How Do Google’s Search Algorithms Work?
Google is a synonym for doing a search online. Google is used for about 90% of all web searches, so it plays a big part in how information moves around the internet.
Last week, President Trump said that Google was abusing its power by removing articles that were positive about his administration on purpose. Trump's claims were debunked by search experts and even some staunch Google critics, but they did reflect a growing worry about how much tech companies control what we see online.
As the Internet has grown bigger and more complicated, so has Google's importance. A small change to its algorithm can send a lot of web traffic in a different direction. Still, not many people know how their search engine works very well. Also, Google tries to keep it a secret for a good reason.
How does Google's search engine really work?
Let's say you want to search for "collard greens health benefits." When you start typing, Google's computers start sorting their index of hundreds of billions of web pages to find the ones that use the words you typed or words that are similar. (According to Google, "the health benefits of kale or collard greens" matches more than 53 million pages in English.)
So, Google uses a secret algorithm to put these pages in order.
This takes a lot of things into account. Google keeps most of the details of how its search engine works a secret, but it has told us a few things. One of the best things about Google's search engine was a formula called PageRank. It was named after Larry Page, who was one of the company's founders and is now the CEO of Alphabet, the company that owns Google. PageRank is based on the idea that the number of sites that link to a page is a good way to figure out how valuable it is. When web search first started, this was a new idea, and it helped Google beat out older competitors like Yahoo and AltaVista.
Over time, the search engine has become more and more advanced. On September 4, it had been around for twenty years. In addition to PageRank, the company has said that the software checks how often and where searched keywords appear on a page, how long ago the site was made (a sign that the information is recent), and the location of the person doing the search.
Google said there wasn't a big difference between how it picks news articles and how it picks other search results, but some things, like the date a page was made, matter more in news searches.
Why doesn't Google tell us how it works?
The tech giant says that if its formula was made public, it would be easier for people to change search results. People who specialize in search engine optimization (SEO) and help their business websites get more traffic already have their own business. With more knowledge, it makes sense that spam sites and ads could push out the most useful pages from the top search results.
Of course, Google also keeps its search engine formula a secret because it owns it. Most of the time, the company became the leader in search because it was better at showing the best results for a given query. Google doesn't want its competitors to be able to figure out its search algorithm, just like Coca-Cola doesn't share its recipe.
According to President Trump, the Google search engine is based against right-wing news organizations Is that correct?
Google said that political beliefs had nothing to do with how its search results were made. It also said that the company doesn't keep track of whether a user is conservative or liberal, and that it doesn't put web pages into groups based on their political views.
But after the 2016 presidential election, a lot of attention was paid to false information, which forced Google to change its search algorithm. Back then, it found that 0.25 percent of its daily traffic was linked to information that was wrong, hurtful, or misleading. It wanted the search results to show more of what it called "trustworthy" content. People said that the change had caused a big drop in traffic. But the group that filed the complaint and all the sites it mentioned lean to the left.
How does Google figure out if the search results can be trusted?
It depends on a huge group of people called "classifiers" to decide how good the search results are. There are 10,000 classifiers all over the world who work for Google. They rate the quality of search results to see if the pages that come up first are good sources of information that can be trusted.
Even though raters can't directly change how the search algorithm works, their feedback can point out problems with specific web pages or blind spots in the search formula. Google makes the rules that classifiers use to decide how good a search is public.
So Google's search engine doesn't favour anyone?
In no way. When it was first made, it only had a list of ten blue links for search results. Back then, the main goal was to help people find what they were looking for as quickly as possible. This has changed a great deal over time. Instead of just showing links, Google is adding more information to its web pages. This is making publishers and other services, like the Yelp review site, who rely on Google to drive traffic to their pages, unhappy. Google has said that people don't just want links when they visit your site; they also want information.
But since Google has added information about travel, shopping, and reviews of restaurants and local businesses to its search results, its competitors say that it is self-distributing or giving more attention to its own services than to those of its rivals. The European Union has said that this kind of special treatment goes against its laws against monopolies.
There is also the issue of bias that is not obvious. These are not questions about bias against a political ideology, as Trump said. Instead, they are about how algorithms or artificial intelligence might unintentionally amplify societal biases against women or other racial groups.
The worry is that because many Google engineers are white or Asian men, they are less likely to notice small problems that affect groups that aren't as well represented. For example, your browser has a feature called "autocomplete" that suggests search terms as soon as you start typing. Some of these ideas used to be based on racist or sexist stereotypes.
Google has said that it knows these kinds of biases could show up in its search results, but it is still working hard to fix these problems. Now, when you start a search with "Donald Trump is," the top three suggestions are "Democrat," "a great president," and "my president."