One of the most fundamental changes in Twitter’s policy ever, was the introduction, at the beginning of March 2017, of proactive moderation. Previously, Twitter had responded to user reports of abuse, offensiveness and spam on an individual basis, but this had largely failed to tackle an endemic problem with low-quality profiles and annoying or distortive spam.
Because Twitter is so huge, the bulk of this new realm of moderation would have to be automated. Twitter thus set to work devising algorithms which could attempt to identify known traits of low-quality or offensive profiles, and then penalise those accounts. Some penalties would be notified to the offending user; others would not. The classic shadowban is not notified to the offending user.
From the start, circa 1st March 2017*, a large number of Twitter profiles were auto-moderated, and dropped out of the search timelines for varying periods of time. The surreptitious measures which rendered many accounts widely invisible, quickly became known as shadowbans.
[*Update 16/4/2019 – I’ve now been able to confirm that tweets were being taken out of search based on shadowban criteria before 1st March 2017, but the measures had their publicised launch on that date.]
EFFECTS OF A SHADOWBAN
There are different types and degrees of shadowban, but most typically, the user’s tweets will drop out of Twitter’s default search results. Depending on why the user is shadowbanned, the following consequences may also apply…
- Other site-search entries relating to the shadowbanned user, become invisible. The user’s profile doesn’t show in a People search or as a search suggestion, for example. On some versions of the Twitter app, it’s possible you’ll see a shadowbanned user’s tweets in search, but their profile will not show in People. A profile with 100,000 real followers, a hundred to one fan/friend ratio, and high incoming engagement, would normally place top or close to top in People priority. But if that profile is shadowbanned, it may be replaced in those pole positions by inactive, default avi users with just a few followers, or “close match” users who don’t have the actual name that was typed. You see a clear distortion of priority, artificially pushing the shadowbanned account into an invisible position.
- The user’s tweets are not seen by followers on their main follow timelines.
- Relevant actions taken by the user do not show up in other people’s notifications. When the user follows someone new, or tweets them, for example.
- The user’s profile is completely invisible in the Followers lists of some or all of the people they follow.
- The user’s replies are not visible beneath the tweets they’re responding to.
- The user’s profile page is blanked out, visitors are given a relevant warning about the profile, and anyone wishing to see the content must accept via a “View profile” button. Some accounts displaying profile warnings are actually locked, but others are only shadowbanned.
Some shadowbans are selectively applied, and dependent on recipient users’ Quality or Sensitive Content settings. For example, an account which has been shadowbanned for low-quality usage may, along with its entire output, still be fully visible to users who have switched their Quality Filter off. And an account which has been shadowbanned for tweeting adult content may still be fully visible to users who have opted in to receiving sensitive matter. But with all settings left at their default, or when readers are logged out, the shadowban applies.
Whether or not a shadowban is selective, it’s not something you want if you’re interested in maximising your reach on Twitter.
THE TRUTH ABOUT HOW TWITTER SHADOWBANS ARE TRIGGERED
Let me stress at the top, that I’m focusing on shadowban offences in this post, and not more serious violations which may result in locking or suspension. Stuff like hate speech, doxing, harassment, threatening behaviour or incitement of violence carry an ultimate or instant penalty of permanent suspension if reports are upheld. I want to be clear in saying that just because I haven’t listed a violation as a shadowbanning offence in this post, it doesn’t mean it’s not a violation of the Twitter rules. Please read the main rules page for a summary of exactly what is or is not permissable on Twitter.
The reasoning behind actual Twitter shadowbans is heavily contested, but read on and I’ll help you clear that up. What we can be sure of, is that Twitter primarily shadowbans due to low quality usage and/or the potential to cause certain types of offence. Low quality usage is stuff like repetitive tweeting, or trivial spam.
GILT-EDGED SHADOWBAN OFFENCES TO AVOID
As of 15th April 2019 – the date of this post, Twitter is clear that the following acts are liable to result in a shadowban…
- Repeated posting of duplicate or near-duplicate content. This applies both to plain text and links in tweets, and it also specifically includes contest entry attempts when such attempts involve repetition. In other words, if someone is asking you to keep tweeting their website link in order to win a prize, and you actually do it, you’re likely to get shadowbanned.
- Aggressive follow and unfollow practices. Or “churn”, as it’s known. Note, however, that this can also result in an account lock or even suspension – especially if automation is used and/or the follow/unfollow interims are short.
- Hashtag and trend hijacking. That’s using a trending hashtag when tweeting about irrelevant subjects, but the rule also applies when the hash symbol is not used and a plain text trend is misused in the same way. This can also lead to more severe penalties if perpetrated in high enough volume – especially when the intention is to drive traffic off-site.
- Automated tweeting and botting. This accounts for a huge volume of shadowbans. If you’re using third party apps, make absolutely sure when you authorise them that they will not post automated tweets to your timeline. Go into the app’s Settings and uncheck any selections that allow it to tweet automatically. If you find any app is posting auto-tweets on your timeline, revoke its access.
- Sync-posting. Tweeting the same, or nearly the same messages across multiple accounts. This one is particularly relevant to shadowbans in the political arena I suspect. If mutiple people tweet the same thing, and Twitter misidentifies the accounts as belonging to the same user or organisation, shadowbans are likely. Social platforms have a deep concern about political groups using sockpuppets to distort the perception of public opinion. And a hundred accounts from the same region tweeting exactly the same, politically-focused message, can easily be identified as sockpuppetry. The solution is to be original. Never just echo what other people are saying, word for word.
