For several years, discussion about bias in search engine results has been fervent. There seem to be two primary concerns: (1) the way that Google reflects existing societal biases and (2) the way that Google actively reinforces biases. The former is seen when searches for general terms return results overwhelmingly of one gender or race. The latter is evident when comparing Google’s response to two image search manipulation incidents.
The broader moral question aside, I was curious whether search engine bias applied to topics related to litigation. My methodology was to type in searches related to litigation and compare the autocomplete suggestions offered by different search engines. (These are thought to depend on the most-searched and most-clicked articles by users like me.) Here’s what I found:
1) Litigation is…
If you query Google about litigation, the suggestions run the gamut from the linguistic (“is another word for” / “is a synonym for what”) to the cautionary (“is expensive”). Google’s suggestions also include conflicting claims: that litigation is and is not a game. This bubbles up both sides of the debate; fairly balanced there.
Bing’s suggested searches cover neither litigation vocabulary nor debate over whether it’s a game. Instead, its autocomplete results imply that potential litigants use the engine, as many of the suggestions highlight the risks and costs of litigation. The rest of the suggestions relate to specific cases or scenarios in which problems emerge during litigation. Not very broadly applicable, but not biased, to my eye.
DuckDuckGo comes out of the gate swinging, with the suggestion to search “litigation is a terrible career choice.” Other phrases are similarly negative, from claims it is “in bad faith” to those underscoring its “adversarial approach.” The results seem to lead the witness, so to speak.
2) Why is litigation…
When it comes to searches for “why litigation,” the results grow more philosophical.
Google’s autocomplete suggestions underscore the difficulty of litigation and ponder where in an organization’s budget it might fit. In a poetic turn of events, Google thinks you might mean “alliteration,” not “litigation,” if your search focuses on purpose. Though that is a bias of sorts, it’s one that encourages literary knowledge-gathering.
Bing’s suggestions for “why litigation,” on the other hand, focus on speed, import, and price. These phrases may bias searchers to hold negative beliefs about litigation. On the other hand, Bing autocomplete results also suggest better understanding litigation, from comparisons to arbitration to explanations of legal holds.
DuckDuckGo suggestions share some of the same biases as Bing’s. However, they do provide one contrasting view: that litigation is bad and that it is good.
3) Is it legal to…
This last one was purely for my own curiosity. I wondered: which laws are confusing and relevant enough to spur people to search for them? Not surprisingly, all three search engines suggested questions about substances like alcohol, tobacco, and marijuana. Each engine also bubbled up a rights issue: whether it is legal to film police officers (on Google), to watch uploaded videos (Bing), or to record conversations (DuckDuckGo).
After that, the suggestions grew stranger. Google searchers were particularly interested in whether they could own different animals – and Bing searchers joined them, when it came to the question of sloths. Both Google and DuckDuckGo suggested searching whether you could legally marry your cousin, and Bing and DuckDuckGo suggested searching whether riding in the back of a truck is allowed. I’m not sure if those suggestions are biased, but they certainly offer insight into what kind of legal advice and classes might prove popular.
One potential takeaway from this experiment? Consider turning off autocomplete, so you alone are biasing your searches – though not your results, of course. Or prepare to lose an hour down a search rabbit hole, trying to ascertain why owning a sloth might be illegal. Not that I have personal experience with that.