Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Fake News Challenge

fakenewschallenge.org

Post reply on HN
Print view View original

31–40 of 233 posts

Re: Fake News Challenge

#31

Post by beatpanda »

I don't think they're going about this the right way.

The first step to identifying 'fake news' (even in NYT or WSJ!) is to be able to pull out attributions from the source text.

This would give you some idea of where the information in an article actually came from. "Fake news" will tend to be either unattributed or misrepresent the thing it's quoting.

But a good first step toward better media literacy would just be giving people a representation of who actually said the the things in an article, and that seems much easier.

Re: Fake News Challenge

#32

Post by jacquesm »

As long as Google still puts Russia Today and plenty of other sites of questionable plumage on display this is a futile effort. It only works if the big names in news aggregation start by culling the worst offenders.

Re: Fake News Challenge

#33

Post by cphoover »

dingo_bat wrote: post #17
sgustard wrote: post #11

Fake news is generally identified by the reputation of its author (New York Times vs Alt-Right Patriot Freedom Observer). Nothing published by the Times or Wall Street Journal (despite any editorial bias) is fake, simply because we know (except in the rare circumstance) that they apply journalistic integrity in sourcing and writing articles. Much of what's published on "fringe" sites, especially if no mainstream outl…

So in your assessment "fake" just means non-mainstream. Haven't seen the word defined like this before.

No I think he means... Articles that are not sourced (citations, reports, interviews, journals).... etc have a higher likelihood of being false, than those that do.

Mainstream news orgs tend to be better about including citations, and sources. They are also better about publishing retractions, when they are caught saying something factually wrong.

Re: Fake News Challenge

#34

Post by lend000 »

sgustard wrote: post #11

Fake news is generally identified by the reputation of its author (New York Times vs Alt-Right Patriot Freedom Observer). Nothing published by the Times or Wall Street Journal (despite any editorial bias) is fake, simply because we know (except in the rare circumstance) that they apply journalistic integrity in sourcing and writing articles. Much of what's published on "fringe" sites, especially if no mainstream outl…

Using "reputation" alone is a dangerous and naive standard -- the most powerful form of "fake news" is the selection/omission of stories and crafting of narratives around those stories, which is just as prevalent on "real news" sources like WaPo/CBS/CNN/NYT as it is on Fox/Breitbart and even 'fringe' sources.

With the issue of opinions derived from factual events, which fit the bill of being 'misleading' (to some groups, at least) -- how do you draw the line? The NYT is certainly not exempt from that (see: half of Paul Krugman's articles).

The only thing you can possibly check for with any reliability is the validity of the base, source event, if there is one. Any higher derivation in the wrong hands (Facebook, for example) is systemic censorship. How can we safely do that? Recursive link tracking, perhaps, to a "primary source," which relies on the AI matching the factual claim to an appropriate source, either a video, image, or text that can be considered the legitimate source based on context (for example, an article about Apple's earnings could link to a primary source of their press release).

Re: Fake News Challenge

#35

Post by cphoover »

dingo_bat wrote: post #17
sgustard wrote: post #11

Fake news is generally identified by the reputation of its author (New York Times vs Alt-Right Patriot Freedom Observer). Nothing published by the Times or Wall Street Journal (despite any editorial bias) is fake, simply because we know (except in the rare circumstance) that they apply journalistic integrity in sourcing and writing articles. Much of what's published on "fringe" sites, especially if no mainstream outl…

So in your assessment "fake" just means non-mainstream. Haven't seen the word defined like this before.

No I think he means... Articles that are not sourced (citations, reports, interviews, journals).... etc have a higher likelihood of being false, than those that do.

Mainstream news orgs tend to be better about including citations, and sources. They are also better about publishing retractions, when they are caught saying something factually wrong.

Re: Fake News Challenge

#36

Post by komali2 »

rdtsc wrote: post #25

Alas we are not supposed to use that term anymore: https://www.washingtonpost.com/lifestyle/style/its-time-to-r... Also somehow CNN shows in the front page of result for "fake news" image search on Google. It wouldn't be totally wrong, depending on how the term is defined.

Probably because the trump subreddit does that super annoying "Upvote this post so that people see CNN when they search for fake news!" thing every few weeks.

Re: Fake News Challenge

#37

Post by Clanan »

Is anyone else put off by this obsession with "solving" fake news? As if we can simply throw technology at such a contentious and controversial issue and then trust the results. It reads to me like the goal is censorship, where only "approved" articles/authors are allowed, all in the name of "fact checking". Never mind that it's incredibly easy to cherry-pick facts or exaggerations to target "offensive" articles (by both/all sides).

Re: Fake News Challenge

#38

Post by lend000 »

How about a 'subjectiveness' ranking instead, with notations for "unable to find link to a source of factual claims"? That's the best we could ever safely do, and we would likely find interesting results (WaPo might be labeled more subjective than Fox News, for example, even though both would generally have low numbers of notations for 'unable to find link to source of factual claim').

As I said in another post: The only thing you can possibly check for with any reliability is the validity of the base, source event, if there is one. Any higher derivation in the wrong hands (Facebook, for example) is systemic censorship. How can we safely do that? Recursive link tracking, perhaps, to a "primary source," which relies on the AI matching the factual claim to an appropriate source, either a video, image, or text that can be considered the legitimate source based on context (for example, an article about Apple's earnings could link to a primary source of their press release).

Regardless, the most dangerous form of "fake news" is by the selection/omission of stories, and that seems pretty impossible to quantify with today's technology (and of course the most prevalent on all of our 'real news' sources).

Re: Fake News Challenge

#39

Post by debt »

DanielBMarkham wrote: post #29

It makes me sad to see yet another tech team go down the road of "machines will help us filter the truth" They will not, and the reason has to do with language. Ludwig Wittgenstein tackled this 100 years ago. The best that machines will do for you is to label something as true or not as if you had consumed the article and decided on your own That is a completely different thing from identifying fake news or truth The…

Wittgenstein wasn't around when a president used Twitter to both control the media and get elected.

Maybe news could rely on first principles approach. Quantify things like political stability, information availability, cultural problems, etc. and maybe apply a first principles approach and maybe an algorithm could tell use what the hell is going on at any given moment and maybe give us some insight into what may go on in the future as well.

I would assume intelligence agencies have something similar to the above. The problem is things happen in realtime and if a regime falls or a state fails or whatever, we have to adjust our models immediately.

Throwing out hands up and saying Wittgenstein solved it seems lazy but idk.

Re: Fake News Challenge

#40

Post by sctb »

coolspot wrote: post #2

It is simple: everything that does not match your beliefs is a fake news.

As far as I can tell, this is a serious project that's likely to have sincere community interest. They define the problem in concrete terms and have created a framework to objectively measure the success of submitted solutions.

When we start with something substantive like this and take it straight up to generic ideological soundbites, we cheat ourselves of the kind of intellectually interesting discussion that we might've had instead.

Post reply on HN