The Dutch data retention implementation law (31145) is on the plenary agenda of the Dutch Senate next Monday and Tuesday, 6 and 7 July 2009. So far, the Senate has been very critical of the implementation law, has held an expert meeting and asked two rounds of questions to the Dutch government expressing its reservations with regard to the fundamental rights of citizens and the effectiveness of the proposed data retention framework.
Our German neighbors are about to vote about a controversial legislative package introducing mandatory internet censorship at the network level. Netzpolitik has an excellent overview of the political debate until now. AK-Zensur, the Working Group on Internet Filtering and Censorship, has a comprehensive overview of arguments against the proposals. Germany would be the first EU country to adopt mandatory Internet filtering for Internet access providers. Some ISPs in some other European countries already filter on the basis of black lists provided to them by government agencies and/or private organizations. Here is part of Markus Beckedahl’s account of the action against the proposed censorship:
The net community did not only oppose the governments plans, but also made constructive suggestions how to deal with the problem of child pornography without introducing a censorship architecture and circumcising constitutional freedoms. The working group on censorship demonstrated the alternatives for instance by actually removing over 60 websites containing child pornographic content in 12 hours, simply by emailing the international providers who then removed this content from the net. The sites were identified through the black lists of other countries documented on Wikileaks. This demonstration underlines the protesters main arguments: instead of effectively investing time and efforts to have illegal content removed from the internet, the German government is choosing censorship and blocking – an easy and dangerous way out. The greatest fear of the protesters is that once in place, the infrastructure will be used to censor other forms of unwanted content, not only child pornography. German politicians already seem to be lining up with their wish-list of content to be censored in future – the suggestions ranging form gambling sites, Muslim web pages, “killer games”, and the music industry cheering up with the thought of finally banning pirate bay and p2p.
You can find a detailed linklist of the zensursula-debate here (in german).
In an important decision the French Constitutional Court has declared the core elements of the proposed three strikes law in France (Hadopi) unconstitutional, considering the freedom of expression and communication, the presumption of innocence, the right to privacy and confidentiality of communications.
This puts further pressure on the Council of Ministers to respect the EP’s adoption of amendment 138, requiring member states to respect the fundamental rights of internet end-users in the context of internet access. The Dutch minister of economic affairs recently wrote a letter to the Dutch parliament, stating the Dutch government did not consider the three strikes strategy to be acceptable. The letter also refers to the new (constitutonally problematic) Dutch notice and takedown code of conduct (my translation):
In the Dutch context, a procedure to deal with end-users with bad intentions is also not necessary, because, partly at the government’s urging, we have adopted a code of conduct on the basis of which internet providers, at the initiative of law enforcement authorities, take action themselves against criminal activity such as child pornography (notice and takedown code of conduct). It will have to be clear that the amendment does not interfere with this code of conduct. The Netherlands wants further clarification of the amendment before it can take a definitive point of view. In particular, it will have to become clear in which cases there should be a prior judicial decision and ion which cases not.
The Dutch government expects the Council to reject the Telecom package as adopted by the EP and the negotiations in third reading to be restricted to amendment 138.
This is a really great lecture! Julie Cohen manages to touch upon almost everything I am interested in, in about half an hour.
Although the positive tone is dominant, there seems to be a little discomfort discussing the initiative and the issues it is seeking to confront, for instance around the issue of working together with law enforcement and answering to subpoenas. Nicole Wong does answer the question about US law enforcement subpoenas, but to say that she is happy about it or deserves compliments for the information she gives would be an overstatement. Google and others should be much more transparent about their interaction with law enforcement in the US and abroad and hopefully GNI will provide the framework do do that. Now their log retention policy FAQ states:
Will governments be able to subpoena server log data after it is anonymized? Will anonymized data still be able to identify an individual user by cookie or IP address?
Google does comply with valid legal process, such as search warrants, court orders, or subpoenas seeking personal information. Logs anonymization does not guarantee that the government will not be able to identify a specific computer or user, but it does add another layer of privacy protection to our users’ data.
Will this policy change make it more difficult for law enforcement to prevent and detect
crime or child exploitation?
No, current laws allow the government to request that companies preserve user data. We regularly comply with such laws.
How many subpoenas for server log data does Google receive each year?
As a matter of policy, we don’t provide specifics on law enforcement requests to Google.
It’s not generally forbidden to provide these data, so let’s organize them and make them universally accessible and useful. UPDATE: Chris Soghoian issued a FOIA request relating to this kind of information.
The GNI was launched about half a year ago. It’s still looking for an executive director and the webiste does not give any information about the board of directors that is supposed to govern the GNI. In other words, formally, the GNI does not really exist yet. To stay positive, this means that once it really starts to function, much could come out of it.
According to the EU observer. And the European Commission is asked to scrutinize the deal from an antitrust perspective.
The motivation (in Dutch) of the preliminary judgment of the Court in Amsterdam about the liability of a website for a wrong suggestion in a Google search result snippet has been published. The problem was that a combination of two sentences ‘suggested’ that defendant Zwartepoorte, a car salesman, was bankrupt.
As was to be expected, the judge concludes that its judgment is very case specific. It clarifies that it involves a balancing act between the freedom of expression of websites and the damages suffered by the plaintiff. According to the Court, the duty of care to remove or restructure the content that caused a combination of snippets to appear in a Google search result existed because:
1. plaintiff had suffered damages because of the wrong suggestion;
2. defendant used the services of Google, had optimized the ranking of its website in Google, and profited from this (through advertisements);
3. Plaintiff had requested to do something about it a number of time, with reference to the damage it was suffering;
4. It was easy for defendant to make a simple change to the website that would have caused the snippet to change;
Ar these circumstances that case specific? I very much doubt it. And are they convincing?
