Protocol | Policy

Twitter's image cropping was biased, so it dumped the algorithm

The META team found the Twitter cropping algorithm occasionally discriminated based on race and gender.

Twitter

Twitter is abandoning its image cropping algorithm after finding racial and gender-based bias.

Photo: Joshua Hoehne/Unsplash

In one of the first instances of total algorithmic transparency from a social platform, Twitter has removed its image cropping algorithm and created new controls for users over how images appear, based on newly published findings of gender and skin color biases.

Members of the Twitter Machine Learning, Ethics and Algorithmic Transparency team conducted a research experiment to review how the image-cropping algorithm chose points of focus in pictures (called "saliency") and found occasional instances of bias in cropping based on the gender and skin color of people in photos. The researchers found that there was an 8% difference in favor of women from demographic parity, a 4% difference in favor of white individuals, and a 7% difference in favor of white women in comparisons of black and white women, META lead Rumman Chowdhury wrote in a blog post Wednesday afternoon. The research found other biases as well. (The company's analysis used subjective applications of "Black," "white", "male", and "female", not the total spectrum of skin tone and racial and ethnic identities).

After analyzing the tradeoffs between the consistency of the algorithm and its potential for harm, Twitter opted away from speed — an unusual, if unheard of, policy change for social networks that usually obsess with reducing friction — and instead designed a new way to share vertical, uncropped images. "This update also includes a true preview of the image in the Tweet composer field, so Tweet authors know how their Tweets will look before they publish. This release reduces our dependency on ML for a function that we agree is best performed by people using our products," Chowdhury wrote.

Twitter first introduced the "saliency" cropping algorithm in 2018 in order to make timeline photos appear more consistent, training the algorithm to crop based on how the human eye sees pictures. The cropping almost immediately drew criticism by people who felt that from their own anecdotal experience, the crops often focused on people with light skin and on women while cutting out others, which could perpetuate both demographic biases and the objectifying "male gaze" toward women. While Twitter at first said the algorithm had been tested for biases before it was released, the company then apologized in September 2020 to the users who were sharing their own experiences with bias on the platform and committed to further assessing the algorithm.

Over the last year, Twitter has heavily emphasized user choice in its public statements about conversational health and safety and in its descriptions of how the company is designing future products. The team did the same for its analysis of the cropping algorithm, asking whether the cropping algorithm caused harm by taking away people's ability to make their own decisions. "Not everything on Twitter is a good candidate for an algorithm, and in this case, how to crop an image is a decision best made by people. They should be able to decide for themselves what part of an image is the most important and what part is the focal point of their Tweet," Chowdhury wrote.

The company also shared a paper published today detailing how the researchers found the biases within the algorithms and how they analyzed the possible harm that could be caused by them, written by Kyra Yee and Tao Tantipongpipat from the META team and Shubhanshu Mishra from the Content Understanding Research team. "The use of this model poses concerns that Twitter's cropping system favors cropping light-skinned over dark-skinned individuals and favors cropping women's bodies over their heads," they wrote in the paper's conclusion.

Chowdhury, a widely-respected leader in the field of algorithmic accountability, was hired in February to lead the new META team after months of widespread criticism toward large tech companies and social platforms for their failures to analyze possible biases and discrimination in their algorithms. She is just one of Twitter's many recent hires of prominent and well-respected leaders in tech ethics and accountability.

Power

How the creators of Spligate built gaming’s newest unicorn

1047 Games is now valued at $1.5 billion after three rounds of funding since May.

1047 Games' Splitgate amassed 13 million downloads when its beta launched in July.

Image: 1047 Games

The creators of Splitgate had a problem. Their new free-to-play video game, a take on the legendary arena shooter Halo with a teleportation twist borrowed from Valve's Portal, was gaining steam during its open beta period in July. But it was happening too quickly.

Splitgate was growing so fast and unexpectedly that the entire game was starting to break, as the servers supporting the game began to, figuratively speaking, melt down. The game went from fewer than 1,000 people playing it at any given moment in time to suddenly having tens of thousands of concurrent players. Then it grew to hundreds of thousands of players, all trying to log in and play at once across PlayStation, Xbox and PC.

Keep Reading Show less
Nick Statt
Nick Statt is Protocol's video game reporter. Prior to joining Protocol, he was news editor at The Verge covering the gaming industry, mobile apps and antitrust out of San Francisco, in addition to managing coverage of Silicon Valley tech giants and startups. He now resides in Rochester, New York, home of the garbage plate and, completely coincidentally, the World Video Game Hall of Fame. He can be reached at nstatt@protocol.com.

While it's easy to get lost in the operational and technical side of a transaction, it's important to remember the third component of a payment. That is, the human behind the screen.

