Guest Contribution By Rigo Wenning (Gastbeitrag von Rigo Wenning) Legal Counsel, W3C
The pending EU privacy regulation has to address search engines
Introduction & Summary
Yesterday, the European Court of Justice decided on Case C-131/12. This decision has some rather surprising characteristics. This counts for the procedure as well as for the content. The first welcoming remarks hailing better privacy protection have faded into a disillusioned critique about the expected collateral damages of the decision with respect to freedom of speech and the added bureaucratic effort. A more down to earth view would suggest that the decision is actually a reminder that search and publicity reserve a lot of challenges, especially concerning the weighing of human rights against each other. It is also a reminder that data protection authorities are often bureaucrats with a regrettable lack of technical understanding. But this also addresses the technical community. I don’t think we can let the courts alone figure it out. In the following, I will try to show that the court was forced to decide the way it decided. This in turn transforms the case into a mandate for the privacy regulation to take the challenge into account. The challenge is to overcome the habit in the legal system to target the central entities in the distributed system that is the web and replace this lazy practice by something that plays well with the system. And there, the IT industry has to get active. Data protection is for the IT industry like environmental protection for the car industry. It makes their technology socially viable and it needs investment, e.g. standardizing the control interfaces for search crawlers.
The case
But let us first look at the underlying case. A local journal in Cataluña, La Vanguardia, has a PDF archive of its printed pages. On page 23 of La Vanguardia from 19 January 1998, there is a right column with small print announcements. It is part of a a database with past events exposed on the Internet. The PDFs are text PDFs, so a search engine can actually index the content. The announcement in question says:
Lesdues meitats indivises dun habitatge al carrer Montseny,8, propietat de MARIO COSTEJA GONZÁLEZ i ALICIA VARGAS COTS, respectivament. Superficie:90 m2.Cárregues: 8,5 milións de ptes. Tipas de sbhasta: 2 milions de ptes.cadascuna de lesmeitats.
Note that I the above paragraph reproduces the content, but contains a <div class="robots-nocontent">
element that tells the search engine not to take up that content. I’ll come back later to this issue of tagging.
Google indexed the PDFs and La Vanguardia has a rather high pagerank. A search on the name would obtain links to two pages of La Vanguardia’s newspaper, of 19 January and 9 March 1998 respectively, on which an announcement mentioning Mr Costeja González’s name appeared for a real-estate auction connected with attachment proceedings for the recovery of social security debts.
Mr González turned to the Agencia Española de Protección de Datos (Spanish Data Protection Agency; ‘the AEPD’) and requested La Vanguardia to remove the pages or alter them to avoid their indexing. He also requested that Google would not show the results anymore. There is no mention of the cached pages feature that Google has. Google refused and the AEPD and Mr González went to court. The Spanish court submitted the questions to the ECJ.
Is it privacy or Google-control? AEPD decision hard to understand
AEPD decided that the content on La Vanguardia could remain as this was justified by the intention to give maximum publicity to the auction in order to secure as many bidders as possible
. I see everybody jump on the contradiction. But it is not the ECJ’s contradiction, it is a contradiction introduced by AEPD. And this way, the question was thus excluded from the decision of the ECJ.
Apart from the question how the AEPD could take this as an argument for the publication of an announcement from 1998, this creates a very interesting situation. There is a page on the on the Web that is currently served by a Web server and that is publicly available. There are a number of search engines that still have La Vanguardia’s archive indexed. Even La Vanguardia itself is still offering a search for names on its archives. But all this remains outside of the question submitted to the ECJ. This means that the AEPD, by its very decision not to further pursue La Vanguardia and other search engines, put the core question out of scope for the ECJ, who could only respond to the questions asked. The core question of the case is the justification of the AEPD that named classified adds from governmental procedures in journals can remain online for archives over a long time. This is a question that already the French legal service had to address in Légifrance and that all official journals have to address. A known and common question where a response on the European level wouldn’t have been a bad thing. But this question was not on the table. But what question was on the table?
It may be human to only look at the annoyance of the moment. And the annoyance of the moment was certainly the prominent listing within Google. The focus on somebody’s image within the services of Google blurs the clear view on the real issue behind the case. By the decision to let La Vanguardia go, the AEPD turned the case from a case about governmentally sponsored official publications with personal information into a case about the say of a data protection authority over Google. This is not a good basis for sensible solutions.
