Like the compass, light bulb and other once-revolutionary tools, the search engine has taken its place in the barely noticeable background of contemporary life. Using Google Search might invite as much contemplation as tying a shoelace: something you do multiple times every day, expending only minimal mental effort—unless you spot some fraying.
In the past year Google released AI Overviews, a platform that answers search queries with artificial-intelligence-generated summaries. Enough of these summaries were so odd that the company addressed them in a blog post (while also noting user satisfaction was high). Searches for certain products, such as air purifiers, kicked up clouds of search-engine-optimized spam. And a judge of the U.S. District Court for the District of Columbia, Amit P. Mehta, ruled that the company used illegal methods to dominate the search engine market. Google plans to appeal that ruling after Mehta determines the company’s punishment, which could include issuing financial penalties or breaking off parts of the tech giant. On top of all that, Google also faces an antitrust lawsuit over its digital advertising practices.
“Google is getting worse, or at least we are seeing it become less useful,” says information scientist Jutta Haider of the University of Borås in Sweden, who, with her colleague Olof Sundin of Sweden’s Lund University, co-wrote the 2019 book Invisible Search and Online Search Engines. “All kinds of work-arounds are popping up” for users who want to tweak their searches. These could include adding the qualifier “site:sitename” to a query to restrict results to a specific website, or formatting Google settings to remove AI responses in favor of traditional hyperlinks.
On supporting science journalism
If you’re enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
Although a March 2024 study observed that major engines are struggling to keep spam out of searches for product reviews, Google still fared better in that analysis than its competitors, a Google spokesperson said in a statement e-mailed to Scientific American. The spokesperson added that Google has since adjusted its results to emphasize reviews from experts with firsthand experience and that “third parties consistently find Google to be of higher quality than other search engines.”
And if you would like to use one of those other search engines, your choices are somewhat limited: nearly 90 percent of U.S. searches go through Google. Consider the other search engines with a U.S. market share of greater than 1 percent, as measured by the Web traffic analyzer StatCounter. There are three. Microsoft’s Bing is used for about 7 percent of searches. Bing’s index, which is essentially an inventory of every website it is aware of, is a main source of information for another engine, DuckDuckGo (which is used for about 2 percent of searches). Also contributing to DuckDuckGo’s results are its web crawler—a bot that scans the Internet to help pull in relevant information—and data from niche search engines. Yahoo! Search (which also represents about 2 percent of searches) has been similarly dependent on Bing’s index since 2009. It, too, has its own crawler, named Slurp.
There are a handful of less popular options, such as Brave Search, which, like DuckDuckGo, emphasizes privacy and limits tracking of user data. Several search engines are aimed primarily at the non-English-speaking world: Google’s Russian-language competitor is Yandex; the Chinese equivalent is Baidu. Some newer search engines employ generative AI to provide answers. These include Perplexity, which has been accused of plagiarism by Forbes and other news outlets. Haider cautions that AI search engines, in general, can offer misleading answers because they lack a savvy human’s media literacy—leading to results that promote climate change denial, for instance.
Some search companies offer alternate ways to access data, such as Berlin-based Ecosia, which donates its profits to tree-planting organizations. Roughly every 50 searches pay enough to plant one tree, according to Pieter Van Midwoud, Ecosia’s chief tree-planting officer. (The exact number of searches needed depends on where in the world users are, he notes, and on how many ads they click.)
Ecosia provides results from Bing, Google and other external sources, but it does “not have much control on what results are served,” says Jade Devey, Ecosia’s head of global communications. It does, however, affix custom labels to certain links. A green leaf icon indicates companies “that have legitimate climate commitment,” Devey says, while “big polluters” are marked with a symbol shaped like an industrial smokestack.
Let Me Bing That for You
How effective are these other engines? They’ll almost certainly give you the answer you seek, if you’re trying to confirm a well-known fact or find a company’s website. “For the majority of queries, it really doesn’t matter which search engines you use,” says Dirk Lewandowski, a professor at the Hamburg University of Applied Sciences in Germany, who studies search engines and the behavior of those using them.
For a study published in 2022 in Proceedings of the Association for Information Science and Technology, Lewandowski and his colleagues compared Google results with those of DuckDuckGo, Bing and MetaGer, a German “metasearch” engine that aggregates data from other engines. Using software called the Result Assessment Tool (RAT), the scientists analyzed the top 10 results for more than 3,500 Google Trends queries in Germany and the U.S. between late 2021 and early 2022. People were searching for news, for events such as the Macy’s Thanksgiving Day Parade and for sports and celebrities (“Sandra Bullock” was trending at the time, the authors noted in the paper).
The study found that certain überpopular websites, such as Wikipedia, Instagram and IMDb, repeatedly appeared among the highest U.S. results. Overall, Google’s top 10 results overlapped with those of the other search engines by about 24 to 25 percent in the U.S. The nonGoogle search engine results were more similar to one another. Bing and DuckDuckGo had 64 percent of their answers in common, which Lewandowski attributes to the shared Bing index.
“For many searches, the difference between Google and Bing or even DuckDuckGo isn’t that big anymore,” Haider says. But Google, she adds, has powerful features the others don’t—Google Maps and YouTube—that are cleanly integrated into its results.
Lewandowski cautions that this study was limited by its reliance on trending searches. “We can rightfully expect that the overlap for rare queries is much lower than for the popular ones,” he says. If the question is too obscure for Wikipedia to answer, different search engines might seek answers in different corners of the Internet. In such cases, he recommends using multiple search engines to dig up more information.
When evaluating a search engine, Haider says she typically queries for a pair of two major cities, such as “Paris Madrid” or “Quebec Toronto”—and the results often put flight promotions at the top. “High-carbon practices [such as flights] are usually ranked higher,” she says. Then she’ll search for something like summer clothes or children’s clothes. Many engines have a “strong consumerist bias” and link prominently to clothing stores or commerce sites, she says, adding that her intent is to expose the engines’ invisible assumptions: “Cities have other relations than an airline,” Haider says. “Clothes have histories and cultures, and they can be swapped or mended.”
Searching for Diversity
Could the future offer a fuller menu of Internet search engines? Even without Google’s market monopoly looming over the field, establishing a new engine isn’t easy. Of a search engine’s parts, which include things such as web crawlers, ranking algorithms and the search interface itself, the index may be the most difficult to create. Google Search executive Pandu Nayak testified last year that Google’s index was “maybe” around 400 billion documents in 2020.
An index of this scale is “extremely expensive” to build and maintain, Lewandowski explains. It must be updated constantly and distributed in ways that permit access to users worldwide. “Nobody has the money to do it,” he says—except, of course, for Google and Microsoft. As a way to circumvent this, Lewandowski, as well as other information and computer scientists, have advocated for a publicly funded index of the Web. This could be the foundation for “thousands of different search engines,” he says. “That’s the way out.”
An open index is technically possible, Haider says, but the politics of building such a thing—isolated from government control or profit motives—remains an open question. In the meantime, she adds, preferences for a particular search engine can make a difference. “Web search engines improve when they are used,” Haider says. “The developers get feedback on what people are searching for, how the searches are going, what is being clicked on, and so on. This then helps to improve the system.” Even something that has faded into the background can change, and that’s especially true for search engines.