Comment

View from Washington: Five takeaways from Zuckerberg’s defence of Facebook

Facebook founder and CEO Mark Zuckerberg finally faced US politicians during two hearings on Capitol Hill.

Facebook founder and CEO Mark Zuckerberg has finally faced US politicians during two hearings on Capitol Hill: a joint session of the Senate Judiciary and Commerce committees and another of the House of Representatives Committee on Energy and Commerce. Here are five takeaways that emerged from Mr. Zuckerberg goes to Washington.

1. It was a very polished performance by Mark Zuckerberg

Under almost 10 hours of questioning it was always likely that Mark Zuckerberg would be rattled at some point – anyone would be. There were two significant incidents: firstly, when he was questioned on Facebook’s links to Palantir Technologies in the Senate hearing – Palantir is a controversial software and services company that happens to be chaired by Facebook board member Peter Thiel – and secondly, when Zuckerbeg was subjected to more rigorous quizzing by the later House committee hearing.

However, Zuckerberg basically stayed on target. He patiently hammered home Facebook’s main talking points – e.g. hiking the number of staff and contractors working on security and content review to 20,000; an increasing use of artificial intelligence (AI) to catch bad content; an audit programme to detect rogue app data harvesting; an openness to regulation.

Zuckerberg also played the necessary role of political punchbag, allowing senators and representatives to slap down Facebook’s careless actions to date on data privacy. He acknowledged their concerns.

Finally, while there were some snippy exchanges – e.g. Senator John Kennedy famously telling Zuckerberg, “Your user agreement sucks”, or Representative Marsha Blackburn cutting him short for a “filibuster” – he did not let things become confrontational. Rather, his focus was on cooperating with legislators on regulatory proposals and following up questions with more information later.

Zuckerberg may well have set the foundations for Facebook actually strengthening its political hand in the medium term, although it’s not all rosy yet.

2. The content did not always match the optics

While Facebook will be far happier than its more aggressive critics on Capitol Hill, the evidence process itself is far from over.

Wired has identified 43 occasions where Zuckerberg said that he would have to get back to his inquisitors with the answers. Some were understandable, e.g. the status of the company’s rural broadband programme. Others seemed strange, given that this was the CEO speaking (and he desperately did not want that title recast as ‘Chief Evasions Officer’).

Zuckerberg appeared remarkably ignorant when it came to discussion of Facebook’s tracking technology (e.g., the ‘pixel’ it uses to monitor users’ ad clicks), the data points it uses for profiling and – a huge bugbear among privacy activists – its default data-sharing settings.

Part of that might be about Facebook protecting its IP. In some cases, it could reasonably provide written answers, but ask for them to be sealed from public view as commercially confidential. Given that privacy issues were paramount at both hearings, however, Zuckerberg’s inability to provide even ballpark answers seemed odd.

Similarly, the CEO’s evidence also seemed to make an unstated but significant distinction between two forms of personal data. Zuckerberg was happy to discuss the ‘Your data’ category in as much as it referred to what a user personally posts or uploads to Facebook. Politicians were repeatedly told that the user owns this and can delete it at any time. Yet the committees heard barely a word about what we might call ‘Data about you’. This is the increasingly controversial category insiders call ‘derived data’ and is based on what is gleaned from tracking tools, ‘Likes’, the membership of ‘Groups’ and other components drawn from what users do and where they go on Facebook and elsewhere.

The company is likely to have to clarify exactly what it means there soon, or the topic could come back with a vengeance.

3. AI is still the new ‘shiny’

At the Senate hearing, Zuckerberg cited Facebook’s growing use of AI tools to detect rogue content in answers to 10 senators, roughly a quarter of those who took part. He claimed that Facebook is already having some success deploying AI to identify and delete terrorist content.

Although Facebook is a leading AI R&D investor, a question mark still hangs over whether the technology will be ready soon enough to answer enough of the company’s immediate content problems.

Even Zuckerberg himself acknowledged that AI today is good in some areas, but not others.

