The Day Yahoo Decided I Liked Reading About Child Murder | Hacker News Reader

The Day Yahoo Decided I Liked Reading About Child Murder

theatlantic.com

38 pointsnsns11 comments

The thing about this is that the "Filter Bubble" isn't actually new. It has existed as long as mass-media has. Technology may well be making the phenomenon worse, perhaps to the point of creating something qualitatively different from what has been seen in the past, but anyone who thinks that the media has ever provided an accurate reflection of actual reality is horribly mistaken.

In fact its difficult to see how the media ever could provide such a reflection. Actual reality is very boring, reading about it doesn't interest people, because the brain has its own filters which lead us to only be interested in that which we find new or surprising.

Unfortunately providing people with information which appeals to them (which is what the media has always done, with or without algorithms) will invariably distort their own sense of reality, just as would training a probabilistic model only on outliers.

Many markets had more than one media outlet. You could choose to read other opinions.

With everybody bubbling, that options is disappearing.

Yes, it happens in traditional media too, albeit more crudely. After all, all of the incentives are already there.

Run stories about rare but terrible crimes -> more people watch your news -> ad revenue goes up -> run more stories about even rarer but more terrible crimes -> even more viewers -> wash, rinse, repeat -> everyone thinks the govt should be "tough on crimes" even though our society is the safest ever.

I totally agree about the pervasive distortion by mass media, but don't you think there's a fundamental difference between most everybody being subject to the same reality distortion versus different bubbles for individual people? Especially when those individual bubbles are completely opaque to the people trapped in them? It's easy to tune out most of the bias of mass-market broadcast culture, but much harder to be aware of what's being pumped to you specifically. If a company is going to tailor their behavior for what I "like", the least they can do is be up front about it and provide a transparent way to change those alleged "preferences".

After reading this story, I was curious how much Google News had personalized my homepage after several years of using it as my primary outlet. Somewhat surprisingly, there wasn't any I could find other than things I had told it to do: more "Science" stories, no "Entertainment", a couple custom news alerts, no Huffington Post or Fox News.

The actual stories under each category were identical, other than a few were linked to different sources.

The worst thing related to this happened to me just today. Google maps gave me different result for address search. Like the little mark on map was placed on different place than it was when I tried it in private tab (and just to make sure in other browser, also not logged in). It is a really weird thing to do (and dangerous too -- I almost sent a friend on a wrong bus).

The reason why I think this happened is different numbering system. You see, I entered street name, number and city. I meant street number, "normal" google maps showed result using the street number, but when I'm logged in I get result not by street number but rather distrcit number (not sure if I'm translating this right). Really bad behavior in my opinion. Also, there was no note on other results, or did you mean, or anything like that.

Something's wrong with the narrative. The author writes, "Going through my search history, I could trace the emergence of stories to the day I read about the Bustamante murder." .

But she didn't use 'search' to get to that article; she just clicked on it on the homepage. So it wouldn't be there in the search history.

Probably just a coincidence but Chrome's history page has a "Search history" button.

Perhaps she meant her browser history.

She probably meant "Searching through my browser history"

Let's also keep in mind that if true, this is personalization done poorly (for the consumer). Presumably, was no correction in Yahoo's model to consider whether the user liked the content after reading it. The first question here is whether or not the author clicked on more articles leading to more ad views as a result. If so, then it "worked" and made business sense. The second question is whether or not it improved the user's experience, which in this case it obviously did not.

Reply on news.ycombinator.com