• TIPLPR

Curbing Search Bias: Lessons to be learnt from the EU-P2B Regulations?

Search Engines- The “Gatekeepers” of the Internet

A “netizen”, a citizen of the internet has access to indeterminable quantities of information. Search engines act as gatekeepers or navigators by directing its users to the most relevant URLs, filtering the rest, with an input of a few keywords. Alternatively, a search engine is a platform- an information service intermediary which connects the “consumer” who is looking for relevant information with the “corporate website user”, the maker of the webpage. Search engines may exercise significant influence on economic, political and personal choices due to their ubiquity, oligopoly in search-related services and the increasing reliance of the current generations on the internet as their primary source of information.

“(Google’s results) increasingly determinant of the ways we author our lives-of our personal autonomy… We are not Google’s product; we are its subjects” [1]

A Biased Objective Decision-maker?

Search engines are often seen as impartial decision makers providing algorithm generated “organic” results. This token presumption of neutrality is important since the users associate impartial results with relevant results and therefore continue to use the search engine. Nevertheless, search-related services are credence goods whereby the user cannot ascertain its utility precisely even after using it. This is an inherent dichotomy- and also perhaps why the users will not shift to an alternate even if there are “tweaks” in the listing of the relevant results.

Search engines give prominence to certain results over others. A “systematic favouring system” has been developed, either by express algorithmic suppression or by including false indicators in the algorithm itself. Rankings on the search engine results page confer an advantage and translate to a higher click-through rate. This submission is fortified by the EU Decision against Google [Para 454-461].

The neutrality of Search Engines, in my opinion, is compromised wherein a conscious deviation from the most relevant results is possibly made due to three factors: i) Ad-service; ii) “Own-service” and iii) Socio-Ethical factors.

Among these three factors- the Ad-service and Socio-Ethical Factors are not subjected to academic scrutiny for the purpose of this blog. Since the first competition law related action against search engines by the Federal Trade Commission in 2002, the principle of clear distinguishing (“paid search results and other forms of advertising should be clearly distinguishable from natural search results”) is well fortified and followed by search engines in all jurisdictions. This lessens the impact of Ad-service bias on consumers. Similarly, socio-ethical bias by search engines is, in my opinion, reasoned and has a positive effect. For instance, sites having largely pirated content are pulled out or down-listed on the search engine results page [Para 3.20 (c)] or engines change racist results generated by its algorithms.

Own-service bias has arisen out of the evolution of search engines; with four factors to be noticed- i) Google is the dominant provider of online search engine; ii) the architectural structure of search engines has changed to both “horizontal search” (generalised results) and “vertical search” (targeted results); iii) “Portalisation” (various zones on the search engine results page) and from merely directing to organic results to offering the answers to the query itself; and iv) search engines have expanded to own certain offerings (per instance, Google’s ownership of YouTube).[2] This changing landscape brought the well-founded allegations as to the search engine exploiting third-party content and changing rankings so as to keep the “consumers within its own offerings”. The most apt illustration for this is Daniel Crane’s allegory for a search engine being a librarian who earlier gave objective book recommendations- to now becoming the author of a few books, which is bound to produce some bias. These tripartite causes when contrasted with the popular claims of non-biasness present an oxymoron: Search engines are biased objective decision-makers.

Indian Jurisprudence- Curbing Search Bias

The solitary decision on search bias by the CCI is the landmark Google-Matrimony case whereby Google was fined 5% of average of total earnings of three years for, inter alia, abuse of dominance pertaining to search bias favouring its own service- Google Flights. The Commission preliminarily held Google to be dominant in the relevant market and outlined a self-imposed restraint from undue interference in the design of the search engine results page [Para 204-205]. It recognised the relation between ranking and higher click-through rate whereby web content providers compete to be privy to the top ranks [Para 244]. Further, relying on the statements depicting commitment to relevance in its search results by Google, it held that consumers presume rankings reflect order of relevance and deviation from this rules constitutes misleading the consumers and hindering market access to other competitors [Para 240].

Additionally, in foreign jurisdictions, entities have made claims of tortious remedies on their dissatisfaction with the algorithmic ranking on certain websites, for the loss of prospective economic advantage. [3] This is yet to have any precedent in India.