Let us first consider the aspect of damage. There was proof that some people thought that plaintiff was bankrupt. They were calling and writing the company and even asking the regional newspaper in Zeeland, the PZC. There was no proof that this caused plaintiff to suffer material damage. It’s unclear whether the Court assumes there was any. From the lack of material damages in the verdict, it seems there was insufficient proof for this. The damage therefore consists of people thinking that defendant was bankrupt.
I wonder what is meant by the statement that defendant used the services of Google under 2. Normally, the inclusion of one’s website in Google does not imply the use of any service. It could mean that defendant used Google’s webmaster tools to optimize the site, so the webmaster should have been aware of Google and the importance of Google’s search results. The other arguments under 2 fit into this reasoning. Not only had plaintiff knowledge of Google’s search service, it optimized it and profited from a prominent listing because it attracted visitors, which in turn increased advertising revenues.
The argument under 3 increases the knowledge of defendant that something could be causing damages to plaintiff. Argument 4 relates to the costs and difficulty of avoidance to cause damage. Defendant did not control the combination of sentences in the search result but by restructuring its website or removing the word bankrupt from the particular page, it could cause the combination to disappear, according to the Court.
In other words, the Court puts itself in the website editor’s shoes. I find this highly problematic from a free speech perspective. And implicitly, the court assumes that there is not a lot of value in the website and the combination of information it contained.
The other basic problem with the Court’s reasoning is in the aspect of assigning the damage to the conduct (negligence) of defendant. This assignment of damages, i.e. ‘legal cause’, is a substantive legal test in (Dutch) tort law, which relies on the foreseeability or the remoteness of damages. It is with regard to this test that the court’s reasoning breaks down. Again, the damage consisted of people thinking that plaintiff was bankrupt. For legal cause to be fulfilled, these thoughts should be somehow reasonable. And freedom of expression makes the test of legal cause particularly important. One can not be liable for a completely wrong conclusion that is drawn from one’s statement. Google search results are search results, they are not an independent sources of information. In particular, Google’s snippets are not an authoritative source of information about bankruptcy. If one wants to check whether a company is bankrupt, there are specific sources where this information can be found. And people that search for [defendant bankrupt] were probably acting on a rumor already. They in particular should be expected to check the underlying information. It also follows that the conclusion that defendant could easily prevent the damage is a little too easy. In principle, it’s a company’s own responsibility to address unreasonable conclusions about its business, by its own statements or conduct.
Of course, the judge had to clarify why defendant, a website, could be held responsible for a combination of (perfectly legal) sentences in a search result which it did not control. The Court argues that it might be true that the website had no control over the functioning of Google but suggests that these questions about the opacity of Google’s functioning should be addressed in a broader context. With reference to the arguments under 2 above, it then concludes that defendant had its own responsibility. The Court evades the question whether Google would have been liable as well, i.e. if it would have refused to take down the result after a request to do so. I doubt it. First, Google would have hired better lawyers, that would have made a better case against it. And second, the judgment fits well into the paradigm to address ‘problems’ in search engines by addressing the sources of information. But seriously, to make websites liable for the impact of legal publications in search engines on unreasonable end-users is a step too far.
It will be interesting to see whether there will be an appeal. Even if there isn’t, I think we can safely assume that this judgment will remain ‘unique’. UPDATE: Defendant appealed, this time with a lawyer.
Via Arnoud Engelfriet.
A Dutch judge has concluded (in a preliminary judgment) that a website is liable for a wrong suggestion in the snippet text of ‘its’ Google search result. The judge has ordered the webmaster to change its website in a way that will stop the wrong suggestion from showing up in Google. Note that the text on the website itself was not unlawful. The problem was that a search for [zwartepoorte failliet] (= CompanyName bankrupt] resulted in the following snippet text:
Volledige naam: Zwartepoorte Specialiteit: BMW … Dit bedrijf is failliet verklaard, het is overgenomen door het motorhuis Ik heb bij Boot Rialto gewerkt …
Complete name: Zwartepoorte Specialiteit: BMW … This company has been declared bankrupt, it has been acquired by the motordealer I have worked for Boat Rialto …
One way to do as the judge ordered is to take down the complete page with information about the company Zwartepoorte. That is what seems to have happened now. Another solution would be to use code language such as is common with the word pr0n.
Clearly the combination of words and the quote in question will show up all over the Web so I don’t see any benefits for the company in question to have won this case. And of course, the judgment (the reasoning of the Court will be published on 27 May 2009) does not make any sense, from a tort law or from a technology perspective. A reason could be that the defendant appeared in person.
I am fascinated by the idea that the existence of search engines, and Google in particular, gives websites some kind of qualified responsibility with regard to the content on their website. I see more and more examples of this, for instance the idea that personal data should be prevented from being indexed through the use of robots.txt instructions.
The French Parliament has adopted the Hadopi Act, containing the controversial three strikes provisions for illegal filesharing. La Quadrature has commentary.
In a recent hearing on the ‘future of journalism‘, news industry officials argued for an exception on antitrust restrictions. The idea is that this would allow an industry conversation that could help it to acquire some leverage to squeeze money out of news search services and aggregators.