3 ms·
I have greater issue with Google Search[1] than I do FB and TWTTR, they should provide a politically unbiased index of the surface web, but unlike DDG, Bing or
by mc32 8y ago
I have greater issue with Google Search[1] than I do FB and TWTTR, they should provide a politically unbiased index of the surface web, but unlike DDG, Bing or others, they curate theirs to further their world view outside that required by law.
Twitter and FBs issue is reliance and optimizing for engagement which leads to clicking and surfacing the most outrageous stuff at the expense of the ordinary.
Google search doesn’t have “an gagement problem”. They have black hat SEO problem, which is different, but they go beyond that and tweak some search results for political reasons. Not talking about terrorism or supremacy groups at all, but somewhat controversial social issues, (reproductive) sometimes tame issues (searching for candidates)[1]. The EU has fined them for “re-ranking” SERs and SSE effects.
>These data showed that Google’s search results favored Hillary Clinton (whom I supported) in all 10 positions on the first page of search results — enough, perhaps, to have shifted two or three million votes in her direction over time.<
Given that, 2-3MM vote swing (vs, what the outside estimate was under 50k votes for Russian bot sway), I’d say Google is the real problem in affecting democratic elections.
[1]https://www.usatoday.com/story/opinion/2018/09/13/google-big-tech-bias-hurts-democracy-not-just-conservatives-column/1265020002/ https://www.usatoday.com/story/opinion/2018/09/13/google-big...
- Ivoirians 8y agoFriendly reminder that Robert Epstein (the author of your article) is a moron who's on an anti-Google crusade because they (correctly) flagged his website as containing malware, and he threw a childish tantrum (threatening lawyers, public letters to the CEO). [1][2] Let me summarize the core studies he ran and builds his case on [3][4] [3]: We built a fake search engine and had people look at search results for political candidates they'd never heard of before. It turns out when the top results on the page presented one of these candidates favorably, the subjects were more likely to favor those candidates. Based on this, we think 20% of undecided voters in a nation can be swayed to support a candidate by search engines, who are clearly biasing their results on purpose. (???) [4]: Google search results pages showed more articles favorable to Clinton than Trump. By this, and the conclusion in [3], we estimate 20% of American undecided voters (this is where the 2-3MM comes from) may have been swayed to vote for Clinton by Google. Is it clear why this is absurd? Especially if you realize that, since the general tone of news published about the election is positive to Clinton and negative to Trump, Google is accurately reflecting the news landscape when the articles it shows are positive to Clinton. [1] https://bits.blogs.nytimes.com/2012/01/05/one-mans-fight-with-google-over-a-security-warning/ https://bits.blogs.nytimes.com/2012/01/05/one-mans-fight-wit... [2] https://bits.blogs.nytimes.com/2012/01/06/readers-and-experts-weigh-in-on-a-site-owner-vs-google/ https://bits.blogs.nytimes.com/2012/01/06/readers-and-expert... [3] https://www.pnas.org/content/pnas/112/33/E4512.full.pdf?with-ds=yes https://www.pnas.org/content/pnas/112/33/E4512.full.pdf?with... [4] https://aibrt.org/downloads/EPSTEIN_&_ROBERTSON_2017-A_Method_for_Detecting_Bias_in_Search_Rankings-AIBRT_WP-17-02_6-1-17.pdf https://aibrt.org/downloads/EPSTEIN_&_ROBERTSON_2017-A_Metho...
- mc32 8y agoThanks for bringing that to light. I was unaware. Still, the EU fined them for SER(p) manipulation. And, independently of mr Epstein, one can do side-by-side comparisons btwn Google and DDG and Bing and one will see the manipulation in action. So, while his conclusions might be disingenuous, his assertions (manipulation) can be independently confirmed.
- Ivoirians 8y agoThat was actually the one line in the article that got me riled up and caused me to start digging. >Is there evidence of actual favoritism in Google’s search engine? Well, the European Union certainly thinks so, having fined Google $2.7 billion last year for having biased search results. The fine was for an anti-trust ruling, where it was determined Google was promoting their own shopping results (showing them at the top of the page) over organic shopping sites, stifling competition. Favoritism for their own service, maybe, but nothing to do with pushing or manipulating results. So that's why I suspected he had an agenda to push and was willing to spin the truth to argue for it. You can argue that study [4] showed Google tended to have more Clinton-favorable results, but I don't personally think that shows manual manipulation. It's worth a critical read--in Issue 7, he seriously alleges that Google somehow realized this study was being conducted halfway through, and started serving unbiased results to subjects with Gmail accounts. If you accept that conclusion, okay, but I'd tend to believe that the results aren't as statistically significant as claimed.
- mc32 8y agoInterestingly, Google disagrees with the EU. So they deny doing what the EU claims they did. Here's a read of some of the things they do out of legal obligation and other they do for selfish reasons and others for other reasons they believe are worthwhile: https://en.wikipedia.org/wiki/Censorship_by_Google#Search_suggestions https://en.wikipedia.org/wiki/Censorship_by_Google#Search_su... compare an autocomplete search for sarah joeng, robert muller, or roger stone on google, ddg and bing and look at SSE results. Bing DDG show relevant political issues, Google is coy about those.