@seawulf575 If you read through that section of their website, you would have a better gist for how their search algorithms work. Significantly, they aren’t directly coding “this website better” or “this website worse.” They’re trying to come up with more general patterns to allow the computer to sort through the immense mass of data that is the internet, whatever someone’s query might be.
The part that is probably most relevant for our purposes is the “Ranking useful pages” topic under “Search Algorithms”: These algorithms analyze hundreds of different factors to try to surface the best information the web can offer, from the freshness of the content, to the number of times your search terms appear and whether the page has a good user experience. In order to assess trustworthiness and authority on its subject matter, we look for sites that many users seem to value for similar queries. If other prominent websites on the subject link to the page, that’s a good sign the information is high quality.
So the algorithms “look” for, among other things: how recently the page was published and how often the search term appears. We’ll get back to this shortly. They also explain a little about how they determine a site’s trustworthiness: do other Google users click on it? Do other websites link to it (this data would be known from the crawlers and indexing process)? These are easily-tracked data points for algorithms. Notice, too, how these are general patterns that allow the computer to sort through large amounts of information quickly no matter a user’s query.
As for the test that supposedly exposes a bias… typing a single word, like a name, into a search bar is a bad way to use a search engine for anything more complex than simply finding that person online. This is why the general search results of “Trump” always lead with Trump’s Twitter and Trump’s personal website, and why Wikipedia is always somewhere near the top if not at the top.
If you type a lone name into the News section of Google (or look specifically at the news articles in the general results)—well, remember that news, as an industry, writes about the negative far more than the positive—that’s the way it works—and expect that to be reflected in the resulting headlines. Type “Clinton” or “Hillary Clinton” into Google, and the results are quite different. Unsurprisingly.
I’m just saying, it’s not a surprising result of the test, especially for someone as active in politics as Trump currently is, thanks to the combination of his job and his lifelong flair for drawing attention to himself.
As for why a “liberal” website like Washington Post would show up on a Google search for Trump more than Fox—well, before we jump to “liberal conspiracy of deliberate anti-Trump bias,” let’s look at how the content of two opposing sites might be sorted by algorithms that use “number of times your search terms appear” and “freshness of the content” as two key factors.
I searched the keyword “Trump” internally on both Fox News’ website and Washington Post’s website (these were the first newspapers that came to mind, so I went with them). I sorted the results first by date to see what Trump-related articles were most recent (or “fresh”), and then by “relevance.”
Fox News “Trump” search (within Fox’s own site), sorted by date:
– Only 2/10 headlines with “Trump” in the title.
– Only 4 with “Trump” in the description (2 of those articles were the 2 with “Trump” in the headline, mentioned above), and that’s counting the phrase “CNN’s Trump-Russia meeting story.”
– 6/10 articles do not have “Trump” in the headline or in the description at all—they are about something else which Trump either only features briefly, or is adjacently related.
Washington Post “Trump” search (within Washington Post’s own site), sorted by date:
– 5/10 headlines with “Trump” in title.
– 9/10 had Trump in the description
– Only 1 that did not have Trump’s name in either location.
When I sorted by relevance on both sites, the name “Trump” was in the headline of every article. The first 4 from Fox News were opinion articles by Ivanka Trump and Lara Trump, with dates from July, June, and March of this year, and May from last year (so Trump in name, but not the Donald). The next most relevant is from the 3rd (today), and the dates march steadily backwards: the 3rd, then Aug 27, 23, 19, 19, 13.
That isn’t ancient history, but if we compare it to the Washington Post results, sorted by relevance… Of their first 10 most “relevant” results, 4 are from the 3rd; one from the 2nd; 4 from Aug 31, and 1 from Aug 29.
Either way we slice it—date or relevance—Washington Post wins by a clear margin.
None of this is conclusive, but it does suggest an alternative, simpler, reason why Fox News might not show up as often as you’d personally like when you search for “Trump”: they’re not publishing as many stories that will catch the “eye” of the mindless algorithms sorting for, among many other things, date of publication and recurrence of search terms. As far as those two factors, for this particular query, Washington Post beats Fox News, and that is enough to explain the results of the “Trump” Google test without resorting to conspiracy theories involving computer programmers wasting their time creating secret pro-Trump/anti-Trump or conservative/liberal rankings of websites.
