On Thursday, an advisory council to Google published its report on the European Union's recently recognized legal principle of the "right to be forgotten" online. The report is the outcome of seven consultations with many experts and the public in Europe from September to November 2014. Luciano Floridi, one of the members of the advisory council, shares his thoughts about the report and the future of the debate.
REPORT SUMMARY: When Google approves a delink request in Europe, the report recommends Google continue its practice of removing the link across all its European versions of Google (Google.fr in France, Google.de in Germany, etc). The report suggests four main criteria that may help Google to evaluate individual deleting requests: 1) the public role of the data subject, 2) the types of information that may bias towards a private or public interest, 3) the source of the information (e.g. a newspaper) and 4) the time frame (relevance of old information). The report also advised that publishers should be notified of delinking requests and should have means to challenge improper delinkings.
In May 2014, the European Court of Justice (E.C.J.) issued a landmark ruling. It stated that in some circumstances, Google must remove (from its search index) links to personal information if this is "inaccurate, inadequate or no longer relevant." The ruling concerned a specific request by a Spanish citizen, Mario Costeja González. His name featured prominently in Google search because of two foreclosure notices published under legal requirement in 1998, when his property was repossessed for debt. In the end, the court accepted Mr. Costeja's claim that providing links to the notices was irrelevant to Google's purposes as a search engine under the 1995 E.U. Data Protection Directive. The links had to be removed.
It seemed a small episode, of no relevance. But it was actually a spark that ignited a lively international debate on how to regulate the availability and accessibility of legally published information online. We soon discovered that two fundamental principles, namely privacy and freedom of speech, had collided. This is why the outcome of such a debate may have long-lasting consequences and represent a watershed in the evolution of the Internet.
Since the May ruling, Google has received about 210,000 requests of removal and has taken down about 40 percent of them (updated figures: Google Transparency Report). Because the ruling concerned the Spanish case, it left unspecified several important aspects of implementation when other cases are in question. In order to identify the right policies to deal with each request of delinking, Google set up an advisory council. After having received contributions and feedback from experts and the public, we have now published our findings and recommendations. I would like to share some comments on two difficult points present in the report on which I was happy to compromise, even if I had slightly different views.
The first point concerns the geographical scope of the ruling -- the so-called territoriality issue. For centuries, roughly since the Peace of Westphalia (1648), political geography has provided jurisprudence with an easy answer to the question of how far a ruling should apply -- that is, as far as the national borders within which the legal authority operates. A bit like "my place, my rules; your place your rules." It may now seem obvious but it took a long time and immense suffering to reach such a simple approach. And it's still perfectly fine today, as long as you operate within a physical space. However, when it comes to the Internet, the space is logical, being made of data, protocols, URLs, interfaces and so forth. Which means that any place is only a click away. The result is that a ruling that concerns the Internet cannot rely on the old, Westphalian solution.
If you ask Google to delink personal information in Spain, all it takes to find the removed links is to check the same search engine in another country. The non-territoriality of the Internet works wonders with the unobstructed circulation of information. In China, for example, the government has to make a constant and sustained effort to control information online. But the same feature proves awkward if you are trying to implement the right to be forgotten.
The report strikes a fair balance, recommending to implement the delinking policy at the European level -- that is, if a request is approved, links are removed from all European version of Google's search engine. Personally, I argued in favor of a more restricted, nation-based delinking. The reasons in favor of this option are pragmatic. Most users never leave their local search engines. Also because of linguistic reasons, Spaniards use google.es, Italians google.it, Germans google.de and so forth. The power of default is enormous.
"I would have preferred a national rather than a European delinking because it would be effective without being excessive."
It follows that if Alice, who is French and lives in Paris, asks Google to delink some legally published information about herself, the most effective implementation is to remove the links from Alice's local search engine, namely google.fr. Over 95 percent of all searches in Europe are on local versions of Google. Thus, it is useless to remove them also from google.pt because virtually nobody in France will ever care to check information about Alice using the Portuguese version of Google, while the very few who may care will not be deterred by a pan-European delinking anyway. Someone who is determined to find a piece of information about Alice will simply use a search engine not based in Europe. Some have bitten the bullet and argued that all this is correct, but this is precisely why the delinking should be worldwide -- that is, applied to all versions of any search engine.
