An algorithm is, in its simplest definition, a set of instructions on how to make a calculation. The word “algorithm,” as you might have inferred based on the “al-“ prefix, comes from the name of 9th century Persian mathematician Muḥammad ibn Mūsā al-Khwārizmī who is also known for the popularization of algebra (al-jabr in original Arabic).
The term is probably most commonly used today in reference to the mysterious processes that deliver us information and media via the internet. In essence, those algorithms are just as described above: we input a search term (or a company inputs our data, as in the cases of targeted advertising and “related” content) and the algorithm is the set of instructions written in computer code that allows a server somewhere to determine, based on those kinds of inputs, what results to provide us.
“The algorithm” has become a kind of shorthand for that whole process, often as a source of awe or squeamishness—ask people about their TikTok “for you” page, and most will report feeling like it was handcrafted to match their interests, sense of humor, and so on. Ask people about the ads they see, and many will half-joke about their phones and computers listening to their conversations and swear they’re getting ads for content, goods, and services that they’ve never engaged with online, but certainly have an interest in. While we know algorithms exist and the principles of how they work, there is often little or no transparency into exactly what sorts of instructions operate search functions on any major online services.
In other words, we can observe the input and the output, but not the processes that take the input and give us the output. We can scroll through the results we get, but we don’t know how algorithms determine which results are first, second, third, seventeen-thousandth, and so on. We don’t know how algorithms determine what results get included or excluded, or how they determine what kind of results are “related” or ones we “might be interested in.”
As a result of this lack of transparency and outside oversight, problems sometimes arise like the race and gender-related issues Safiya Umoja Noble identified in her book Algorithms of Oppression, and others have documented in scholarship and popular media elsewhere. We can never be sure if a company “fixes” the problems by making a structural change to the algorithms to prevent racist and sexist results from appearing, or if they’re sort of “patching” the algorithms, letting them continue to work as normal, but adding an exception to the rules to prevent objectionable results from appearing when popular search terms are used. In other words, we can’t know if a given tech company is treating the symptom and not the cause.
Algorithms are, after all, sets of instructions that are crafted by humans and given to computers to carry out. When searches are processed, the fact that they’re being processed by computers gives the whole experience the veneer of being objective, neutral, and unbiased or unaffected by the politics and beliefs of humans. In reality, the search algorithms we interact with on a daily basis are never neutral—they reflect and have embedded within them the beliefs, ideologies, biases, and blind spots of their creators. So, the most important thing for average users to know about algorithms in general is that they are rarely there to give you the best possible results, they’re there to give you the result that will maximize your engagement with the web service you’re using. Stay critical of your search results!