Dr. King, who also leads the Institute for Quantitative Social Science at Harvard, developed the industry-academic model that will be used for the Facebook initiative. He recently sat down with Chris Bingham, Crimson’s Chief Technology Office, to discuss the project and answer live questions about its timeline, objectives and potential concerns.
It’s been a whirlwind few weeks for the world’s largest social media platform. Amidst stories about Cambridge Analytica improperly gaining access to millions of Facebook users’ data and Mark Zuckerberg’s first appearance before Congress, one piece of news about the social media giant may have gone under the radar, but could end up being the most important of the bunch: the announcement of a new academic research initiative designed to determine if and how Facebook influences democratic elections.
This question is at the center of a growing debate about social media data, and is incredibly important for us at Crimson Hexagon for two reasons. First because Crimson Hexagon closely analyzes the real world impact of online consumer conversations. And second, because the partnership will be spearheaded by Crimson Hexagon’s Founder and Chair of the Board, Gary King.
3 Highlights from our Q&A with Gary King about the new Facebook research initiative
How is the new partnership set up and what are its goals?
There’s quite an enormous amount of data that exists out there in the world. Social scientists that use these data to try to create social good and understand, and perhaps ameliorate, some of the great challenges that affect human society need these data to make progress. We have more data than ever before, but we also have access to a smaller fraction of data than ever before because large companies, like Facebook for example, have massive amounts of data that they don’t or aren’t willing or aren’t able to make generally available.
If it’s possible to come up with a way that’s incentive-compatible, we thought, for Facebook or other companies to give access to individuals or academic researchers, then we could really do something important. Ultimately, I worked with the executive team at Facebook to establish a two-part structure to make this possible:
- The commission. Highly respected, senior academic researchers who will go in Facebook, sign a nondisclosure agreement and have access to absolutely everything. They will then be allowed to find a question within the general category that’s agreed to — the effect of social media on democracy and elections — and a set of data that will allow them to answer their question, or series of questions.
- Requests for proposals. Academics all over the world can apply, some smaller number of them, after the peer-review process, will be approved, and they’ll be given grants and access to the data at Facebook.
What happens if Facebook disagrees with the findings?
One of the most important parts of this initiative is that there will be no pre-publication approval process. Facebook will not be able to review any of the findings before they are released. Indeed, if Facebook doesn’t allow the commission to ask certain questions, doesn’t allow them to have an RFP for certain questions, doesn’t give them access to certain data they feel they need, doesn’t give them access to certain people, then the commission is allowed to, and indeed obligated to, go to the public and say that Facebook wasn’t keeping up its end of the bargain.
Facebook is not going to be aware of the research before it’s published. They’re not going to be permitted to check on the research, they cannot stop it from being published.
How will the data be kept private?
With the Cambridge Analytica story, one researcher violated Facebook’s rules about sharing data. We’re not going to make it an individual responsibility to follow rules about data security, we’re going to make it a collective responsibility. So we totally know how to completely protect the most secret data you can imagine. Some researchers will only have access to data if they’re in Menlo Park, California. They’re in a locked room with an “air gap”, which means no connection to the internet. We can also use more convenient ways like a virtual clean room, a locked down laptop that can only do one thing, there’s no data actually stored on it, it communicates via high-level encryption back to a private server. Everything’s audited by the system, which is what we mean when we talk about collective responsibility. If a researcher types the letter ‘k’ everyone knows about it.