Extract from The Guardian
The integrity of our future elections depends on structural reforms to how Facebook is allowed to operate
Mark Zuckerberg’s testimony to the European parliament
in Brussels appeared to hold the company accountable to the more than
500 million people represented by 28 countries in the EU, but was this
just an optical illusion?
In fact, only a small group of MEPs spoke with him, and they had almost no time to prepare. The format allowed Zuckerberg to cherrypick his responses rather than reply to each individual point. As a former Google design ethicist who spent last week in Brussels with EU policymakers, I question whether this is real accountability or just a minimal concession.
Zuckerberg’s EU hearing, originally to be in private but ultimately livestreamed after public uproar, excluded many EU officials with the deepest understanding of technology and social media. He accepted the invitation just six days in advance and before a public holiday that left Brussels empty from Friday until just before Tuesday’s hearings. It’s a familiar pattern – Congress was on recess in the two weeks before Zuckerberg’s testimony in the US, with little time to prepare.
Facebook has repeatedly appeared to treat these issues as a crisis of PR, not of democratic integrity, making mostly the small changes needed to avoid further scrutiny when the real solution to the problems at hand involves deep, structural changes to Facebook’s thinking and design, which enabled these problems in the first place.
It should have been a historic moment, this first time that Zuckerberg appeared in the EU to answer questions in a democratic process. He likely won’t visit again. Yet Facebook will continue to powerfully influence the elections, mental health and culture of EU nations. Ireland is already in the grip of controversy over foreign actors’ Facebook ads in its forthcoming abortion referendum. The UK parliament has repeatedly asked for him to testify in its continuing inquiry into the Brexit referendum and Cambridge Analytica, but he has no plans to do so – yet many questions still need an answer from the Facebook boss.
In fact, only a small group of MEPs spoke with him, and they had almost no time to prepare. The format allowed Zuckerberg to cherrypick his responses rather than reply to each individual point. As a former Google design ethicist who spent last week in Brussels with EU policymakers, I question whether this is real accountability or just a minimal concession.
Zuckerberg’s EU hearing, originally to be in private but ultimately livestreamed after public uproar, excluded many EU officials with the deepest understanding of technology and social media. He accepted the invitation just six days in advance and before a public holiday that left Brussels empty from Friday until just before Tuesday’s hearings. It’s a familiar pattern – Congress was on recess in the two weeks before Zuckerberg’s testimony in the US, with little time to prepare.
Facebook has repeatedly appeared to treat these issues as a crisis of PR, not of democratic integrity, making mostly the small changes needed to avoid further scrutiny when the real solution to the problems at hand involves deep, structural changes to Facebook’s thinking and design, which enabled these problems in the first place.
It should have been a historic moment, this first time that Zuckerberg appeared in the EU to answer questions in a democratic process. He likely won’t visit again. Yet Facebook will continue to powerfully influence the elections, mental health and culture of EU nations. Ireland is already in the grip of controversy over foreign actors’ Facebook ads in its forthcoming abortion referendum. The UK parliament has repeatedly asked for him to testify in its continuing inquiry into the Brexit referendum and Cambridge Analytica, but he has no plans to do so – yet many questions still need an answer from the Facebook boss.
Accountability
Facebook affects the public health and elections of more than 180 countries, yet has answered to the public only where the media has been loudest. In lesser-reported countries like Sri Lanka, where fake news on Facebook has unintentionally amplified violence, Facebook’s response has been criticised as insufficient. What will you change to ensure greater accountability to democratic governments?Levels of harm
Metrics reorganisation
The only way Facebook monitors how changes it makes to its platform impact two billion people is through metrics. How have previous metrics caused the wrong thing to happen? By what metrics do you monitor social goods and harms? Because metrics are developed internally and subject to those teams’ blind spots, how will you use independent advisers or academics to scrutinise them and make recommendations?Team reorganisation
It seems that even if a Facebook employee knows that a part of the service will cause harm, it’s hard for that person to make change – especially if it hurts any way that teams measure success. What kinds of internal reorganisation of your teams will be necessary to make the company capable of anticipating the harms and goods you see as most relevant? How are your employees incentivised to anticipate these harms and goods rather than just to focus on performance goals?Protection of elections
Social media ads can be purchased from anywhere, disguised as normal social media posts, micro-targeted to polarise voters, and loaded up with sensational and divisive messages. How do you plan to address these problems far in advance of important elections instead of at the last minute, as in Ireland? When will society’s interest in free and fair elections trump business interest in maintaining hyperefficient programmatic advertising?Artificial intelligence
Facebook believes that using AI in addition to content moderators will successfully combat fake news and malicious ads. But the exponential diversity of threats in languages your engineers don’t even speak – millions of advertisers running millions of campaigns, targeting users with immeasurably diverse content (posts, groups, ads) through automated systems – means that automated detection of new issues will lag behind the harm they cause. This potential for continued harm suggests a need for approaches besides AI. What are those approaches?Corporate governance
The Federal Trade Commission in the US and the Commons digital, culture, media and sport committee in the UK are investigating whether a consent decree and laws were broken in regard to Cambridge Analytica. Facebook has no independent board or risk committee, and a shareholder group has questioned the work of the company’s audit committee. What changes, if any, do you think Facebook needs to make to its governance system to avoid any such mistakes in the future?The public and governments across the world have had enough with Facebook tinkering in optics and small changes. The future of our elections urgently needs structural reforms of the platform to reflect the interests of governments, civil society, and ultimately Facebook itself.
• Tristan Harris is a former design ethicist at Google
No comments:
Post a Comment