Google, Facebook and Twitter have control over an enormous proportion of the nation’s political conversation. Increasingly, as these platforms have started moderating political content, their power has been distorting and degrading the quality of American democracy, says Francis Fukuyama, senior fellow at Stanford University’s Freeman Spogli Institute for International Studies (FSI) and director of FSI’s Center on Democracy, Development and the Rule of Law (CDDRL).
“Big private corporations have neither the capacity nor the legitimacy to decide what types of information should be amplified or suppressed,” Fukuyama says.
In a white paper published by the Stanford Cyber Policy Center’s Working Group on Platform Scale, Fukuyama and his colleagues including Ashish Goel, Stanford professor of management science and engineering, and Stanford HAI International Policy Fellow Marietje Schaake, offer a novel proposal to deal with this problem: outsourcing content moderation to a layer of competitive middleware companies that would offer users of these platforms the ability to tailor their search and social media feeds to suit their personal preferences or objectives. “This would give users control over what they see rather than leaving it up to a nontransparent algorithm that is being used by the platform,” Fukuyama says.
The middleware proposal was featured as part of Stanford HAI’s “Policy and AI: Four Radical Proposals for a Better Society” conference, held November 9-10, 2021. Below, watch the full presentation.
Middleware: What Is It?
Middleware is software that rides on top of an existing internet or social media platform such as Google, Facebook or Twitter and can modify the presentation of underlying data. So, for example, a middleware provider might rate the credibility of various sources of information, or filter product searches for items that are eco-friendly or made in America – to fit a particular user’s preferences. “The point would be to give users of these platforms control over what they see in their searches or news feeds, rather than leaving that up to a platform that is completely nontransparent,” Fukuyama says.
Currently, at least one startup company, NewsGuard, has entered this niche in partnership with Microsoft. Their middleware, which can be bought as a browser plugin, rates the credibility of more than 6,000 news and information sources on a numeric scale.
“NewsGuard is an example that we would like to see more of,” Fukuyama says. “Consumers should have a choice among a variety of middleware options rather than leaving it all up to the internet platforms.”
Challenges to Making Middleware a Reality
Technological, business and governmental challenges would need to be overcome for the middleware proposal to be implemented, Fukuyama says.
On the technological front, a relatively small organization would have to be capable of developing AI that could sort through internet platform data in a timely way. The existing platforms argue that only they are up to this task, Fukuyama says. “Until someone actually tries to do it, we won’t know whether that’s true or not.”
Read all the proposals:
As a business matter, since NewsGuard seems to be the lone example of middleware that evaluates content accuracy, perhaps there isn’t enough consumer demand or enough of a commercial incentive to drive this business model, Fukuyama says. He and his colleagues are in the process of thinking through potential government regulations that would make it profitable for companies to offer middleware services to consumers. Perhaps, for example, to make these kinds of products viable, the government would have to require the platforms to share some of their ad revenue. And because the United States government doesn’t currently have the capacity to do this sort of thing, Fukuyama says, “We think there should be a new specialized agency for digital regulation.”
Middleware’s Potential Impact on Democratic Debate
Critics are concerned that the middleware proposal might make the online misinformation ecosystem worse, Fukuyama says. While some middleware companies would filter out what the mainstream media thinks is fake news, other middleware options would intentionally accelerate fake news.
Given that we live in a country governed by the First Amendment, the objective of public policy cannot be to prevent people from saying things that are false or misleading, Fukuyama says. Middleware doesn’t stamp out fake news, but it would hopefully prevent the big platforms from artificially amplifying it among mainstream Americans.
“Because these internet platforms are so large, their content moderation decisions have an oversized impact. We just want to reduce that power, not create an alternative source of power that will somehow cleanse American political discourse of falsehoods,” Fukuyama says. “We don’t think that kind of power is safe for anyone to deploy.”
Watch the Presentation
Stanford HAI's mission is to advance AI research, education, policy and practice to improve the human condition. Learn more.