The court responds to 3 questions
Now there are two questions left:
- Is a search engine processing personal data?
- Can persons request removal of their personal data?
Additionally, the court addressed the question of territorial competence of data protection law. If one factors out the Google hype around the question, those are reasonable questions to ask.
In response to the questions, the ECJ undertakes a very detailed review of the various conditions and requirements of the legal texts in question. The English text (translated from Spanish) does not only sound like a very detached lawyer-like geeking exercise. In fact, the court remains deep in the weeds of detailed questions. One hardly gets the feeling the ECJ is addressing the higher exercise of weighing freedom of expression against privacy. Art. 10 of the ECHR or Art. 11 of the Charter of Fundamental Rights of the European Union are not cited a single time. The ECJ has simply not addressed that question, because the questions to the ECJ where such that it did not have to address it. So what are the questions the court addressed?
Is a search engine processing personal data?
Of course it does. But lawyers can’t be that simple. Instead, there is a detailed analysis whether Google can be a data controller. Of course they control what they process and what they provide to the public. They even make money with knowing it by providing detailed advertisement matching the search users are executing. The question about whether Google is a data controller serves for the court’s argumentation to determine whether Google can actually accomplish the request brought to them.
And Google? Google responds with the argument that made me laugh hard already when it was used to excuse the reading of people’s emails for the gmail advertisement: We do not actually read your email, the engine is doing that
. Referring to uncle Harry or the man on the moon for things that one has diligently programmed and that one fully controls is a joke and will always remain a non-argument when confronted with real courts. If I would be the NSA, I would use the same argument in the current debate. From a legal point of view it was impossible for the ECJ to give in to this argumentation. Google said: We are only a cache
and the court responded no you aren’t
. It was inevitable, given the argumentation.
But Google is not a European company. One can’t follow all the rules of the world simultaneously. This counts even for Google. The international dimension was scrutinised and the court found a reasonable set of conditions to apply European or Spanish law to Google. In fact, Google is targeting the European market, makes money here, has even a Spanish subsidiary and thus has to respond to those rules. Reading the comments on the decision, many people think Google will now close the subsidiaries and contract directly from the US. But this isn’t so obvious as economic activity in a market as large as the European one needs some management and organisation that responds to local questions. If all goes wrong, such a targeted market will address the money flows to enforce its rules. While it is nice to read some of the romantic visions about the Internet in the blogosphere, it is prudent to keep a realistic view on things.
The right to be forgotten
Remains the question about the merits of a request to erase an entry in a database and to add a URI to the blacklist for the crawler. The Spanish court submitting the questions to the ECJ anchored the scope of the right to be forgotten within Article 14.a of Directive 95/46EC. Can a citizen ask a search engine to remove an entry?
Google is of the opinion that a search engine should only react once the publisher has erased the content. The Austrian Government supports this and addresses exactly the point about freedom of speech that Thomas Stadler was making, namely that (quoting from paragraph 64 of the judgement):
a national supervisory authority may order such an operator to erase information published by third parties from its filing systems only if the data in question have been found previously to be unlawful or incorrect or if the data subject has made a successful objection to the publisher of the website on which that information was published.
To all of us professionals of the Internet, this immediately makes sense and it addresses the end-to-end principle and works well with the distributed nature of the Internet in general and the Web in particular. But the ECJ did not follow that argument and here we are confronted with the most controversial question of the case. In articles about this subject matter there is often an undertone about courts not understanding the Internet. The counter argument to the Austrian position is of a rather legalistic and systemic nature which may not be understood, this time, by the netizens. I can only try to translate.
Mr González, the Italian, the Spanish and the Polish government do not accept the hierarchy. They say a person can request the removal from the search engine without having to pursue the publisher first. The Commission says that the fact that the information has been published lawfully and that it still appears on the original web page has no effect on the obligations of the search engine. Here the Polish government joins Austria and removes the obligation to act.
