Conservatives: Stop Crying Wolf on Tech Bias
This article by Lincoln Network head of policy Zach Graves originally appeared at Techdirt.
In an article picked up by Drudge Report and then tweeted by President Donald Trump himself, PJ Media editor Paula Bolyard makes the shocking claim that Google deliberately manipulates its search results to favor left-wing views and undermine the President.
In supporting this allegation, she goes to Google and looks through the first hundred listings on the search engine results page. Therein, she finds that 96 percent of results for “Trump” are from liberal media outlets. Bolyard remarks:
I was not prepared for the blatant prioritization of left-leaning and anti-Trump media outlets. Looking at the first page of search results, I discovered that CNN was the big winner, scoring two of the first ten results. Other left-leaning sites that appeared on the first page were CBS, The Atlantic, CNBC, The New Yorker, Politico, Reuters, and USA Today
She adds that other than Fox News and the Wall Street Journal, traditional right-leaning outlets didn’t make the cut:
PJ Media did not appear in the first 100 results, nor did National Review, The Weekly Standard, Breitbart, The Blaze, The Daily Wire, Hot Air, Townhall, Red State, or any other conservative-leaning sites except the two listed above.
Aha! A big tech company caught red handed pushing its progressive agenda. Well…not so fast. Rather than uncovering compelling evidence of bias, this article’s author and its promoters merely reveal their ignorance of how search engines work.
First, the author seems to conflate Google Search and Google News, two products which use different algorithms and serve different functions. Google News is a searchable news aggregator and app (with some overt editorial functions), whereas Google Search tries to give users the most useful and relevant information in response to a query.
In order to determine what constitutes a relevant and useful result, search engines use complex algorithms to rank the quality of different pages based on a variety of signals such as keywords, authoritativeness, freshness or site architecture. A big part of this quality determination is based on outside links to a site – an idea going back to Larry Page and Sergey Brin’s work at Stanford in the late 1990s that culminated in the creation of the PageRank algorithm.
Page and Brin realized that incoming links to a site served as a proxy for quality markers like authoritativeness, trustworthiness and popularity. Today, Google Search is much more complex, utilizing complex machine-learning functions like RankBrain and an evolving set of algorithms with names like Hummingbird, Panda, Penguin and Pigeon. However, incoming links are still a key factor. Additionally, while Google uses manual quality raters to test new algorithm changes, they do not use them on live search results.
Google News’ approach to ranking results is also driven by algorithms that use a number of the same signals (you can get an idea from their patent), with a couple exceptions where manual input is used for editorial features, major events, and cross-over results from Google Search for particular topics.
With this in mind, it should be no great surprise that outlets like the New York Times, CNN, and Washington Post trounce outlets like PJ Media, National Review, and the Weekly Standard in organic search. The sites in the latter group don’t have metrics that support them rising to the top of the search algorithm. Of course, PJ Media found Fox and WSJ weren’t affected by this “bias” because their numbers are actually comparable to the former group of “left-wing” outlets (see below).
(Data from Alexa.com)
This approach to ranking quality isn’t just a Google thing. If you look at competitors like DuckDuckGo or Bing (which PJ Media didn’t seem to bother doing), you’re going to see pretty similar results. Maybe this says something about the media landscape. But it’s not a good reason to storm Mountain View with pitchforks.
PJ Media’s conspiracy-mongering is based on an avoidable misunderstanding that could throw gasoline on the techlash and lead to policies that chill American innovation (although at least for now, conservatives still think a Fairness Doctrine for the Internet is a dumb idea).
It’s worth saying that libertarians and conservatives aren’t totally unreasonable in wanting to investigate whether they’re getting fair treatment by tech companies. After all, Silicon Valley is a very liberal placethat doesn’t always reflect their norms or values (I also say this as someone with generally right-leaning views who has worked for organizations like the Cato Institute and R Street). That being said, if you’re going to make an allegation that there’s a big conspiracy, you should do your due diligence. This means taking time to understand the underlying technology before jumping to conclusions.
On Google’s part, given all of the tensions around bias lately, they would probably be wise to be more transparent about how their news algorithm works and do more proactive outreach to avoid future misunderstandings.