4 ms·

Let's address the elephant in the room: the problem lies with Google. Prior to this recent NY Times article, why is a family brawl with minor injuries and no c

by cirno 7y ago

Let's address the elephant in the room: the problem lies with Google. Prior to this recent NY Times article, why is a family brawl with minor injuries and no charges that took place in 2008 shown on the first page of results when someone is looking up a man's pizza and seafood restaurant? Is that really the most relevant result that is in the public interest? I would bet money the original story didn't even have any backlinks. Bing and other search engines are not nearly as awful, but Google in particular has a notoriously bad habit of promoting defaming content to the top of its search results. And right now, they absolutely dominate the search market. It is my belief that Google's AI is optimized to prioritize "scandalous" content, because it grabs readers' attention, in the same way National Enquirer headlines at the grocery store checkout lane do. Google does the same sort of thing with YouTube, promoting extremist content above moderate content, and it has a real effect on people: https://www.nytimes.com/interactive/2019/06/08/technology/youtube-radical.html https://www.nytimes.com/interactive/2019/06/08/technology/yo... There is a real problem with online harassment and defamation, and Google is contributing heavily to it by rewarding sites that peddle defaming content.

zaroth 7y ago
I love this hypothesis because not only does it fit in perfectly with my world view of Google, but because it’s entirely plausible to consider how their algorithm could inadvertently optimize for such a result simply by looking at click through and bounce rates. People are likely to click on the most sensational result on the page, and in cases like this they are pretty damn sure going to pick a different search query afterward, which to the algorithm will look like success. It seems like something easy to study and prove to a high degree of confidence if it is occurring. Particularly the theory that other search engines don’t surface the same negative content, or don’t rank it as highly.
- cirno 7y ago
  I have done a significant amount of research into this, and compared results through multiple search engines. I am confident this is mostly specific to Google. I would prefer not to name the sites in question, because they are borderline illegal and I do not wish to give them exposure, but on the off chance any journalists are reading this and are interested, please get in contact with me here and I will share my findings.

by cirno 7y ago

zaroth 7y ago

I love this hypothesis because not only does it fit in perfectly with my world view of Google, but because it’s entirely plausible to consider how their algorithm could inadvertently optimize for such a result simply by looking at click through and bounce rates. People are likely to click on the most sensational result on the page, and in cases like this they are pretty damn sure going to pick a different search query afterward, which to the algorithm will look like success. It seems like something easy to study and prove to a high degree of confidence if it is occurring. Particularly the theory that other search engines don’t surface the same negative content, or don’t rank it as highly.

cirno 7y ago
I have done a significant amount of research into this, and compared results through multiple search engines. I am confident this is mostly specific to Google. I would prefer not to name the sites in question, because they are borderline illegal and I do not wish to give them exposure, but on the off chance any journalists are reading this and are interested, please get in contact with me here and I will share my findings.