship policy of removing followers unjustly. For example, the hashtag #TwitterLockOut spread early in 2018 when conservatives lost huge numbers of followers overnight.

However, it looks like this “purge” was due to Twitter flushing out colonies of Russian bots, or bots made for sale to Twitter users. Still, the allegations won’t go away.

4. Twitter censoring tweets – does it happen?

But what if Twitter could either remove or suppress individual tweets at will? Wouldn’t this give the company absolute power to determine which subjects gain exposure, and which wither away?

As we’ve already noted, there is definitely evidence of users suffering from “shadow banning” users based on their previous tweeting patterns. However, there’s not much evidence of Twitter censoring tweets on an individual basis.

But wait. Twitter has long been associated with filtering tweets to suit the governments in certain countries. Back in 2012, the company was attacked for blocking anti-government messages in countries like Bahrain, Egypt, Saudi Arabia, and Iran.

In Thailand, for instance, Twitter admitted that it could remove tweets deemed critical of the monarchy, making them invisible to Thai users.

So yes, the platform has the capacity to censor individual tweets. Does it do so in most countries? Probably not. But it could.

How to avoid falling victim to censorship on Twitter

If you are a concerned Twitter user and want to ensure that your posts are as visible as possible, it’s vital to learn about the platform’s rules. Knowing where you stand will make it easier to avoid lapsing into “hateful” or spammy speech, which could result in forms of ban.

It may also be handy to use a VPN (Virtual Private Network) to conceal your location and identity. Twitter uses location information to determine whether users are genuine, but it’s not a precise strategy. So if you move around a lot, a VPN could help you stay on the right side of the platform’s algorithms.