Search engines have gotten so good at providing information, that we forget to question how that information was selected for us, and if that selection could be biased. Search engines use algorithms to sift through billions of web pages and rank them based on relevance. But who gets to decide what’s relevant?
Google, Bing, and other mainstream search engines use a variety of factors to assess relevance, which in turn shapes our interactions with the digital world. Their selection of results is inherently biased—there’s no way of getting around it. Understanding the nature of this bias and considering alternative search engines can help you get the most out of search.
What does bias mean in the context of a search engine?
Search engines want to show results that people will click on, which means that they are biased towards engagement. When people click on a particular search result and spend time on the page, they send a signal to the search engine that that result was relevant and engaging. This signal makes the search engine more likely to surface that result again. But is engagement always a signal of quality content?
Consider, for example, the top ranking results for personal health queries. Advertising driven sites like WebMD often outrank government sites like National Institutes of Health. This is because WebMD is better at getting people to read through their content and click around.
“It doesn’t matter what symptom I have, ad-supported medical sites generally tell me I’m dying”, explains Sridhar Ramaswamy, co-founder of Neeva. “That’s because they have an ad-driven model. They tend to sensationalize everything, because they want you to click around.”
The more time you spend on sites like WebMD or Healthline, the more ad revenue they make, and the higher their pages rank in search results. Government sites like NIH, on the other hand, have very little incentive to increase engagement, since its model is not based on advertising. Although the NIH provides more factual information, Google is biased towards melodramatic WebMD.
The same goes for sites that review products. You might be looking for the best sneakers, but the top results are likely sponsored by the very products they recommend. These sites have a big incentive to get you to interact by clicking on a link to buy something quickly. This helps them rank high, get more views, and pull in advertising money—without providing real service to the consumer.
Other types of bias in search
There are other factors that can contribute to bias in search results.
- Location. Mainstream search engines use your location to provide you with more personalized results. That’s why, if you search for “larb”, you’ll not only see recipes and a Wikipedia entry for the Lao meat salad, but a list of nearby restaurants that serve larb. Many people enjoy this feature, but it does mean that search results differ depending on where you are in the world.
- Search history. Another aspect of personalization is the way search engines use your browsing history to serve you more relevant results. You may appreciate the way Google knows that you, an oceanographer, mean “dolphins” the animal, not “Dolphins” the Miami football team. But using your search history to personalize results can sometimes create the phenomenon of a “filter bubble”.
- Offensive content. Search engines may deliberately filter out content that is illegal—which they are legally compelled to do, which is why there’s no such thing as an “uncensored” search engine—or potentially offensive. Google, for example, altered its algorithms after it received complaints about the antisemetic and racist results that appeared when searching for “jews'' and “black girls”.
What are popular search engines like Google biased towards?
The algorithms that assess, sort, and rank results for search queries are inherently biased—that is, they give priority to results that fit certain criteria.
Google’s algorithms are not available to the public, but they do share best practices, which reveal the types of content they are biased towards. “Google-friendly” sites are those that:
- Include keywords. Google’s algorithms prioritize sites that share detailed, useful information. How does an algorithm determine what’s useful? By looking for specific keywords that it believes are relevant to the search query. That means the high-ranking results you see are rich in keywords, but not necessarily in information.
- Get backlinks. To identify content that is both popular and authoritative, Google prioritizes pages that a lot of other pages link to. Websites with a high volume of content can manufacture the appearance of authority by linking to its own pages. This explains why when you search for a recipe, the top results aren’t necessarily from the sites with the best recipes, but the ones with the most recipes.
What about ads?
Advertising on the search engine itself adds another layer of complexity. Most search engines make their money through online ads, and it’s difficult to tell exactly how those ads impact users’ searching experience. Google advertising accounts for about 80 percent of revenue for its parent company, Alphabet. Even alternative search engines, like DuckDuckGo, the most popular private search engine, are powered by ads.
When it comes to bias, ad-powered search poses two main issues. The first is conflict of interest. Ad-supported search engines allow advertisers to buy their way to the top of the results page for any given search term. This is especially true for retail queries—that is, queries for which the intent is to buy something.
It is a relatively well known fact that, for a given query, the quality of the ad shown up on top is lower than the quality of the first organic result. This bias towards paying results is a serious source of bias in commercial search engines.
The bias created by the presence of ads is compounded by the fact that ads and organic results are often indistinguishable, especially when searching on a mobile device. In January 2020, Google changed the way paid advertisements look on the page, causing many users to unknowingly click on ads. The backlash was so severe that Google reversed some of the changes within the week.
How to get the most out of search
Here’s how to get better, less biased results when using search engines.
- Check the source. Just because a website ranks high on search results does not make it a reliable source. When looking for product recommendations, seek out independent research sites like Consumer Reports and Wirecutter, which do not accept payment for their reviews. For other queries, check government sites, nonprofits, and encyclopedias.
- Do a fact-check. Fact checkers are people whose job it is to check whether something that appears in print is true. When possible, seek out publications and other sources that employ fact checkers to review articles. Some sites name their fact checkers, or discuss their fact checking practices on their “About” pages. You can also become your own fact-checker by using your search engine to confirm facts with primary sources and other trusted sources. You can also use non-partisan fact-checking websites such as FactCheck.org, Politifact.com, and Snopes.com.
- Beware the “echo chamber” effect. Mainstream search engines like Google use your search history to personalize your results, so it’s possible that your search results could start to reflect your own bias. That’s one reason why people are moving away from Google and towards search engines that give them more privacy options.
- Try an alternative search engine. Different search engines will yield different results. Some users prefer the depersonalization of the results offered by anonymous search engines. Others prefer an ad-free experience like Neeva, where relevant web pages don’t have to compete with ads for space at the top of the results.
Top alternative search engines
All search engines use algorithms to filter and sort results, but some offer alternatives to Google search that can reduce different types of bias.
- Anonymous search engines. If you’re concerned about getting stuck in a “filter bubble,” you may want to consider an anonymous search engine. Anonymous search engines like DuckDuckGo and Startpage do not collect users’ personal information. Since they don’t know anything about you, anonymous search engines eliminate the possibility of location or search history bias. What they don’t eliminate is the conflict of interest that comes from having an ad-based model. Their results are also subject to engagement bias—you’ll find this one nearly impossible to avoid.
- Metasearch Engines. Metasearch engines, like Searx and Metager, pull search results from hundreds of different sources, so they reflect a different set of biases from Google and Bing. Metasearch engines can be a bit clunky to use, but they’re an interesting option for anyone who wants to delve deeper into the possibilities of search.
- Private search engines. As the only subscription-based private search engine, Neeva eliminates the conflict of interest that plagues so many advertising-based search engines. Neeva balances privacy with personalized search by collecting some data to improve user experience, while giving users control over their data.
Neeva is the world’s first private, ad-free search engine, committed to showing you the best result for every search. We will never sell or share your data with anyone, especially advertisers. Sign up today and try Neeva for yourself: neeva.com/signup.