Hacker News Reader

Comment by mic47 | Hacker News Reader

I worked for FB anti-spam team, and believe me, nobody trusted user feedback absolutely (i.e. it's good to send to review, but not automatically take down). I guess that with fake news the signal to noise ratio is even worse.

The whole article sounds more like focusing on one implementation details of review queue prioritization (i.e. what is the difference of definition of "fake news" for this user vs Facebook's definition), rather than bigger picture.

View on HN

To me, fake news also includes slanted coverage of true events (i.e. covering events in such a way as to convey a sponsored message, biased polling methodology, etc.). That encompasses statistically approaching 100% of trashy sources you’re likely to encounter as 'sponsored content', as well as a healthy portion of more established sources' content. I'd be likely to report a majority as fake news in perfectly good faith if given the option.

Maybe the term isn’t as narrowly defined as your (former?) employer would like.

Edit: upon a bit more reflection, I do take for granted a desire to protect life and liberty (in that order), and anything that appears to threaten that view will draw special scrutiny, and potential abuse of reporting features.

The problem with conflating biased coverage with fake news is that the latter phrase becomes meaningless.

"Fake" means "false", not "stuff I disagree with".

But the parent commenter's issue does imply a solution! Just add "Biased/slanted coverage" to the list of reasons to report, and now your "Fake News" section is much higher signal to noise.

Everything is biased. True objectivity is effectively impossible, and it's not even clear that perfect objectivity is desirable.

People will report any negative coverage of their favorite party/politician as biased, especially now that the current U.S. president hoists that banner every day.

Certain content might be more or less blatant about it, and as seymour1 suggests, gathering that additional data and cleaning up the currently actionable data really only has upsides.

One avenue would be to put those who report the same things as biased in cohorts, to be shown or not shown similar content.

> One avenue would be to put those who report the same things as biased in cohorts, to be shown or not shown similar content.

That just reinforces the current self-silo effect we see, with people only seeing the point of view with which they agree.

It's not like being exposed to content biased contrary to our view changes our view; it's just more readily detected as biased, and in certain ways may reinforce our view as we defend it.

In any case, the ratio of bias reports/views for content is likely to be inversely related to its value and general appeal.

Exactly. A statistically literate person should be able to make reasonably accurate inferences about the broader distribution of some event class by reading about the most newsworthy events and extrapolating (taking into account that newsworthy events should be among the highest-impact in the class). However, practically all mainstream news sources fail this basic test. You can't get very close to an accurate picture of reality these days without reading e.g. individual bloggers who dedicate a significant portion of their lives to highlighting relevant patterns that mainstream news sources on both 'sides' are intentionally omitting or burying.

I think the use of "trustworthiness" is misleading. Really it's more "how likely is this person to accurately flag an article that matches Facebook's internal definition of fake news."

That would mean that your number would be on the lower end, but that's not a reflection of your character. More likely it just means that whatever queue is set up for review will wait until more and higher-scoring individuals also flag the article before bumping it up the queue.

Funny how this article is something I would flag if I used your definition of fake news. To me this article is trying to allude to some nefarious Black Mirror nonsense.

> Really it's more "how likely is this person to accurately flag an article that matches Facebook's internal definition of fake news."

Couldn't say it better.

Well said, and I agree completely about this story.

There's still an aspect of this with which I'm unsatisfied, however. What's the character of leadership who would address misuse of the feature by devising weights intended to devalue the opinions of large swaths of their engaged user base? Significant numbers of people are expressing a sentiment through their alleged abuse of the feature, and instead of trying to channel that constructively, the company is devising ways to ignore it.

> "Significant numbers of people are expressing a sentiment through their alleged abuse of the feature, and instead of trying to channel that constructively, the company is devising ways to ignore it."

What's the character of leadership who would unjustly silence group A just because significant numbers of people are expressing sentiment to silence group A? Not sure if not doing what they are doing is actually better.

On another note:

This is not necessarily abuse. Even with no abuse, you can have different definition of fake news, than FB have. In such case, you report, someone at FB reviews, and nothing happens. Having weight on you does not change this. The "leadership" is "ignoring you" in a same way as before. Apart from the fact that it's not "ignoring you". They just have different opinion than you.

Note: I used "you" but I don't think that you necessarily have different definition than FB, if was just more convenient to formulate it this way.

By using "significant numbers" alone, you're vulnerable to coordinated information suppression efforts (by state actors, for example).

Certainly there are more than bad faith actors abusing the feature, which could be separated out through changes to the interface.

I think it boils down to regulation of AI. Only the EU seems to be challenging Silicon Valley with actual penalties and laws to keep them a bit more straight. Why can't the U.S. regulate their own companies?

The thing is that pretty much all reporting has bias. Interpretation is inextricable from observation. Bias in and of itself is not a bad thing -- it provides a framework for meaningful discourse. IMO the real issue is being unaware of these biases. Introspection and constant self-evaluation is key.

Reporting to me is largely facts with a baseline humanist bias (i.e. AP, Reuters, AFP wire reports) while editorial is opinion seasoned with facts, or all too often, fact-like flavor enhancers. All I'm saying is flavor enhancer sensitivity is real, and Yelp makes tons of money helping you avoid low quality fare (not only due to the protection racket they're running).

Its going to be very difficult to report news without any bias.

I subscribed to a lot of contradictory crap on facebook after hearing about their data mining (Fox News plus Huffington Post, Obama as well as Trump). Its actually a lot more interesting seeing how both sides try to manipulate the truth.

Reply on news.ycombinator.com