Monday, January 6, 2014

Meta Search Engines Likely to Infringe Database Right by Data Scraping

The Christmas time drowned out publication of a particularly interesting CJEU decision - Innoweb C‑202/12 - dealing with the EU-specific database right and its applicability to meta search engines.

The preliminary reference before the CJEU came from a Dutch proceedings of Wegener group against Innoweb BV and its operation of a dedicated car meta search engine that enables searches to be carried out on third party websites, including Wegener’s website, where a collection of car sales advertisements (‘car ads’) is displayed.

8        Through its website (‘the AutoTrack website’), Wegener provides access to an online collection of car ads, together with a list, updated daily, of 190 000 to 200 000 second-hand cars. Approximately 40 000 of those advertisements are to be found only on the AutoTrack website. The other advertisements can also be found on other advertising sites. With the help of the AutoTrack website search engine, the user can carry out a targeted search for a vehicle on the basis of various criteria.

9        Via its website,, Innoweb provides a meta search engine dedicated to car sales (‘GasPedaal’). A ‘meta search engine’ uses search engines from other websites, transferring queries from its users to those other search engines – a feature which differentiates meta search engines from general search engines such as Google. A meta search engine described as ‘dedicated’ is designed to enable searches to be made in one or more specific subject areas. GasPedaal is such a dedicated meta search engine, designed to search for car ads: through a single query on GasPedaal, the user can simultaneously carry out searches of several collections of car ads listed on third party sites, including the AutoTrack website.

10      By means of the GasPedaal dedicated meta search engine, it is possible to search through the AutoTrack collection on the basis of different criteria, including not only the make, the model, the mileage, the year of manufacture and the price, but also other vehicle characteristics, such as the colour, the shape of the chassis, the type of carburant used, the number of doors and the transmission and, second, ‘in real time’, that is to say at the time when a GasPedaal user enters his query. GasPedaal carries out that query in ‘translated’ form; in other words, it translates the query into the format required for AutoTrack’s search engine.

11      The results thrown up by the AutoTrack website – that is to say, cars meeting the criteria chosen by the end user – which are also to be found on the results pages of other sites are merged into one item with links to all the sources where that car was found. A webpage is then created with the list of the results thus obtained and merged, which shows essential information relating to each car, including the year of manufacture, the price, the mileage and a thumbnail picture. That webpage is stored on the GasPedaal server for approximately 30 minutes and sent to the user or shown to him on the GasPedaal website, using the format of that site.

12      The total number of advertisements on websites searched through GasPedaal is approximately 300 000.

13      Every day, GasPedaal carries out approximately 100 000 searches on the AutoTrack website in response to queries. Thus, approximately 80% of the various combinations of makes or models listed in the AutoTrack collection are the object of a search daily. In response to each query, however, GasPedaal displays only a very small part of the contents of that collection. In every case, the contents of those data are determined by the user on the basis of criteria which he keys into GasPedaal.
The referring court basically asked whether an owner of a website with car ads from users can prevent a third party from data scraping its database of user car ads and thus effectively prohibit the meta search engine operation.

After several remarks on how technologically dedicated (real time) meta search engines differ from general search engines like Google or Bing (§ 24-29), the Court court, relying mainly on contextual (§ 33-34) and teleological arguments (§ 35-36), comes to the following conclusions:

37      In the light of that purpose, the concept of ‘re-utilisation’ as used in Article 7 of Directive 96/9 must be construed as referring to any act of making available to the public, without the consent of the database maker, the results of his investment, thus depriving him of revenue which should have enabled him to redeem the cost of the investment (see The British Horseracing Board and Others, paragraph 51). ‘Re-utilisation’ accordingly refers to any unauthorised act of distribution to the public of the contents of a protected database or a substantial part of such contents (see The British Horseracing Board and Others, paragraph 67; Case C‑545/07 Apis-Hristovich [2009] ECR I‑1627, paragraph 49; and Football Dataco and Others, paragraph 20). The nature and form of the process used are of no relevance in this respect (Football Dataco and Others, paragraph 20).
Note that tautology par excellence in CJEU language: "re-utilisation .. refers to any unauthorised act of distribution" = it is like saying that a concept of prohibited acts refers to what is prohibited under the law.

