Names given to Donald Trump’s opposition by the US Republican candidate himself have been found to be appearing on Google’s autosuggest feature.
While instances of ‘lying ted’ were shown when the term ‘lying’ was searched for in Google last week, the equivalent slur Donald Trump coined for Hillary Clinton, ‘crooked hillary’, did not appear to show.
Search Engine Land’s Danny Sullivan sourced a response from Google querying the seemingly unequal results, and received a response, which read:
Autocomplete predictions are produced based on a number of factors including the popularity of search terms. Our systems are periodically updated to improve Search, and our users’ search activity varies, so the terms that appear in Autocomplete may change over time.
Additionally, our systems automatically filter a small set of offensive or inappropriate content from autocomplete predictions.
There is no difference between how our removal filters operate among candidates.
It therefore appears that the one-sided autosuggestions being picked up in Google are a result of actions out of its control, and influenced primarily by user behaviour.
Though Google autosuggest is influenced by previous searches, language, location, freshness and more, the favouring towards one term, despite having a lower search volume, has raised questions. With ‘crooked hillary’ achieving more than 10,000 additional searches in April in the US alone, some users appeared sceptical about Google’s answer.
This isn’t the first time that Google’s autosuggest feature has come under scrutiny for its apparent political bias. In February 2016, users spotted a disparity between autocomplete results for “Conservatives are…” versus “Labour are…”, with Google seemingly heavily favouring conservatives. The results returned for the labour search were full of defamatory suggestions, while the conservatives saw no such results.
In fact, the mystery surrounding results returned by the feature has even spawned the online game, Google Feud.
What does it mean for brands? Should they be worried about the potential for misuse by detractors? If Google’s explanation as to the mechanics of how the autosuggest feature works is to be believed, then the reasoning behind seemingly skewed or biased results could potentially be explained by dedicated manipulation by the opposition.
Reputation management is important in these scenarios, and while brands can’t directly control what autosuggest returns, they can use both paid and organic search results to ensure that users encounter a positive impression of their brand wherever possible online. Social media can also play its part. As well as this, it is important to note that autosuggest features are updated regularly, and can return just as quickly to their previous results as they changed to the new ones.
Whether or not this will be the case for “lying Ted” or “crooked Hillary” remains to be seen.