We rely on search engines to navigate the world on a daily basis and in a variety of forms. Google alone saw at least 5.5 billion daily searches in 2016 and holds over two-thirds of the United States search market. Whatever it is we search for and on whatever search engine we choose, we expect the results we get to be relatively unbiased, ordered by their relevance to the search. Each search engine chooses and orders its results differently based on algorithms most hide from the public. A search engine, especially one the size of Google, has the power to make sites disappear into relative obscurity by listing them lower or not at all.

Although we might expect our search results to be relatively unbiased, there isn’t a guarantee that they will be. In 2014, a San Francisco court ruled that the way Google orders its search results is protected by Google’s first amendment rights. In other words, our search results are completely up to the company behind the search and what the company, or its algorithms, decide to censor. And they do drop sites from results; search engines lower the rankings of websites that load slowly, carry malware, or are deemed untrustworthy by their algorithms. They don’t always get it right, however. Google has surfaced conspiracy theories in its instant answers on more than one occasion and manually steps in to correct those results. In some cases, sites have figured out how to game the results to promote a questionable agenda, despite the best efforts of companies like Google.

The fact that major search engines can make websites almost disappear from the Internet is as disruptive to a business as it is to the ability to be informed. Aside from well-meaning corrections and exclusions, websites are dropped from search in questionable circumstances. Foundem, a search engine launched by UK entrepreneur Adam Raff was “penalized” by Google, effectively making it disappear from the Internet from 2006 to 2009 from anyone who relies on Google Search. Being penalized by Google means being lowered in rankings or dropped completely which means being found and clicked on less.

Google is open about some of the reasons it decides where to rank results in search. In recent years, the search giant has announced that how mobile friendly sites were or whether they were available over secure protocols were factored into where a site showed up in search. That is, of course, in addition to “relevancy” which has been a factor in Google’s algorithms for a long time. Generally, Google search results are high quality and relatively similar from person to person, without a wide degree of interest targeting. However, in addition to its issues with Foundem, Google has been involved in multiple lawsuits over low rankings or even blacklisting of certain sites and competitors.

The differences in algorithms between search engines are the factor that distinguishes search engines from each other. They’re important to the identity of a search engine and are what make some search engines better for certain searches than others (or, better at finding what we want to find). While there isn’t anything wrong with that, it does mean there is little to no transparency in how search engines choose their results, which is a problem. As search engines get further ingrained into our lives for everything from finding a good restaurant to finding the latest on a political candidate, that lack of transparency may severely harm our ability to be informed. Without knowing what biases are built into a search engine–intentionally or not–we don’t know how much we can trust search results especially when it comes to news and politics.

Though trying to compel a search giant to publish its search algorithms is a waste of time and arguably, not terribly worthwhile, more transparency in how it decides the relevance of sites is. It’s hard to know what the motivations, political leanings, or even just software bugs will be in a search engine’s future.

In the meantime though, we can choose search engines that are dedicated to unbiased and non-personalized results to make sure our search results are not keeping us in a filter bubble. Unfortunately, even for some search engines that are committed to those principles, such as DuckDuckGo and StartPage which both highlight their commitment to privacy and neutrality, the details of how they order search results are relatively secret (or, in the case of StartPage, are based on Google). Others, like Gigablast, are fully open source so anyone can see their inner workings. That makes it easier to hold them accountable for their search rankings so that we know they’re being reasonably fair and unbiased no matter what we search for.