How Do We Know if the Laws are Working?
About the author: Sarah Myers ’21 is an FSI The Europe Center intern at the Lisbon Council in Brussels, Belgium.
So far, the work I’ve done at the Lisbon Council has revolved around finding information. I’ve spent a lot of time researching particular policy areas, trying to find information about specific policies and their effects. Sometimes, this can be challenging.
One of the projects I’m working on is about intermediary liability. Intermediary liability, in this case, means the extent to which an online service provider is liable for content that it hosts. At the turn of the century, the US and European Union both decided that online service providers are not liable for content that they passively host. This allowed several tech giants, including Facebook and Google, to grow. Now, however, these governments are reconsidering this policy. In the last few years, Facebook has struggled to avoid passively hosting hate speech, disinformation, incitement to terrorism, and copyright infringement, and Google has sometimes directed people to the most popular websites rather than the most accurate ones. Both companies have taken steps to address these problems. Facebook has increased its efforts to remove damaging content, and Google has introduced a rating system for websites which is designed to move misleading entries lower down in the results list. Both companies, along with Twitter, voluntarily signed on to the European Commission’s 2016 Code of Conduct for social media platforms. Since then, all three have increased the speed at which they remove hate speech.
Some policymakers and scholars believe that making these companies liable for the content they host will force them to take a more active, and therefore more effective, approach to removing illegal or damaging content. Unfortunately, as I’ve discovered while working on this topic, we have almost no evidence about how changing intermediary liability laws might change the internet. Some people believe that increasing liability for intermediaries will lead to a safer, better internet. Others believe that it will create an overwhelming and damaging censorship regime, as companies err on the side of caution to avoid legal problems and censor anything remotely controversial. But, because the EU and US adopted nearly identical intermediary liability laws and have not changed them since, we do not have real world evidence about the potential impacts of different laws.
China has adopted full intermediary liability, meaning that online service providers are liable for all content. This has effectively outsourced censorship of the internet to private companies, creating an effective and cheap (for the government) censorship regime. But many people in the US and EU view this as exactly the kind of censorship-run-amok scenario which we should strive to avoid.
In order to find information about the possible effects of different intermediary liability regimes, I spent about a week researching the policies EU member states have enacted which affect intermediary liability. Fortunately, Stanford’s World Intermediary Liability Map provides a list of all such policies. All I had to do was search every policy listed, for all 28 EU member states, in both English and the relevant national language. The problem was, I didn’t find anything. With few exceptions, no evaluations, impact reports, or other relevant studies had been published about these laws.
Ironically, the work I did is still useful, because it presents convincing evidence that the EU has not yet done enough to evaluate how its policies play out in the real world or to evaluate how proposed changes to intermediary liability might play out. But I’ve started wondering about how countries evaluate their laws more generally. There are, of course, some laws whose effectiveness isn’t really relevant. Laws against murder, for instance, must exist. But laws like the ones about intermediary liability, where we have certain outcomes in mind but have little to no good evidence for how to achieve those outcomes, present more difficult questions. It does not help that, perhaps because we are accustomed to writing and relying on the first kind of law, most governments do not have processes or institutions which can effectively and regularly evaluate how laws are affecting the real world.
Some progress has been made; the United Kingdom now has a program called What Works, which publishes recommendations about specific policies which have been shown to work in well-designed studies. The EU does, at least, perform impact assessments predicting the impacts of proposed policies before enacting them. The US has the Congressional Budget Office and Congressional Research Service. The problem is mainly that these programs need to have broader scopes, ones which include follow-up studies on laws after they’re enacted. If nothing else, it would make my project a whole lot easier.
