The other day I decided to search for the word “cookie” on Google. Literally: “cookie.” No context, because I just wanted to see what would come up on top. And indeed, the results surprised me a bit. There was a nutritional table off to the right side and a picture of some pretty mediocre-looking chocolate chip cookies. Down in the actual search results, though, the first two results were about browser cookies- the kind that store info on your computer. The rest of the top 10 was a mix of browser cookies, and edible cookies. That surprised me, because I just naturally think of edible cookies. But is it fair to assume that’s “natural?”
Whenever we type something into that search bar, Google has to predict what we’re asking about. Sometimes the results are spot-on. Other times, there’s a disambiguation problem, a spelling problem, or just not enough information for Google to do its job correctly. Searching a generic term like “cookie” didn’t really give Google enough information to help me out. Imagine you were a non-native speaker researching for a project and all you knew is you were supposed to do a report on “cookies.” But you didn’t know what kind. How would you handle these mixed results? Would you decide to write about food, or web lingo? Have you ever seen those lists of Google auto-completes? It’s funny to see what Google expects us to say, and it’s also funny to see what people actually are searching for.
The fact that I expected to see relevant results actually points out a perspective bias. I think about edible cookies a lot more than any other sort, so I expect Google to do the same. Plus, Google has a ton of data on my search habits and browser history, so I thought maybe that’d play a factor in how the results index. But, web cookies are likely more relevant to Google’s own bread and butter. Perhaps its indexing engines take that into account when they stack the results? I’m not really sure how their system works, of course, but I’m so intrigued by how the different factors must get weighed.
I do a lot of consumer research and whenever we ask people how they look up information on a given topic, they tell us that they go Google it. Google is a wonderful and powerful tool. I use it constantly for my own market research at work. But we have to remember that even the Great Google has bias in its results. No source can ever be completely un-biased; it’s just not possible. There is always some system classifying the information, and that classification order imparts bias. Whether it’s Google telling us which kind of cookies are most important, or a news source choosing what facts to share, we always have to dig deeper. You can’t just settle for the first few answers you find- you have to try to determine if they’re really the best answers. Imagine you were researching a controversial topic and the top 5 results all claimed the same point of view. Perhaps results 6-10 rebutted this point of view entirely. But if you never clicked past #5, you might assume there was only one possible answer and walk away mis-informed.
Just for kicks, here is a fun video mocking our Google searches:
And one last thing. Wondering about this post’s title? Check out Let Me Google That For You. This is snarky, so use it wisely. But when someone asks you a silly question they could just Google in 5 seconds… this lets you tell them that.