Twitter reminds us that while the above are generally shadowban offences, it’s feasible for them to result in penalties up to suspension in certain cases.
Twitter cites a variety of other spamming practices as rule breaches, but deems them to be punishable with a notified penalty (such as an account lock or suspension) rather than a shadowban. These include list spamming, attempting to sell followers, posting rogue links, non-conversational or irrelevant replying in high volume, industrial scale Liking/RTing with the ultimate intention to drive traffic off-site, etc. Pretty much anything that attracts a significant volume of valid spam complaints from other users is liable to get you into bother. To be valid, spam complaints generally need to come from the actual people you’re messaging. Not third parties.
With this in mind, I’d strongly suggest you avoid replying to tweets that have a very high number of users tagged in, when you know your message will annoy most of those users. If the users you annoy are savvy, they’ll all report you for spam, and as I understand the rules, those reports will be valid, because each person was a direct recipient of your message, and not a “bystander”.
Although Twitter doesn’t list it on the Search Policy page, there’s also a shadowban attached to the posting of sensitive media or language. It’s not a universal shadowban, because the tweets can still be found in search by logged-in users who’ve consented in their Settings to seeing sensitive media. But for any reader logged out or on default settings, the effect is exactly the same as a shadowban – the tweets or profiles are not visible in search at all. Incoming tweets to the sensitive account may not be visible in that search either – even though those other users’ tweets could still show in searches for their own usernames.
It should additionally be noted that some sensitive media is not permissable at all, and across the board sensitive media must be marked as sensitive by the account holder. Even otherwise acceptable sensitive media must not be included in a profile pic or header, or live video. Failure to observe these rules can result in permanent suspension. See Twitter’s intimate media page for details on unacceptable adult matter and the contexts surrounding its posting.
Broadly speaking, that covers the shadowbanning offences. So how come Twitter shadowbanned you for something else? Keep reading…
Twitter stresses that it may remove profiles from search whilst it investigates a potential violation from a range of categories. So if you’re reported or flagged for an offence, you may be shadowbanned while the matter is in the moderation queue. If it then turns out that you’ve breached no rule(s), no action will be taken, and you won’t know why you were shadowbanned. This can explain a lot, particularly in heated debates, where two enemies are seething with anger and doing everything they can to get their opponents into trouble. It’s no secret that people try to get each other suspended on Twitter, and it can be hard to determine what sort of reports they’re filing.
The danger of interim shadowbans can be drastically reduced just by being polite to people. Twitter doesn’t consider one-off, non-discriminatory, non-harmful insults to be abuse by default, and it takes context into consideration. But while calling one person an idiot, on one occasion, is probably not going to meet the hateful conduct / abuse threshold (which carries a penalty of suspension), the user may still report you for it, and depending on the type of report they file, you may go on shadowban until Twitter has looked into it.
The lesson is, don’t call anyone an idiot. What positive outcome can it possibly achieve? You’re never going to convince them they’re an idiot – all you do is annoy them. If you’re getting shadowbanned and you’re not committing one of the specific shadowban offences, the likelihood is that you need to look at the way you address people or how you behave towards them. Make sure you’re not building up a high volume of trivial complaints, basically.
It’s well worth reading about the online disinhibition effect. Some of the people involved in Twitter debates have completely lost sight of responsible behaviour, acceptable conduct, and face-to-face tolerances, and are talking to random adults in ways that an angry parent wouldn’t even talk to their own child. If the arbiter of your behaviour is an automated bot whose job it is to decide whether or not you’re creating a “bad user experience”, it’s obviously not a good idea to provoke complaints on the regular.
Shadowbans can also feasibly be the result of misreporting. One of the problems with huge online platforms is that most people don’t take the time to read the rules, and they may therefore not know…
- What is or is not a violation of the rules.
- How to correctly report a specific violation of the rules.
Additionally, in the reporting mechanism, some reports are easier to file than others. The incidence of people reporting their concerns in the wrong category (especially the spam category) is therefore inevitably high. It’s not difficult to see how misreporting could result in a shadowban.
The final thing to consider is whether people who say they’re shadowbanned, actually are. People can mistakenly believe they’re shadowbanned, or they can make a false claim deliberately. The motivations behind the latter vary. Some make the claim in a bid to labour a political point (portray themselves as a victim of bias, essentially), some use it as a ploy to get sympathetic retweets… It’s just human self-interest overriding the truth – common practice on Twitter. The lesson is, don’t automatically accept that because someone says they’re shadowbanned, they are shadowbanned. Check for yourself. The number of times I’ve done so and found the person to be fully visible, logged out, is unreal.
IDENTIFYING A SHADOWBAN
Personally, I don’t use apps to identify a shadowban. My test is just to log out and search my own @username on Twitter. If I see my account in the search box suggestions and my own tweets come up in search, I’m not shadowbanned to any extent I need to worry about. If I don’t see myself, I’m shadowbanned.
However, this Shadowban Tester app is open source, and offers a nice discovery method for anyone who’s looking for more detail.
My preferred way of avoiding shadowbans is to think less about actual, individual rules, and more about the experience of other people. Simply, am I doing something that’s likely to annoy anyone? And how would my actions play out on a traditional, moderated forum? What could I get away with there? Would the admins/mods tolerate me invading every thread with a repetitive prompt and a link to my website? No. Would they be okay with me adding value to the board and providing a link, where relevant, for those who wanted more detail? In most cases yes. Twitter’s outlook is going to be much the same.