2 ms·

Here's an easy example: http://abcnews.com.co/donald-trump-protester-speaks-out-i-was-paid-to-protest/ http://abcnews.com.co/donald-trump-protester-speaks-out-

by sulam 10y ago

Here's an easy example: http://abcnews.com.co/donald-trump-protester-speaks-out-i-was-paid-to-protest/ http://abcnews.com.co/donald-trump-protester-speaks-out-i-wa... Google is probably just as easy to fool as Facebook. This was the first hit on a search for "protesters paid to protest", a story that is demonstrably fake: https://www.washingtonpost.com/news/the-intersect/wp/2016/11/17/facebook-fake-news-writer-i-think-donald-trump-is-in-the-white-house-because-of-me/?hpid=hp_rhp-top-table-main_fake-news-845a%3Ahomepage%2Fstory https://www.washingtonpost.com/news/the-intersect/wp/2016/11... Immediate red flags should be: This isn't ABC News's website. There are statements in the article like: "No one still has an AOL email address except people that would vote for Hillary Clinton" and from a made up Trump supporter quoted saying "I knew those weren’t real protesters, they were too organized and smart..." and (awesomely) "I think I was paid more than the other protesters because I was white and had taken classes in street fighting and boxing a few years back". I mean, this may well be parody, but it gets picked up as real news and he's making real money off it, impersonating a real news site, which is the point at which it cross the line from parody to fake, IMO.

garysieling 10y ago
You can tell this site has no credibility based on the other articles on the site. I think the trick with a machine learning approach to identifying these sites is not letting the model fixate on particular words (e.g. Obama or Moon Landing could be terms with the right dataset; though if they occur together maybe that's a good indicator after all...)
nkurz 10y ago
I mean, this may well be parody "May well be" is being unnecessarily conservative. Maybe you didn't make it far enough into the article to read the faked interview with the author of the the Snope's article I posted: David Mikkelson, founder of Snopes.com, a website known for its biased opinions and inaccurate information they write about stories on the internet in order to generate advertising revenue, told ABC News that he approves of what a story like this is accomplishing. “You have to understand that when a story like this goes viral, and we spend a minute or two debunking it, we make lots of money. Stories like this have helped put my children through college, buy a new car, a home and even get the Silverback gorilla my wife Barbara always wanted since she was a child,” Mikkleson said. Buried, but actually rather funny. In a way, this is both a fabulous and a terrible example. Read in its entirety, it's clearly satire. Yet I'm sure that many people never read it closely enough to notice. Should Facebook be censoring this? No. Should they be labelling it? Yeah, it might be a good idea. Is this the sort of story that the author is referring to? I don't know, but it would be nice if he'd tell us explicitly.
- hga 10y ago
  Given who Snopes assigned political "debunking" to and the predicable results, you could also label this item as "Fake but Accurate", the "website known for its biased opinions and inaccurate information" is accurate enough for enough of their political items. One things I don't get, though, is why so many organizations destroyed whatever credibility they had remaining in a futile attempt to win this election for their preferred candidate, I mean, this was an "it's my turn" race, a bit like Dole in 1996 and Romney in 2012. Especially prior to Trump winning the Republican nomination, was it really that important? Note that this goes all the way to the DNC for the party itself....

by sulam 10y ago

garysieling 10y ago

You can tell this site has no credibility based on the other articles on the site. I think the trick with a machine learning approach to identifying these sites is not letting the model fixate on particular words (e.g. Obama or Moon Landing could be terms with the right dataset; though if they occur together maybe that's a good indicator after all...)

nkurz 10y ago

I mean, this may well be parody "May well be" is being unnecessarily conservative. Maybe you didn't make it far enough into the article to read the faked interview with the author of the the Snope's article I posted: David Mikkelson, founder of Snopes.com, a website known for its biased opinions and inaccurate information they write about stories on the internet in order to generate advertising revenue, told ABC News that he approves of what a story like this is accomplishing. “You have to understand that when a story like this goes viral, and we spend a minute or two debunking it, we make lots of money. Stories like this have helped put my children through college, buy a new car, a home and even get the Silverback gorilla my wife Barbara always wanted since she was a child,” Mikkleson said. Buried, but actually rather funny. In a way, this is both a fabulous and a terrible example. Read in its entirety, it's clearly satire. Yet I'm sure that many people never read it closely enough to notice. Should Facebook be censoring this? No. Should they be labelling it? Yeah, it might be a good idea. Is this the sort of story that the author is referring to? I don't know, but it would be nice if he'd tell us explicitly.

hga 10y ago
Given who Snopes assigned political "debunking" to and the predicable results, you could also label this item as "Fake but Accurate", the "website known for its biased opinions and inaccurate information" is accurate enough for enough of their political items. One things I don't get, though, is why so many organizations destroyed whatever credibility they had remaining in a futile attempt to win this election for their preferred candidate, I mean, this was an "it's my turn" race, a bit like Dole in 1996 and Romney in 2012. Especially prior to Trump winning the Republican nomination, was it really that important? Note that this goes all the way to the DNC for the party itself....