Hacker News Reader

Comment by danso | Hacker News Reader

The last line FTA:

> Yet even that hasn’t solved the problem. See for yourself: Do a Google search for “Guy Babcock.”

The first "traditional" result for Babcock's name is now the NYT story, but the first 5+ (even moreso on mobile) are thumbnails [0] showing his image with a text overlay of his name and "paedophile".

The first 3 thumbnails link to the sites cheaterboard dot com, cheaters dot news, thecheatalert dot com. The fourth thumbnail is a Pinterest account linking to a ripoffreport post.

All these "top" results feature the same "content" [1] – the victim's name spammed repeatedly, along with "paedophile" and "racist POS".

I know spam abuse detection isn't trivial, but this seems like the most basic kind of spam abuse that Google claims it knows how to downrank.

[0] https://i.imgur.com/gosroWQ.jpg

[1] https://i.imgur.com/kFI9XNc.png

View on HN

Yeah I’m not sure it was a good idea of the author to include a link to that search from the article (seems like that would boost the existing results).

I think I'd argue that it was a good idea. This isn't me linking to that from my blog that gets 100 views to a post. This is the NY Times linking to that. Love them or hate them, them linking to something will get attention, and I would bet that attention will lead to changes.

Good point.

I don't see how linking to a search would boost the results.

Some percentage of the many people following the link from NYT would do so while logged into Chrome. Or maybe Google Analytics is on the web site. Or some other Google property comes into play.

Google would then use its apparent popularity to boost its search engine ranking.

Google has to have a method of determining whether a search query originates from a manual request, or if it is a link to the results, such as here.

There are many ways this could be detected: The Referer header from old browsers, the massive amount of data that the search query URL contains (the parameters that are not "q"). Probably Google even has session tracking, and the session would be missing everything before the results page.

Given that Google applies ML so heavily, and that this query would be an outlier in so many ways, I can't imagine how they wouldn't detect this.

>> Google has to have a method of determining whether a search query originates from a manual request, or if its a link to the results

The issue was about how it could be risky to one's reputation by linking from a NYT article to a disparaging page.

My parent asked how such a risk might arise. I assumed they could not imagine how one's browser, NYT and the target site could conspire to affect rankings.

I explained how this might happen. So Google knows there's traffic. Some traffic ranks higher than none, and much traffic ranks higher than some.

So, to allow Google to be aware that a well-trafficked site is slating you, is riskier than if it's a site with few visitors because that site will now rank higher.

Or is down ranked, innit? The problem was (before the NYT article) that this person is not a public persona so there's no content to uprank above it.

Reply on news.ycombinator.com