I was recently reading comments on a post related to COVID-19, and saw a reply I would classify as misinformation, bordering on conspiracy. I couldn’t help but ask the commenter for evidence.
Their response came with some web links and “do your own research”. I then asked about their research methodology, which turned out to be searching for specific terms on Google.
As an academic, I was intrigued. Academic research aims to establish the truth of a phenomenon based on evidence, analysis , and peer review.
On the other hand, a search on Google provides links with content written by known or unknown authors, who may or may not have knowledge in that area, based on a ranking system that either follows the preferences of the user, or the collective popularity of certain sites.
In other words, Google’s algorithms can penalize the truth for not being popular.
Google Search’s ranking system has a fraction of a second to sort through hundreds of billions of web pages, and index them to find the most relevant and (ideally) useful information.
Somewhere along the way, mistakes get made. And it’ll be a while before these algorithms become foolproof – if ever. Until then, what can you do to make sure you’re not getting the short end of the stick?
One question, millions of answers
There are around 201 known factors on which a website is analyzed and ranked by Google’s algorithms. Some of the main ones are:
- the specific keywords used in the search
- the meaning of the keywords
- the relevance of the web page, as assessed by the ranking algorithm
- the “quality” of the contents
- the usability of the web page
- and user-specific factors such as their location and profiling data taken from connected Google products, including Gmail, YouTube and Google Maps.
Research has shown users pay more attention to higher-ranked results on the first page. And there are known ways to ensure a website makes it to the first page.
One of these is “search engine optimization”, which can help a web page float into the top results even if its content isn’t necessarily quality.
The other issue is Google Search results are different for different people, sometimes even if they have the exact same search query.
Results are tailored to the user conducting the search. In his book The Filter Bubble, Eli Pariser points out the dangers of this – especially when the topic is of a controversial nature.
Personalized search results create alternate versions of the flow of information. Users receive more of what they’ve already engaged with (which is likely also what they already believe).
This leads to a dangerous cycle which can further polarise people’s views, and in which more searching doesn’t necessarily mean getting closer to the truth.
A work in progress
While Google Search is a brilliant search engine, it’s also a work in progress. Google is continuously addressing various issues related to its performance.
One major challenge relates to societal biases concerning race and gender. For example, searching Google Images for “truck driver” or “president” returns images of mostly men, whereas “model” and “teacher” return images of mostly women.
While the results may represent what has historically been true (such as in the case of male presidents), this isn’t always the same as what is currently true – let alone representative of the world we wish to live in.
Some years ago, Google reportedly had to block its image recognition algorithms from identifying “gorillas”, after they began classifying images of black people with the term.
Another issue highlighted by health practitioners relates to people self-diagnosing based on symptoms. It’s estimated about 40% of Australians search online for self-diagnosis, and there are about 70,000 health-related searches conducted on Google each minute.
There can be serious repercussions for those who incorrectly interpret information found through “Dr Google” – not to mention what this means in the midst of a pandemic.
Google has delivered a plethora of COVID misinformation related to unregistered medicines, fake cures, mask effectiveness, contact tracing, lockdowns and, of course, vaccines.
According to one study, an estimated 6,000 hospitalizations and 800 deaths during the first few months of the pandemic were attributable to misinformation (specifically the false claim that drinking methanol can cure COVID).
To combat this, Google eventually prioritised authoritative sources in its search results. But there’s only so much Google can do.
We each have a responsibility to make sure we’re thinking critically about the information we come across. What can you do to make sure you’re asking Google the best question for the answer you need?
How to Google smarter
In summary, a Google Search user must be aware of the following facts:
- Google Search will bring you the top-ranked web pages which are also the most relevant to your search terms. Your results will be as good as your terms, so always consider context and how the inclusion of certain terms might affect the result.
- You’re better off starting with a simple search, and adding more descriptive terms later. For instance, which of the following do you think is a more effective question: “will hydroxychloroquine help cure my COVID?” or “what is hydroxychloroquine used for?”
- Quality content comes from verified (or verifiable) sources. While scouring through results, look at the individual URLs and think about whether that source holds much authority (for instance, is it a government website?). Continue this process once you’re on the page, too, always checking for author credentials and information sources.
- Google may personalize your results based on your previous search history, current location, and interests (gleaned through other products such as Gmail, YouTube or Maps). You can use incognito mode to prevent these factors from impacting your search results.
- Google Search isn’t the only option. And you don’t just have to leave your reading to the discretion of its algorithms. There are several other search engines available, including Bing, Yahoo, Baidu, DuckDuckGo and Ecosia. Sometimes it’s good to triangulate your results from outside the filter bubble.
This article by Muneera Bano, Senior Lecturer, Software Engineering, Deakin University is republished from The Conversation under a Creative Commons license. Read the original article.