38      The second part of the definition given in Article 7(2)(b) of Directive 96/9 – ‘by the distribution of copies, by renting, by on-line or other forms of transmission’ – and, in particular, the alternative ‘or other forms’ also make it possible to construe that definition broadly, by reference to the objective of Article 7, considered in paragraphs 35 and 36 above.
The objective being "to ensure that the person who has taken the initiative and assumed the risk of making a substantial investment in terms of human, technical and/or financial resources in the setting up and operation of a database receives a return on his investment by protecting him against the unauthorised appropriation of the results of that investment" [again? protecting against unauthorisied?].

39      As regards the activity of the operator of a dedicated meta search engine such as that at issue in the main proceedings and, in particular, the facet of that activity that is of relevance to the present case – that is to say, making available on the Internet a dedicated meta search engine for translating queries (keyed into that meta search engine by end users) into the search engines of the databases covered by the service of the meta search engine in question – it should be noted that that activity is not limited to indicating to the user databases providing information on a particular subject.

40      The purpose of that activity is to provide any end user with a means of searching all the data in a protected database and, accordingly, to provide access to the entire contents of that database by a means other than that intended by the maker of that database, whilst using the database’s search engine and offering the same advantages as the database itself in terms of searches, as can be seen from paragraphs 25 and 26 above. The end user no longer has any need, when researching data, to go to the website of the database concerned, or to its homepage, or its search form, in order to consult that database, since he can consult the contents of that website ‘in real time’ through the website of the dedicated meta search engine.

41      That activity on the part of the operator of a dedicated meta search engine such as that at issue in the main proceedings creates a risk that the database maker will lose income, in particular the income from advertising on his website, thereby depriving that maker of revenue which should have enabled him to redeem the cost of the investment in setting up and operating the database.

42      As the end user no longer has any need to proceed via the database site’s homepage and search form, it is possible that the maker of that database will generate less income from the advertising displayed on that homepage or on the search form, especially to the extent that it might seem more profitable for operators wishing to place advertisements online to do so on the website of the dedicated meta search engine, rather than on one of the database sites covered by that meta engine.
I find this finding to be a bit puzzling. You can try the disputed meta search engine yourself here. If this is the form in which it worked before the litigation, then I see only regular list of results leading to original websites. If a customer wants to see the car ad, he needs to consult the website of the source with all its advertising. To me, it seems that operation is no different than one of flight tickets (e.g. or general price comparison websites (e.g.

43      As regards, furthermore, database sites displaying advertising, sellers – aware that, with the dedicated meta search engine, searches will be made simultaneously in several databases and duplications displayed – may start placing their advertisements on only one database site at a time, so that the database sites would become less extensive and therefore less attractive.

44      The risk that the making available on the Internet of a dedicated meta search engine such as that at issue in the main proceedings deprives the database maker of revenue cannot be ruled out by force of the argument that, in order to have access to all the information relating to a result found in a database – that is to say, in the case before the referring court, access to all the information on a car featured in an ad – it is still necessary, as a rule, to follow the hyperlink to the original page on which the result was displayed.
OK, so it seems that operation is really like one of price general comparison websites. Then what kind of revenue will be lost by the operation of a meta search engine? I can understand a business reasons when Raynair was suing screen scraping websites as it was trying to build an image of cheapest European airline. The image that meta search engines could effectively undermine. But the database right can not protect any kind of business model build around the collection and subsequent provision of databases.
45      First, the information displayed by the dedicated meta search engine enables the end user, to some extent, to sort the results found and to decide that he does not need further information on a particular result. Second, it is possible that the end user will access more detailed information on a result without following the link to the database concerned, if that result is thrown up by a number of databases covered by the dedicated meta search engine, since that search engine reveals duplication by grouping all doubles together.
So what? I don't see why it is a "problem" that a user will decrease its transactions costs in search for useful information. To the contrary, I would list it as a benefit of the service. The viewpoint of the database owner can not be the only decisive one.

46      Of course, the protection under Article 7 of Directive 96/9 does not cover consultation of a database (see The British Horseracing Board and Others, paragraph 54, and Directmedia Publishing, paragraph 51). In consequence, if the database maker makes the contents of that database accessible to third parties, even if he does so for a consideration, his sui generis right does not enable him to prevent such third parties from consulting that database for information purposes (see The British Horseracing Board and Others, paragraph 55, and Directmedia Publishing, paragraph 53).