Shortcomings

Firstly, Indian jurisprudence regarding search engine biases and the digital economy (albeit there are a few decisions by the CCI) is not fully developed. There is also an innate hesitation by the competition watchdog to intervene in the digital space, fearing it being in the nascent stage or that such intervention would lead to curbing innovation and consequently, cause disadvantages to consumers.

Secondly, via the linkage of liability of search bias with abuse of dominance, the Commission has inadvertently given a “free-pass” to other search engines which are although not dominant but present bias in their search engine results page.

Lastly, drawing from the second point, Indian law against search bias has developed like a horse with blinkers- with the focus only on competition law remedies. The right to internet has been recognised, but there is an apparent lacuna in regulation, which could restrict search biases and ensure that the access to internet is truly “fair”. Such regulations could also reduce the “opinion manufacturing” externalities that a search engine could induce, thereby allowing individuals to reclaim personal autonomy.

Remedy Models for the Regulator and EU-P2B Regulations- Examples India could imbibe.

A restrained yet active approach by CCI, the regulator, is warranted to ensure a competitive, fair, and transparent online ecosystem which increases consumer welfare. There are certain remedy models that Benjamin Edelman suggested; their efficacy is evaluated hereinafter:

§ Avoiding improper ranking factors- The algorithms are formulated, designed and owned by search engines who are consequently in control of these algorithms. This was highlighted by the Investigation Report of the Director General [Google-Matrimony case, Para 174]. The search engines are in a position to intervene and impact the relevance and rankings of results. Therefore, there must be disclosure regulations and improper ranking factors- say, factors preferring links related to the search engine, should be avoided.

§ Compulsory disclosures of conscious readjustments and suppression of algorithmic results- whenever there is deviation from selection on the basis of relevance via algorithmic rules of general application, the facts and motivations behinds such notifications should to be reported to the regulator.

§ Treating integrated components as bundling- treat portalisation and services like “one-boxes” as bundled with the search engine. Endelman suggests that no default option (which favours the search engine) should be allowed, rather the consumer should have options to choose their service provider. While this suggestion is well-rooted in economics, yet its usefulness in the Indian context is limited since the CCI seems to have accepted Google’s contentions that such instructions are firstly not feasible due to lack of technical systems and would cause time lags which are against consumer interests. [Google-Matrimony case, Para 223]

The EU Regulations on Platform-to-Business Relations (EU P2B) recently became applicable from 19 July 2020 and attempt to regulate the relationship between corporate website users and the online search engine. Preliminarily, it recognizes that rankings have an important impact on consumer choice and commercial success of the corporate website users and offers protection even in the absence of contractual relationship. It defines the term “online search engine” (undefined in India) and makes it technology-neutral and inclusive. Article 5 directs search engines to provide a publicly available description of its ranking parameters, their relative importance and access to notification to the corporate website user which alters rankings. Article 7 requires a description of any differentiated treatment to the consumers between the websites and services offered by the provider themselves (or under their control) and other corporate website users. These regulations must be carefully analysed by the CCI, and treated as an example which must be imbibed in Indian jurisdiction as well.

Conclusion

Given the potential of search engine bias to dictate personal choices of the consumers, it is a pressing concern that Indian law is grossly inadequate. Google’s bias towards its own services is not a solitary occurrence but is rather a practice adopted by search engines, yet under Indian law, it is the only one which is actionable.

The regulator, keeping in mind the interests of the consumers, must intervene to assimilate measures from the EU such as public disclosure of ranking factors, inclusion of improper ranking factors in the algorithm and differentiated treatment. Nevertheless, this must not be done via an interventionist approach but rather via a liberal/restrained disclosures approach.

[1] M. Thompson, In Search of Alterity: On Google, Neutrality, and Otherness, 14 TUL. J. TECH. & INTELL. PROP. 137, 140 (2011). [2] Observed with specific reference to Google by E. Goldman, “Revisiting Search Engine Bias”, 38 Wm. Mitchell L. Rev. 96, 102-105 (2011). [3] Christopher Langdon v. Google Inc, 474 F Supp 2d 622 (D.Del. 2007); Kinderstart.com LLC v. Google Inc, Case No. C 06-2057 (N.D. Cal. 2006).

The title image has been used only for creative visualisation and doesn't intend to solve any other purpose.


The article has been written by Purva Anand who is a penultimate year student at Gujarat National Law University, (GNLU), Gujarat.