Power of Search Engines: Some Highlights of Berlin Workshop
ø
I’ve spent the past two days here in Berlin, attending an expert workshop on the rising power of search engines organized by Professor Marcel Machill and hosted by the Friedrich Ebert Stiftung, and a public conference on the same topic.
I much enjoyed yesterday’s presentations by a terrific group of scholars and practitioners from various countries and with different backgrounds, ranging from informatics, journalism, economics, and education to law and policy. The extended abstracts of the presentations are available here. I presented my recent paper on search engine law and policy. Among the workshop’s highlights (small selection only):
* Wolfgang Schulz and Thomas Held (Hans Bredow Institute, Univ. of Hamburg) discussed the differences between search-based filtering in China versus search engine content regulation in Germany. In essence, Schulz and Held argued that procedural safeguards (including independent review), transparency, and the requirement that legal filtering presupposes that the respective piece of content is “immediately and directly harmful” make the German system radically different from the Chinese censorship regime.
* Dag Elgesem (Univ. of Bergen, Department of information science) made an interesting argument with regard to the question how we (as scholars) perceive users as online searchers. While the shift from passive consumers to active users has been debated in the context of the creation/production of information, knowledge, and entertainment (one of my favorite topics, as many of you know), Dag argues that online searchers, too, have become “active users” in Benkler’s sense. In contrast, so Dag’s argument, much of our search engine policy discussion has assumed a rather passive user who just types in a search term and uses what he gets in response to the query. Evidently, the question of the underlying conception of users in their role as online searchers is important because it impacts the analysis whether regulatory interventions are necessary or not (e.g. with regard to transparency, market power, and “Meinungsmacht” of search engines.)
* Boris Rotenberg (DG Joint Research Center, European Commission, Sevilla) linked in an interesting way the question of the search engine user’s privacy – as expression of informational autonomy – with the user’s freedom of expression and information. He argues, in essence, that the increased use of personal data by search engine operators in the course of their attempts to personalize search might have a negative impact on freedom of information in at least three regards. First, extensive use of personal data may lead to user-side filtering (Republic.com scenario). Second, it might produce chilling effects by restricting “curious searches”. Third, personalization tends to create strong ties to a particular (personalized) search engine, hindering the user to use alternative engines (“stickiness”-argument).
* Benjamin Peters (Columbia University) used the Mohammed cartoon controversy to explore three questions: (1) As to what extent do search engines eliminate the role of traditional editors? (2) Do algorithms have any sort of in-built ethics? (Benjamin’s answer, based on David Weinberger’s notion of links as acts of generosity: yes, they have). (3) What are the elements of a “search engine democracy”?
* Dirk Lewandowski (Department of information science, Heinrich-Heine Univ.) provided a framework for assessing a search engine’s quality. He argues that the traditional measurement “precision” – as part of retrieval quality – is not a particularly useful criterion to evaluate and compare search engines’ quality, because the major search engines produce almost the same score on the precision scale (as Dirk empirically demonstrated.) Dirk’s current empirical research focuses on the search engine’s index quality, incl. elements such as reach (e.g. geographic reach), size of the index, and actuality/frequency of updates.
* Nadine Schmidt-Maenz (Univ. of Karlsruhe, Institute for Decision Theory and Management Science) presented the tentative results of an empirical long-term study on search queries. Nadine and her team have automatically observed and analyzed the live tickers of three different search engines and clustered over 29 million search terms. The results are fascinating and the idea of topic detection, tracking, and – even more interestingly – topic prediction (!) highly relevant for the search engine industry, both from a technological and business perspective. From a different angle, we also discussed the potential impact of reliable topic forecasting on agenda-setting and journalism.
* Ben Edelman (Department of Economics, Harvard Univ.) empirically demonstrated that search engines are at least in part responsible for the wide spread of spyware, viruses, pop-up ads, and spam, but that they have taken only limited steps to avoid sending users to hostile websites. He also offered potential solutions to the problems, including safety labeling of the individual search results by the search engine providers, and changes in the legal framework (liability rules) to create the right incentive structure for search engine operators to contribute to overall web safety.
Lot’s of food for thought. What I’d like to explore in greater detail is Dag’s argument that users as online searchers, too, have become highly (inter-)active, probably not only in the sense of active information retrievers, but increasingly also as active producers of information while being engaged in search activities (e.g. by reporting about search experiences, contributing to social search networks, etc.)
