Tell Congress to Use the CRA to Save Net NeutralityLearn More About the CRA
Google's announcement that it is going to penalize sites that receive a lot of "valid DMCA notices"--that is, accusations of copyright infringement that are structured correctly (if not necessarily legally justified)--raises a lot of questions. Just this week, we've seen how Google's Content ID system has been abused by people claiming rights over content they do not own, and how the system itself has flaws that encourage invalid takedowns. Google needs to work hard to assure the Internet-using public that any new system it adopts does not have these same problems.
Voluntary measures such as this do not expand the law's requirements
Before getting to some of the potential drawbacks of Google's announcement it is necessary to clarify a few things. Internet companies are free to go above and beyond what the law requires to try to help copyright owners police for infringement, and they shouldn't be criticized just for doing that. But there's a trend where extraordinary measures that some companies with a lot of resources can afford become expected of everyone else--and this is a problem. Google can afford things that DuckDuckGo can't. It may be worth it for a company like Google that is involved in many activities, many of which (such as digital media stores) need the cooperation of media companies, to buy a little goodwill. But these extraordinary measures should not be expected of every website, and they should not become legal requirements.
It's significant that Google is announcing this change on the same day that public comments to the Intellectual Property Enforcement Coordinator (IPEC)--the government official in charge of supervising the government's various IP enforcement activities--are due. For a while now, the IPEC has been actively encouraging web companies to go above and beyond what the law requires. While it may be appropriate for the government to encourage different industries to work together instead of dragging each other to court, the government shouldn't pressure, cajole, or threaten sites for failing to do more than they are legally required to do with regard to IP enforcement. Private actors are free to tell individual companies they should do more, but when the government starts "encouraging" certain things this encouragement is often accompanied by an implied threat that if companies don't cooperate they may be legally compelled to.
A similar problem arises when plaintiffs in copyright infringement cases point to the extraordinary measures that some sites take, or that may be theoretically possible, as the minimum standard that any site needs to meet in order to qualify for DMCA safe harbors. Luckily, judges are usually skeptical of such claims, but this doesn't stop people from trying to turn content filters, automated takedown systems, or some other system into a legal obligation. In many cases, advocates for increased copyright enforcement fail to distinguish what companies legally must do from what they, in their opinion, ought to do. This has an unfortunate ratcheting effect where companies are expected to do more and more, regardless of the cost or social benefit. What Google has announced today should not become the baseline by which other companies are judged.
These are notices that Google is receiving, not notices that the sites themselves are receiving
Many people reading this will know that a site that hosts user-posted content has to comply with the DMCA process to avoid liability for copyright infringement. A site like Vimeo or a regular web host must disable access to copyrighted material upon receipt of a DMCA notice, otherwise it becomes liable for the content in addition to the user who posted it. The user then has the option to put the content back up by issuing a counter-notice, and if the copyright owner has a problem with that, it can take it up with the user directly. This system is far from perfect and sites and copyright owners sometimes don't do the right thing--sites may not make it easy for users to issue counter-notices, and "copyright owners" may issue takedowns for content that is not theirs, or is not infringing. Nevertheless, at least in theory, the system sets up a process that allows users, copyright owners, and sites to work out disputes.
However, the law also requires "information location tools"--like Google--to respond to DMCA notices if they are "referring or linking users to an online location containing infringing material or infringing activity." One difference between notices sent to "information location tools" and user-generated content sites and storage providers is that information location tools do not necessarily have any sort of relationship with the creators of the information they are indexing. If you have a website, and Google gets a notice about it, Google has to disable access to your site. Since you don't need to have any sort of relationship with Google for it to index your site, Google might lack any way to notify you that links to your site have been disabled.
To Google's credit, with its webmaster tools it does provide a means for sites to file counter notices, and as PK understands it notices that have had counter-notices filed no long count as "valid" notices for the purpose of this new search signal. Despite this, in practice, it is harder for sites to challenge questionable notices sent to search engines, than it is for sites (and the users of sites) to challenge notices they themselves have received. Because notices to search engines might not be challenged, entities with questionable copyright claims might be more willing to send such notices. And because being highly-ranked on Google can be so important, there's a strong incentive for entities to send DMCA notices to search engines to suppress their rivals. There's a danger that, in good faith, Google is setting up a process that can be abused. To counter this, Google should take steps to minimize the effect of fraudulent and bad-faith notices, and make sure that sites that are subject to these notices are aware that they can file counter-notices.
This new search signal may punish sites that are not disproportionally infringingPopular or Large Sites Should Not Be Penalized for Being Popular
Popular websites receive disproportionate attention and may receive a disproportionate number of copyright notices. This does not mean that these sites are disproportionately infringing. Sites shouldn't be penalized for being popular. Similarly, large sites that host a lot of content may be subject to many DMCA notices. This does not mean that they content they host is, on average, more likely to be infringing. Sites shouldn't be penalized for being large, either. Google should make sure that its system accounts for the fact that some sites will simply receive more notices than others, even though the content those sites host is not more likely to be infringing.Lawful Categories of Site Should Not Be Penalized
Sites that allow users to upload and share content are more likely to host infringing content than a site for a government agency or a restaurant. A photo-sharing site is more likely to host photos that infringe than the New York Times--not that large media companies don't have their own copyright troubles from time to time. But one of the purposes of the DMCA was to protect sites that host user content and allow them to operate free of legal uncertainty. Lawful sites shouldn't be penalized simply for their business model. Google should make sure that its new system does not disproportionally penalize sites with lawful business models, such as file-storage services or sites with user-generated media.
If Google's new policy helps users find legitimate sources of content, protects the valid interests of copyright holders, and doesn't penalize lawful sites, then it's a win all around. But any new system such as this has potential dangers and unintended consequences, and can be abused. Google is undoubtedly aware of this--but it remains to be seen how it will respond to problems that arise, and whether it will continue to put the interests of users first.