Questionable Search Engines: Why and How They Are Biased

Recently, Berlin called on Google to disclose the details of its secret algorithms that had allowed it to monopolize web search in Europe, in a move that was likely to be welcomed by competitors and fiercely resisted by Google. Actually, Claims of bias are nothing new to Google. According to The Economic Times, "pressure is growing on Google to lift the veil on some of its inner workings as European regulators investigate complaints that Google's search engine unfairly discriminates against certain websites" by the European Union.

Increasingly, the quality of Google's search results and business practices are being called into question, and almost all the blames are pointed toward the its algorithm, the “secret sauce recipe” that has enabled it to dominate the field of web page searching. In computer science, an algorithm is the description of an automated solution to a problem. Some algorithms are inherently biased, such as those designed to generate a random number, which exceed the capability of a computer, while most of them are artificial. The artificial bias of searching is often referred to the practice of causing a web page to rank highly in search engine results for unrelated or off-topic search terms by linking heavily, known as Google Bombing. So how and why algorithms of such searching engine are intentionally designed to be 'biased'?


1. A result of user preference

Search engines now aspire to develop "perfect recall" of all of their users. To do this, information providers must not just track their users, they must also "build technical infrastructures and business models that link individual sites into a suite of services, or an even broader ecosystem. And then create incentives for users to remain within it." (The Relevance of Algorithms, Tarleton Gillespie) Today we call the algorithms lies behind these behaviors "big data", storing user preferences in the Cloud, so any device with a network connection and a method to "identify" the user can use these preferences to provide most relevant and desirable results of searching.

Even though sometimes repetitively appeared relevant searching result and advertisement are pretty annoying, and excessively recording user preference is now considered to be immoral and potentially fatal to information security, I still believe that big data has a broad prospective in better one's searching experience, accompanied by more advanced governmental regulation, and solutions to ethical and security problems.

2. A result of economic and commercial processes

Baidu, the most influential search engine in China owning world's second biggest share of the search engine market, has always been controversial about its algorithms and its way of making profit "Baidu Popularization". Searching the word "bicycle" on Baidu, you will find that the top ranking results are almost all advertisement - links direct you to bicycle companies and their product pages. It also provides the ranking based on how much money those companies are paying to promote, and even forces some small companies to pay by putting their links far far behind where they should be.

Besides promoting its own business, by studying what types of ranking manipulations a search engine is using, a company can also provoke a search engine into lowering the ranking of a competitor's website by purchasing Google Bombing services for the website of a competitor. The attacker provokes the search company into punishing the competitor by displaying their page further down in the search results, forming an unhealthy and potentially illegal part of the booming technological community.

Google Bombing is also potentially a result from inter-companies competition. Broadly acknowledged as the first ever Google bomb came about back in 1999 when many people were still on dial-up connections. A search for "more evil than Satan himself" resulted in the Microsoft homepage as the top result. In 2011, Bing defined the search for “more evil than Satan himself” as 10^100 (a googol, the word the Google founders used as the basis of their company’s name). These definitions were obviously chosen by both teams, instead of results of errant algorithms.
In stead of being modified by users or the search engine itself, sometimes a third party, more specifically the website that wants to boost its popularity, can also make the algorithm or the search result biased. Owner of an Internet domain name could set up the domain's DNS entry so that all sub-domains are directed to the same server. In this way a page full of desired Google search terms, each linking to a sub-domain of the same site will be generate.

3. A result of political processes

During the 2004 U.S. presidential election, Google bombs were used to further various political agendas. Searching "miserable failure" on Google would link you to George W. Bush's White House biography and "waffles" to John Kerry's website. Names of some public figures were also being displayed when searching words like "scandal", "rumor" and "private photos". It was not until 2007 that Google changed their indexing structure so that Google bombs such as "miserable failure" would "typically return commentary, discussions, and articles" about the tactic itself, instead of prospectively depreciating individuals.

Google is also adapted to local policies of different countries and regions, especially in China and European Union. Censorship is designed to protect copyright of books, music and movies, as well as private information, adult content and unethical contents.


Is algorithm biased? In my opinion, the answer is YES. Algorithm embodies in search engines is human-written and human-oriented, composed with not only programmers' frosty codes, but also with their faith and opinions.