Slashdot Mirror
Google AdSense Banned a Random Webpage About a 32-Year-Old Bill Because It Was About Sexual Abuse (vice.com)
An anonymous reader quotes a report from Motherboard: Earlier this week, an algorithm made an absurd choice. Google AdSense, Google's advertising program that makes up the bulk of the tech giant's advertising revenue, decided that a web page about a decades-old bill about sexual abuse was "adult content," and wasn't allowed to display ads anymore. The page, which is at least six years old and contains strictly legislative information about a bill called the "Child Sexual Abuse and Pornography Act of 1986" on free legislative research and tracking website GovTrack.us, tripped the AdSense algorithm that decides what pages are allowed to run ads. This single, very dry page being flagged as "adult content" is most likely a minor fluke in the AdSense algorithm, but it's a perfect example of how a tiny tweak in the way a platform uses automation to enforce policies can send a ripple through seemingly-unrelated parts of the internet. The page was flagged by Adsense as "policy non-compliant" on Monday, with Google citing the page's "violations" in a summary of the AdSense adult content policy. Here's what Google told GovTrack: "As stated in our program policies, we may not show Google ads on pages with content that is sexually suggestive or intended to sexually arouse. This includes, but is not limited to: pornographic images, videos, or games; sexually gratifying text, images, audio, or video; pages that provide links for or drive traffic to content that is sexually suggestive or intended to sexually arouse." The GovTrack page contains none of these, yet the page still can't run AdSense.
bots/algorithms make mistakes, get over it.
Activists + technical incompetence + blind reliance on technology.
The holy trinity of how to fuck things up.
Confucius say, "Find worm in apple - bad. Find half a worm - worse."
This proves that they (google inc that is) is not fit to manage this sort of thing.
Years ago, a search that I did was blocked because it contained the letters SEX. It wasn't for sex but Middlesex but google decided in its wisdom that this was not allowed.
Middlesx is the name of a county in the UK, the name of a University and a County Cricket Club.
But the Google Puritans will keep on trying to censor our lives.
There's just too much internet to have any hope of filtering it manually, and yet filtering is required both for legal reasons and to avoid social outrage. So automated filtering is the only way - and we all know how well that works. It's got high rates of both false positives and false negatives.
It's leading to a lot of conspiracy theories though. If you read politically slanted news sites, you will see that those on the right are full of stories about how Facebook and Google are striving to force conservative voices off the internet - and if you flip the other political faction, the news sites on the left run many (though not as many) about how Facebook and Google are striving to force liberal voices off the internet. Just throw in a bit of reporting bias and everyone things the algorithms are out to get their team.
Google could turn the filtering off - but then it would be about a week before stories start appearing about how Google "supports prostitution" or somesuch, and politicians would start to threaten regulation again.
What's the downside here?
AI = Artificial Idiocy
GovTrack.us is not a taxpayer funded site.
https://www.washingtonpost.com...
Stories like this are designed to make us think... "censorship algorithms make mistakes".
They also make us think... "censorship on the internet is needed to protect us".
And, "most people want censorship".
The problem is censorship. The problem is that the algorithm mistakes are a smoke screen to cover the actual intentional censorship that is taking place. Facebook and Twittter want to make sure that anti-globalist people like Trump are not elected. Facebook and Twitter and Google want to make sure they stay in the good graces of the communist party of China.
Face it, Facebook and Twitter have become public forums. There is no place for censorship in a free society. The algorithms are designed and tuned to deflect attention from intentional censorship, which is a threat to democracy everywhere.
Why would they bother with AI? They won't increase the worth of AdSense by including a few more sites that now are falsely flagged, for Google this is a non issue.
It looks like Twitter and Facebooks are Public Forums in at least some circumstances, according to a Federal judge
https://irontrianglepress.com/2018/05/25/presidents-twitter-account-constitutes-public-forum/
Would it be OK if a privately owned toll bridge required drivers to remove all their pro-obama bumper stickers before crossing the bridge? Would it be OK for the bridge company to require people to remove their hateful anti-Islamic bumper stickers from their cars before crossing the privately own bridge?
And, suppose that I'm offended by your subject line due to my personal religious beliefs (I know many sincere Christians would be offended by your use of Jesus's name as an expletive). Do you think would be reasonable if you were censored from participating on Slashdot because of my sensitivity? Why is it OK for you to offend me with your religious hate speech?
Definitely legal language is something which children shouldn't be exposed to. Otherwise they never learn write and speak on human language correctly.
Bots and algorithms are trained by people. The people who manually intervene at Facebook and Google have shown a remarkable consistency in letting their personal views inform their decisions. Case in point: the demonetizing of Prager University by YouTube. The fact that Google did not intervene and fire a segment of the YouTube workforce over that manual decision is a perfectly valid data point to call into question the culture training these models.
You said when it comes to flagging suspicious content." I think that's an important word, suspicious. There are now more when pages than there are people on the planet. Billions of web pages to run ads on. Pages which include several words related to child sexual abuse are suspicious. Why put your ads on suspicious pages?
More specifically, on a page that has sexual words, so seems to probably be about sex, what kind of ads would be run? Perhaps a Trojan condoms ad? How pissed would Trojan be when the screenshot was going around Facebook of Trojan advertising on a page about sexual abuse of children?
This reminds me of The Illustrated Presidential Report of the Commission on Obscenity and Pornography, back in 1970.
As Wikipedia says:
The report was resoundingly rejected and denounced by congress.
In response to that, along with continued attempts to, nevertheless, enforce existing, and impose new, anti-pornography laws (and otherwise harass publishers of erotic images), Earl Kemp published an illustrated version of the report, consisting of the report's text but "replete with the sort of photographs the commission examined.".
For distributing this book he was sentenced to a year in prison, and served the federal minimum of three months and one day. (This brings up the question of how one is supposed to have a right to view something it's a crime to provide.)
Nearly half a century later the same sort of attacks on free speech continue on the new Intenet medium. And a handful of copies of the ... Illustrated Report ... are available on Amazon (with the asking price of an unused copy of over half a grand.)
Bantam Dominique roosters crow a four-note song. Once you've heard it as "Happy BIRTHday" you can't NOT hear it that way
The problem is when there is no human oversight to correct the inevitable mistakes that bots make either before or after the fact.
Web plublication (some of it automated) occurs at computer speed and there's a lot of it. _Live Science_ estimated total internet traffic at a zettabyte per year, and that
Bantam Dominique roosters crow a four-note song. Once you've heard it as "Happy BIRTHday" you can't NOT hear it that way
In some ways, yes. In others, the whole problem in Westworld is that the bots went and did their own thing incorrectly and no humans remained in charge.