Twitter Fights Back Against Harassment and Hate, But Is It Enough?

I‘ve been on Twitter for over a decade, and I love it for the laughs, the breaking news, the memes, and the sense of community. But as a woman online, I‘ve also faced my share of trolls and abusive accounts. From unsolicited sexual advances to threats of violence to coordinated campaigns of harassment, the dark side of Twitter can take a toll on your mental health and make you think twice before tweeting.

I‘m far from alone in this experience. A 2017 Amnesty International study found that an abusive tweet is sent to a woman every 30 seconds on Twitter, with women of color and LGBTQ women facing even higher rates of harassment. And a 2021 survey by Pew Research Center revealed that 23% of U.S. adult Twitter users have been targeted with online abuse, with younger adults and Hispanic users experiencing higher levels.

For years, Twitter was rightly criticized for not doing enough to protect its most vulnerable users and prevent hate speech from running rampant on the platform. Its reporting tools were cumbersome and ineffective, abusive accounts proliferated with impunity, and public pressure mounted with each high-profile incident of harassment.

But in recent years, Twitter has begun to take substantive steps to better combat abuse and foster a safer environment for users. From expanding muting and blocking capabilities to implementing AI systems for detecting problematic content, the company has made harassment prevention a clear priority. But are these measures enough to turn the tide against Twitter trolls? Let‘s dive in.

The Seven Prongs of Twitter‘s Anti-Abuse Strategy

Twitter‘s current approach to fighting harassment revolves around seven key changes rolled out in early 2017 and refined in the years since:

1. Expanded notification filters: Users can now fine-tune which accounts can contact them based on factors like profile completeness and prior interactions. For example, you can opt out of notifications from accounts that haven‘t confirmed their email or phone number, making it harder for bad actors to create throwaway accounts to harass people.

2. Improved mute options: In addition to muting accounts, users can now block specific keywords, phrases, and even emojis from appearing in their mentions for set time periods. This gives users more control over their experience and can provide much-needed respite from triggering or hateful language.

3. Greater transparency around reporting: In the past, users who reported harassment often felt like their concerns were being shouted into the void. Twitter now sends clearer in-app and email notifications informing you when they take action on one of your abuse reports. You can also view your reporting history to track repeat offenders.

4. Timeouts for abusive accounts: Twitter implemented a strike system where users who repeatedly violate the platform‘s rules may have their account temporarily locked or have their tweets hidden from non-followers. The goal is to limit the reach of bad actors without resorting to permanent bans, which they often circumvent by creating new accounts.

5. Safer search results: Twitter uses machine learning algorithms to proactively identify and collapse potentially abusive or "low-quality" replies so you see the most relevant content first. Of course, you can still choose to view those hidden replies, but they won‘t clutter your searches by default.

6. Downranked tweets: In a similar vein, Twitter also deploys AI to predict which individual tweets are likely to be abusive or spammy and limits their visibility in users‘ conversations and timelines. The tweets remain available if you seek them out, but they won‘t be given free amplification.

7. Stopping serial harassers: To prevent banned trolls from simply opening new accounts, Twitter has gotten better at spotting signals that a user was previously suspended and barring them from the platform completely. Specific methods are kept under wraps to prevent bad actors from gaming the system, but this could include fingerprinting devices and detecting patterns in IP addresses, phone numbers, and email addresses.

Taken together, these changes represent a significant shift toward proactively identifying and mitigating potential abuse rather than relying solely on user reports after the damage is done. But implementing them is a massive undertaking given the scale of Twitter‘s platform and the complexity of online harassment.

The Promise and Peril of AI Abuse Detection

Many of Twitter‘s anti-harassment initiatives hinge on machine learning algorithms that analyze billions of tweets and account interactions to flag suspicious activity in real-time. By looking at signals like an account‘s age, follower ratio, and posting frequency, these models try to determine the likelihood that it‘s engaging in abusive behavior and take automated actions like hiding its tweets or preventing it from being recommended.

There‘s no question that AI is an essential tool for content moderation at scale. No human team could hope to manually review the sheer volume of Twitter activity for policy violations. And properly tuned algorithms can identify problematic patterns and prevent abuse before it escalates in a way that reactive user reports simply can‘t.

However, AI moderation comes with its own set of risks and challenges. Models can inherit the biases of their human designers and pick up on spurious correlations in their training data, leading to unfair or inconsistent enforcement. Automated systems also struggle to grasp nuance, irony, and contextual cues in the same way human moderators can.

There have been instances where Twitter‘s algorithms have mistakenly flagged innocent accounts as spam or temporarily limited users for discussing contentious topics like sexual assault. The company maintains that human review is still involved in any permanent account suspensions, but a lack of transparency around these AI systems fuels valid concerns about unintended censorship.

As Twitter continues to refine its machine learning approach to harassment, it will need to prioritize fairness, accountability, and explainability. Users should have clear ways to appeal moderation decisions and receive explanations of why their content was flagged. And independent audits of the algorithms can help detect bias and suggest areas for improvement.