47      However, it is important to note that the activity of the operator of a dedicated meta search engine such as that at issue in the main proceedings does not constitute consultation of the database concerned. That operator is not at all interested in the information stored in that database, but he provides the end user with a form of access to that database and to that information which is different from the access route intended by the database maker, whilst providing the same advantages in terms of searches. By contrast, it is the end user keying in a query in the dedicated meta search engine who consults the database by means of that meta search engine.

48      Moreover, the relevant aspect of the activity of the operator of a dedicated meta search engine such as that at issue in the main proceedings – that is to say, making that search engine available on the Internet – comes close to the manufacture of a parasitical competing product as referred to in recital 42 of the preamble to Directive 96/9, albeit without copying the information stored in the database concerned. In view of the search options offered, such a dedicated meta search engine resembles a database, but without having any data itself.

49      It is sufficient for the end user to go to the website of the dedicated meta search engine in order to gain simultaneous access to the contents of all the databases covered by the service of that meta engine, as a search carried out by that meta engine throws up the same list of results as would have been obtained if separate searches had been carried out in each of those databases which, however, are presented using the format of the dedicated meta engine’s website. The end user no longer has to go to the website of the database, unless he finds amongst the results displayed an advertisement about which he wishes to know the details. However, in that case, he is directly routed to the advertisement itself and, because duplicate results are grouped together, it is even entirely possible that he will consult that advertisement on another database site.

50      It follows from the foregoing considerations that the act on the part of the operator of making available on the Internet a dedicated meta search engine such as that at issue in the main proceedings, into which it is intended that end users will key in queries for ‘translation’ into the search engine of a protected database, constitutes ‘making available’ the contents of that database for the purposes of Article 7(2)(b) of Directive 96/9.

51      That ‘making available’ is for ‘the public’, since anyone at all can use a dedicated meta search engine and the number of persons thus targeted is indeterminate, the question of how many persons actually use the dedicated meta engine being a separate issue.

52      Consequently, the operator of a dedicated meta search engine such as that at issue in the main proceedings re-utilises part of the contents of a database for the purposes of Article 7(2)(b) of Directive 96/9.
The Innoweb C‑202/12 decision is in a stark contrast with decision Automobil­ Onlinebörse I ZR 159/10 handed down by the German Federal Supreme Court (BGH) in 2011. Unlike CJEU, BGH took position (§ 26) that a provider of a software that enables comparison of different websites with car ads does not use that data himself and hence can not be a direct infringer. Instead, BGH focused on conduct of users and their possible violation of the database right of scraped websites (so did Dutch court in the reference "makes it possible for the public to search"). This led BGH to rejection of an infringement of a database right as to both re-utilization of substantial part of the database and systematic re-utilization of its insubstantial part. As seen above, the CJEU without any arguments, considered similar real time scraping to be acts carried out directly by the meta search engine operator. The Court basically implicitly attributed all the searches to the operator of the website. One can only speculate about the reasons. It however shows, how thin the ice between primary and secondary liability under different intellectual property rights can be.

Even more troubling that above doctrinal issue is a practical impact of the decision. This decision effectively outlaws operation of most of the socially beneficial websites that help consumers to compare the prices or qualities of different goods across the Internet. Unlike general search engines that channel in more traffic than they could possibly take out, the comparison websites are very likely to be nuisance for at least bigger providers/sellers, which probably receive less than loose. I am not sure if it is a good innovation policy for the EU to make operation of such websites dependent on mere tolerance by the big players. Especially when smaller competitors are possibly greater beneficiaries of this comparison websites. It is not that I would favor all forms of meta search engines, but this is a too broad ban of innovative services, which EU can not afford.

PS: interestingly enough: the decision was handed down by the Fifth Chamber with alleged help of General Advocate P. Cruz Villalón. His opinion, however, if delivered at all, was not published until today [reader Juan spotted in the ruling that the Court decided to proceed to the decision without the Opinion].


Juan José Marín said...

Hi Martin, according to English version of the Innoweb decision, "having decided, after hearing the Advocate General, to proceed to judgment without an Opinion". This is the reason why the General Advocate opinión is not published!

Huťko said...

Many thanks Juan for spotting this! It somehow slipped my attention.