Over the last two years, many retailers have seen the benefit of investing in new, flexible payments. Ones that reflect the changing lifestyles of younger spenders, who are increasingly holding onto their cash — despite reports to the contrary. This means it's more important than ever for merchants to take note of the latest payment innovations so they can tap into the savings of the COVID-19 generation.

Keep Reading Show less
Antoine Nougue,Checkout.com

Antoine Nougue is Head of Europe at Checkout.com. He works with ambitious enterprise businesses to help them scale and grow their operations through payment processing services. He is responsible for leading the European sales, customer success, engineering & implementation teams and is based out of London, U.K.

Protocol | Policy

Why Twitch’s 'hate raid' lawsuit isn’t just about Twitch

When is it OK for tech companies to unmask their anonymous users? And when should a violation of terms of service get someone sued?

The case Twitch is bringing against two hate raiders is hardly black and white.

Photo: Caspar Camille Rubin/Unsplash

It isn't hard to figure out who the bad guys are in Twitch's latest lawsuit against two of its users. On one side are two anonymous "hate raiders" who have been allegedly bombarding the gaming platform with abhorrent attacks on Black and LGBTQ+ users, using armies of bots to do it. On the other side is Twitch, a company that, for all the lumps it's taken for ignoring harassment on its platform, is finally standing up to protect its users against persistent violators whom it's been unable to stop any other way.

But the case Twitch is bringing against these hate raiders is hardly black and white. For starters, the plaintiff here isn't an aggrieved user suing another user for defamation on the platform. The plaintiff is the platform itself. Complicating matters more is the fact that, according to a spokesperson, at least part of Twitch's goal in the case is to "shed light on the identity of the individuals behind these attacks," raising complicated questions about when tech companies should be able to use the courts to unmask their own anonymous users and, just as critically, when they should be able to actually sue them for violating their speech policies.

Keep Reading Show less
Issie Lapowsky

Issie Lapowsky ( @issielapowsky) is Protocol's chief correspondent, covering the intersection of technology, politics, and national affairs. She also oversees Protocol's fellowship program. Previously, she was a senior writer at Wired, where she covered the 2016 election and the Facebook beat in its aftermath. Prior to that, Issie worked as a staff writer for Inc. magazine, writing about small business and entrepreneurship. She has also worked as an on-air contributor for CBS News and taught a graduate-level course at New York University's Center for Publishing on how tech giants have affected publishing.

Protocol | Workplace

Remote work is here to stay. Here are the cybersecurity risks.

Phishing and ransomware are on the rise. Is your remote workforce prepared?

Before your company institutes work-from-home-forever plans, you need to ensure that your workforce is prepared to face the cybersecurity implications of long-term remote work.

Photo: Stefan Wermuth/Bloomberg via Getty Images

The delta variant continues to dash or delay return-to-work plans, but before your company institutes work-from-home-forever plans, you need to ensure that your workforce is prepared to face the cybersecurity implications of long-term remote work.

So far in 2021, CrowdStrike has already observed over 1,400 "big game hunting" ransomware incidents and $180 million in ransom demands averaging over $5 million each. That's due in part to the "expanded attack surface that work-from-home creates," according to CTO Michael Sentonas.

Keep Reading Show less
Michelle Ma
Michelle Ma (@himichellema) is a reporter at Protocol, where she writes about management, leadership and workplace issues in tech. Previously, she was a news editor of live journalism and special coverage for The Wall Street Journal. Prior to that, she worked as a staff writer at Wirecutter. She can be reached at mma@protocol.com.
Protocol | Fintech

When COVID rocked the insurance market, this startup saw opportunity

Ethos has outraised and outmarketed the competition in selling life insurance directly online — but there's still an $887 billion industry to transform.

Life insurance has been slow to change.

Image: courtneyk/Getty Images

Peter Colis cited a striking statistic that he said led him to launch a life insurance startup: One in twenty children will lose a parent before they turn 15.

"No one ever thinks that will happen to them, but that's the statistics," the co-CEO and co-founder of Ethos told Protocol. "If it's a breadwinning parent, the majority of those families will go bankrupt immediately, within three months. Life insurance elegantly solves this problem."

Keep Reading Show less
Benjamin Pimentel

Benjamin Pimentel ( @benpimentel) covers fintech from San Francisco. He has reported on many of the biggest tech stories over the past 20 years for the San Francisco Chronicle, Dow Jones MarketWatch and Business Insider, from the dot-com crash, the rise of cloud computing, social networking and AI to the impact of the Great Recession and the COVID crisis on Silicon Valley and beyond. He can be reached at bpimentel@protocol.com or via Signal at (510)731-8429.

Latest Stories