Gauging the Effectiveness of Twitter‘s Efforts

So are Twitter‘s anti-harassment measures actually making a difference? It‘s a challenging question to answer definitively, given the complex dynamics of online abuse and the lack of comprehensive data from the notoriously tight-lipped company. But there are some encouraging signs.

In a 2019 blog post, Twitter claimed that its machine learning systems were identifying and removing 2.5 times more abusive accounts on a daily basis compared to the previous year. And it said that 50% of tweets it removed for abusive content were surfaced proactively without a user report.

A 2021 study by the University of Michigan School of Information also found that Twitter‘s efforts to limit the visibility of abusive tweets were having a measurable impact. Tweets predicted to be abusive by the platform‘s algorithms received around 30% fewer likes and retweets compared to other tweets. The researchers concluded this was an encouraging sign that Twitter‘s interventions were "reducing the popularity of content it deems harmful."

However, despite this progress, harassment and hate remain entrenched issues on Twitter. Amnesty International‘s 2021 Twitter Scorecard, which assessed the platform‘s human rights impact, found ongoing problems with abuse reporting mechanisms, algorithmic bias, and transparency. And high-profile trolling campaigns like the racist attacks on Black England soccer players after the 2020 Euro Championship show how Twitter can still be a megaphone for hate.

Ultimately, no set of technical solutions will eliminate online harassment as long as the underlying societal prejudices and power dynamics driving it persist. But by continuing to iterate on its systems, listen to user feedback, and measure meaningful outcomes, Twitter can make strides toward becoming a safer and more equitable space.

Putting Twitter‘s Progress in Context

Twitter is far from the only social platform grappling with how to handle harassment and hate speech. From Facebook and Instagram to YouTube and TikTok, user safety and content moderation have become pressing concerns for the entire industry.

In recent years, we‘ve seen these companies take significant steps to bolster their anti-abuse arsenals. They‘ve hired more human moderators, established oversight boards, expanded user controls, and invested heavily in AI systems for flagging problematic posts. Governments around the world have also begun to introduce legislation aimed at holding platforms accountable for addressing illegal content.

But online harassment is ultimately a reflection of offline prejudice and discrimination. Combating it requires not just better platform policies, but broader social and cultural changes. We need comprehensive education on digital citizenship, bystander intervention, and allyship. We need to challenge the normalization of abusive language and hold public figures accountable for stoking hate. We need stronger legal protections and support resources for victims of cyberbullying and severe online harassment.

There‘s also an important balance to strike between protecting vulnerable users and upholding principles of free expression. When does an edgy joke cross the line into targeted harassment? How do we distinguish good-faith political discourse from hateful propaganda? Who gets to make these determinations and how can we ensure the rules are applied equitably? As much as we may want a one-size-fits-all approach, content moderation involves challenging judgment calls and evolving societal standards.

Twitter‘s recent changes are a meaningful step forward for the platform, but they‘re just one piece of the larger puzzle in making the internet a safer and more inclusive place. As users, we too have a role to play in shaping the culture and norms of the online communities we inhabit.

Strategies for Staying Safe and Sane on Twitter

So what can you do as a Twitter user to have a more positive experience and support your fellow tweeters? Here are a few key tips:

1. Master your privacy settings: Take advantage of the granular controls Twitter provides to limit who can see your tweets, who can message you, and what information is publicly available on your profile. Remember that even if your account is private, any of your followers can still take screenshots of your tweets.

2. Curate your notifications: Muting keywords and accounts liberally can go a long way in shielding you from triggering or spammy content. Don‘t hesitate to block accounts that are harassing you – you don‘t owe abusers your time or attention.

3. Use third-party tools: Services like Block Party and Bodyguard offer additional capabilities for bulk-blocking trolls, filtering abusive language, and auto-moderating your mentions. While it‘s unfortunate that users have to rely on outside tools to feel safe, they can provide much-needed peace of mind.

4. Document and report harassment: If someone is persistently harassing you or making threats, take screenshots of their tweets and file reports with Twitter and, in severe cases, your local law enforcement. Even if you‘re not sure it rises to the level of illegality, it‘s important to create a paper trail.

5. Lean on your support network: Online harassment can be incredibly isolating and demoralizing. Don‘t hesitate to reach out to friends, family, or a therapist for support. There are also online communities like HeartMob where you can share your experiences and get advice from people who‘ve been through similar situations.

6. Amplify positivity: One of the best ways to counteract negativity on Twitter is to boost the voices and content that bring you joy. Share uplifting stories, crack jokes with your friends, post cute animal videos – whatever helps remind you of the good parts of the platform.

7. Know when to log off: If Twitter starts to feel more draining than rewarding, it‘s okay to take a break. Your mental health should always come before any social media platform. Unplug, focus on offline hobbies and relationships, and come back if and when it feels right for you.

Ultimately, creating a safer and more inclusive Twitter will require a sustained commitment from the company, its users, and society at large. We all have a part to play in modeling digital citizenship, calling out abuse when we see it, and extending empathy and support to those targeted by harassment. While there‘s no quick fix, each small step we take together can help make Twitter a platform where more voices can thrive.