The court, confronted with the two opposite positions, makes a very systemic argument out of the legal nature of Directive 95/46EC. Processing of personal data is prohibited in Europe unless it is allowed. The ECJ draws from this fact that everybody who processes personal data must have a legal ground or the permission by the data subject to process that data. The journal La Vanguardia was already out of the question as described above because the AEDP declared they had a legal ground to publish the personal information. The ECJ could not reverse that decision as this was not a question it had to answer. Now they were looking for a reason for Google to process that data. And they couldn’t find one. So the prohibition prevails. And if someone requests removal and there is no legitimate grounds for processing (whatever that means), the person can ask to remove the data from the index.
Of course, the court ornaments all that with a flood of words of caution that this only counts in this very case and that this may be overridden by a legitimate interest of the Internet users concerning that information and that even a journalist could potentially use that information and that the court eventually did not feel very well with a mere prohibition and opened every window it could while closing the door. They already prepared the escape line.
Opinion: Who is responsible for the wrong result? Not the ECJ!
The decision has merits. It states the obvious: Google can’t just claim to be a cache and escape responsibility while making business on top of the functionality it controls. The decision is a clear warning for Google not to continue the argumentation around it’s not us, it’s just the engine
. Search engines are processing personal data and they control the processing of that personal data. That means they are part of the relevant actors and can’t just hide in the woods of being an innocent middlebox that does nothing.
The decision identifies an additional impact on the privacy of individuals by the search engine over the publishing of the information. A fair assessment will follow the court in this analysis. But for the non-lawyers reading this: Having said that something impacts my right to privacy doesn’t say anything about the consequences that impact triggers or is supposed to trigger. Those consequences are not implied by the Charter of Fundamental Rights of the European Union. It only means we have to address the issue somehow. Nobody really questions that.
The decision shows why the Directive 95/46EC is outdated and needs urgent replacement. A court is not a parliament. A court is supposed to apply law, not to replace it. The ECJ did strictly apply the law. It could have been less strict and follow the Austrian position, but that would have been at the very limits of a possible argumentation. The Austrian position had also two branches that remained unexplored: One could argue that the search engine will only be forced to remove the entry from the index once the publisher has removed the content or one could argue that the search engine is only responsible if all other publishers have already removed the content. The latter would be a factual denial of the protection as duplication is so easy and frequently used in scenarios that follow the Streisand effect. The court hinted at this and made it one reason to have an independent right against the search engine operator. As Thomas Stadler already indicated, the funny thing here is that the ECJ is by its prominence responsible for the Streisand effect on Mr González. So the entire procedure had the exact opposite effect on him. While some people in Catalyuña knew about his difficulties of 1998 in the past, now the entire European union knows about it. Should we remove the court decision from the index?
If there is a feeling that the result of the decision is wrong, it is due to two factors: On the one hand, the AEDP’s decision to have the PDFs of La Vanguardia kept online is putting all the burden on the search engine while keeping the initial harm. This is hard to understand but not the fault of the ECJ. On the other hand, the search engines are an important part of the Web infrastructure. Manipulating search engines or opening the way for private persons to manipulate search engines is very sensitive. There is a feeling that the Directive 95/46EC on Data Protection is just too simplistic and one-sided in this respect. This in turn contributes to the finding in this opinion that Directive 95/46EC is the wrong tool for the Internet as it does not take into account the infrastructural responsibility of search engines. It also shows that if the privacy regulation is not decided by this parliament and has to be brought back into the next EU parliament, there could be a window of opportunity to introduce a provision establishing the Austrian position that a search engine is not infringing if the original source is not protected against indexing with tags or other means.
Finally, the decision has merits as it forces all search engine operators to think about their social responsibility and how they can contribute to the infrastructure they are living from. Search engines are the only actors of the Web that do not participate in standardisation in any way. Elements to control the crawlers are scattered all over the place. Every search engine has their own tags. In such a situation it is very hard for a new legislative effort to provide the right framework to put control into the hand of end users. The decision encourages all search engine operators to convene and discuss about controls and how to use them. The timing is right: The IETF talks about the next generation HTTP and W3C is still working on the HTML5 web platform. There is a decision to make whether to use the RDFa framework or something else. But at the end of the day, if the search engine operators remain inactive in this field, I predict that the courts will inflict things upon them on a case by case basis. This makes lawyers happy as it is very expensive. I don’t know whether it will make the search engine operators happy