“Today, as we sit here, 99 per cent of the ISIS and Al Qaeda content that we take down on Facebook, our AI systems flag before any human sees it. So that’s a success in terms of rolling out AI tools that can proactively police and enforce safety across the community,” he told Senator John Thune.

“Hate speech? I am optimistic that, over a five to 10-year period, we will have AI tools that can get into some of the nuances - the linguistic nuances of different types of content to be more accurate in flagging things for our systems.”

OK, the Facebook CEO did add the caveat to some of his comments, but there was still a sense that he was blinding the politicians – and, by extension, the public – with science, something Internet companies have been too eager to do of late concerning AI.

Remember, YouTube rolled out its own AI-based system to catch terrorist content last year and quickly had to pull it for flagging too many false positives from legitimate news organisations. Zuckerberg did not talk – nor was he asked to talk – about what mistakes Facebook’s anti-Islamic State system has made, although he did have to address other blocks including posts from the pro-Trump African-American vloggers Diamond and Silk.

Is Facebook still prone to over-promising and under-delivering? It needs to be careful here, too, given its recent history.

4.  Regulation is coming

No big company welcomes regulation unless it is inevitable. Given that, one exchange between Zuckerberg and Senator Lindsey Graham was key.

Graham: You embrace regulation?

Zuckerberg: I think the real question, as the Internet becomes more important in people’s lives, is, ‘What is the right regulation?’ Not whether there should be or not.

Graham: But you, as a company, welcome regulation?

Zuckerberg: I think, if it’s the right regulation, then yes.

Going back to the measured nature of Zuckerberg’s testimony, this encapsulates what was almost certainly the company’s main medium-term goal for the hearings.

Right now, Facebook has few friends in government, even as it ramps its lobbying spend to massive levels. At the same time, it knows that – despite what some think – there is little political appetite for breaking up or even closing the company. It is an American success story, so Congress is in no mood to go there.

So, Zuckerberg set out to mend fences, but also position his company at the forefront of ‘advising’ on what this now-inevitable regulation should cover. He explicitly undertook to review the Browser Act already being prepared in the House of Representatives. He also repeatedly brought up the General Data Protection Regulation about to enter force in the EU, probably not as an exact template but certainly a valuable model.

5. Technology needs to learn as much about politics as politics needs to learn about technology

The charge that national governments contain too few tech-savvy politicians feels ancient. Still, criticisms that the ‘middle-aged’ politicians quizzing Zuckerberg seemed digitally ignorant were widespread after the hearings. Yet they were not entirely fair.

While some questioning during the Senate hearing was (being charitable) anaemic, the House hearing was far more pointed and aggressive. Beyond that, though, many of the charges levied by techies actually exposed their own ignorance of the political process. If regulation is inevitable, that will have to change.

Remember, the hearings limit each questioner to just four or five minutes - hardly enough time for a forensic examination. The sessions are organised in speaker order rather than by topic, so the subject will often switch back and forth. Finally, even those ‘dumb’ questions are often not what they appear.

An old trick among Washington (and Westminster) politicians is to phrase something that sounds a bit daft, but which the witness then goes at length to correct – often giving away much more than he would have, had the point been made more precisely. The traditional thinking is that technology folk are among those most likely to fall for it: one of the most famous victims was Bill Gates during Microsoft’s anti-trust hearings.

Zuckerberg, by contrast, is known to have had an A-team of lobbyists briefing him for his hearings and dodged the trap. That tells us a lot, too.

Facebook is looking to turn the current scandal around by making itself the legislators’ main point of contact with Silicon Valley. I am not sure it will ultimately succeed. There are still many unanswered questions that could have very damaging answers. It needs to execute internally far more effectively than ever before.

It has, at least, now figured out how a DC sausage is made. And has it just leap-frogged over the competition as a result?

Recent articles

Info Message

Our sites use cookies to support some functionality, and to collect anonymous user data.

Learn more about IET cookies and how to control them

Close