鶹ý

Trump is Right: Google is Biased

— Search giant picks winners and losers -- driven by everyday users

MedpageToday

Last week, President Trump accused Google of rigging its search results, and Sen. Orrin Hatch suggested that the Federal Trade Commission should investigate. In this video, , agrees that Google is rigged -- but not in the way Trump suggested.

A transcript of his remarks follows:

At the risk of alienating some of you, I have a confession to make: it's not all that often President Trump's public statements on Twitter deeply resonate with me. A glaring exception occurred recently when he posted a Tweet about the Google search engine. Yes, Mr. President, you're mostly correct, Google is rigged. Or should I say perhaps more accurately, Google is biased.

Wait a minute. There's a big but here, a but way bigger than even Santa Claus's fanny. Evidence abounds that Google isn't deliberately rigged against the President or conservative media. No, it's biased for the masses. And also for making money. Allow me to demonstrate.

Let's do a couple of common online searches, starting with use of the search phrase heart attack. Google offers a special tool that tells me internet searchers use this phrase 246,000 times each month. In comparison, less than 7,000 searchers use the term acute myocardial infarction monthly, about one one-thirty-fifth as often. Of special interest, when I search the term heart attack, the top-ten Google results show links to website pages published by the American Heart Association, the Mayo Clinic, WebMD, and the National Institutes of Health and their subsidiaries.

Now we'll try another popular search term, breast cancer, which Google says searchers use 165,000 times each month. I get very similar results in the search engine results pages, or SERPs: the top rankings go to articles published by famous entities such as the Mayo Clinic, the American Cancer Society, Medical News Today, and so on. Note that none of these online publishers are names or institutions unfamiliar to the public.

Meanwhile, in my internet searches I can't find any rankings for scientific journal articles on either heart attacks or breast cancer anywhere in the first five Google results pages. As much as we in the healthcare sector venerate the Journal of the American Medical Association, the New England Journal of Medicine, Lancet, and the British Medical Journal, they generally don't rank highly in these kinds of searches. However, you do see them rank better when you use clinical search terms such as hormone-receptor-positive, HER2-negative breast cancer – which gets a mere 10 searches per month – and/or you expand the search term to heart attack scientific journal. But how many common folk know and use such terms? Not many, obviously.

Why do search engines favor known names? Back in the 1990s, two Stanford University graduate students designed the Google search engine to mimic scientific and academic research. The idea is that the most credible and respected journal articles are those that other journal articles later cite most often over time. Lots of citations mean lots of people are reading those articles. This is largely how search engines work as well but on a far more public basis. Although search engine algorithms are secret and typically change hundreds of times each year, search engine optimization specialists know generally how they work. For example, two major factors in search rankings are, firstly, how many internet searchers have visited a page or pages on a website, and, secondly, how many other popular websites link to that page or site in reference. In essence, search engines are online popularity contests, with a few ads sprinkled in.

What does all this mean to healthcare providers? That the internet and search engines are biased toward the famous rather than ordinary experts. Which is why TV docs such as Sanjay Gupta and Mehmet Oz will never want for patients. They are brand names with high public awareness, just like Mayo, the NIH, and the American Cancer Society. No big surprise, of course, but I do see some injustice, a la Mr. Trump's complaint. You might be the world's leading oncologist in treating hormone-receptor-positive, HER2-negative breast cancer, but your expertise is all but invisible on the internet. With a couple of exceptions, your website page on this subject has little to no chance of ranking ahead of Dr. Oz's multitude of breast cancer videos on his website – even though his specialty is in cardiothoracic surgery.

The two exceptions bring good news to all us aggrieved nobodies. Google's second bias is for income. You can jump ahead of Dr. Oz's rankings by paying for advertisements to appear at the top of search engine results pages for specific search terms in specific geographic locations. Moreover, search engines are getting more localized – Google and others are paying more attention to searchers' individual locations. This is why you get different search results in Rochester, New York, than in Tucson, Arizona. And why you, an oncologist in Wisconsin, have a chance of your website biography ranking highly whenever an internet user uses the search term Green Bay breast cancer doctor.

Let's face it, Mr. Trump. Humans tend to trust the familiar, which is why I'm often tempted to drive past locally owned coffee shops on my way to Starbucks, even though as a small business owner I try hard to support other small businesses. As much as you may label the New York Times and CNN as fake news, their websites still draw millions of eyeballs each week. This explains why their online articles go to the top of internet search results – along with those of the British Broadcasting Company, CBS News, The Hill, Fox News, and other media of various stripes and slants, along with the President's own Twitter feed. Search engines only reflect the biases of their users, not of executives at Google.

You might say that Google is basically a product of a democratic, open, free-market society. It simply gives the masses what they want, not unlike a reality TV show. You don't have a problem with that, do you, Mr. President?