In the case of Google, this means delinking the information in question also from, for example, google.br (Brazil). I disagree. Why? Remember: my place my rules, but your place your rules. How could one explain to Brazilians that some legally published information online should no longer be indexed in a Brazilian search engine because the European Court of Justice has ruled so? Would the opposite also apply? Could Brazilians appeal? And how could one determine what is of public interest in this or that country? Maybe I am an investor from Brazil, and I do need to know whether a person has (for example) had some properties repossessed in the past.
Some have called for Google to extend its delinking to all its global search sites, since they are accessible within Europe. However, consider the following scenario. The day after some worldwide delinking starts being implemented, nothing will stop undemocratic and illiberal places from hosting a search engine that provides links to all information anyway. It would be ironic if we were to find information using a search engine based in North Korea because it was more complete than the local ones. Geographical space is no longer the solution; so the approach recommended by the report is a good compromise that adapts an outdated answer to a new question. It does not work very well but it is the classic "better than nothing" solution. Opting for a global delinking would be, instead, the classic "perfect is the enemy of good." It would be just another way of killing the Westphalian approach by asking the world to adapt to European decisions.
"It would be ironic if we were to find information using a search engine based in North Korea because it was more complete than the local ones."
When it came to finalizing the text of the report, I was happy, pragmatically, to concede the point because a pan-European delinking simply adds nothing to a national one, in terms of effective protection of individuals. It would be a different story if one were to argue that some legally published information online should be removed (the information itself, not just the link) altogether or blocked at the source (for example, by not allowing any search engine to index it in the first place). I am not against similar options, but I suspect that, in order to consider them, we would have to have a serious debate about how harmful the information in question needs to be to justify such a drastic solution. But this is something with which not everybody in favor of the "right to be forgotten" seems to be willing to engage.
The second point concerns the publishers, and it is simpler. I am of the view that publishers should be fully involved in the evaluation of a delinking request. They should have the right to know about whether someone has requested a search engine to delink some information that they legally published; to be informed about what decision has been taken by the search engine with regard to such a request; and to appeal, if they disagree with the delinking decision. All this applies even more strongly if a worldwide delinking approach were to be adopted.
Of course, the risk is that, by informing the publishers, one may enable them to re-publish the same contents in ways that can bypass the ruling and the delinking decision itself, both of which concern only personal information and hence "name and surname" searches. Yet this is a case in which I would recommend a principled approach. One could certainly implement disincentives, but the fact that publishers may misuse the meta-information about a delinking request is not an argument against their right to know and hence being able to appeal. This will be even more obvious the further we move towards a situation in which not being indexed by a search engine simply means "not being," full stop. In this case too, the report has found a fair balance, by recommending Google to follow the good practice of notifying the publishers "to the extent allowed by the law." It is a bit vague, and I would have liked to see an even more incisive position in favor of a full involvement of the publishers throughout the process, but it is a satisfactory compromise.
At the end of our consultations and internal discussions, once the report had reached a final version, each member of the advisory council had the possibility of adding a dissenting opinion. This is common practice but, given that the report is a finely balanced compromise that has been reached through long consultations and difficult negotiations, I was in favor of not taking advantage of such a possibility. So I invited all members to make an extra effort to agree on the outcome. Some of us decided to opt for such a conciliatory approach. Compromises have the distinctive property of leaving each party a bit dissatisfied. Our report is not an exception. But I hope that those who will discuss it will use it not to take it apart, but to make further progress on an issue so vital for the future of the Internet.
We are the transitional generation. In the future, both people in front and behind a desk during a job interview, for example, will be digital natives. When everybody will be on the other side of the divide, embarrassing pictures on Facebook may just be normal and acceptable. An analogy may be drawn with prenuptial sex -- something normal today, but still scandalous only a couple of generations ago.
How our culture and our notions of privacy and freedom of speech will change is very hard to guess but change they will. They are dynamic features of our social life and will evolve with it. I hope that they will change for the better, in favor of more relaxed and tolerant views of what, in the future, will be our personal information online. And I trust that more ideas, better technological solutions and new legal frameworks will provide for a reconciliation of privacy and freedom of expression -- two necessary pillars of any liberal democracy.