Advertisers are suspending ads on YouTube

due to extremism concerns

[nonstop]

Major advertisers on YouTube are suspending their Ads because of extremism concerns.

Advertisers are pulling their Ads because they are appearing next to "inappropriate" material on the video-sharing site.

A recent investigation by the Times found adverts were appearing alongside content from supporters of extremist groups, making them around £6 per 1,000 viewers, as well as making money for the company.

Ministers have summoned Google for talks at the Cabinet Office after imposing a temporary restriction on its own ads - including for military recruitment and blood donation campaigns - appearing on YouTube.

[bbc.co.uk...]

An investigation by The Times found ads for dozens of leading firms have been shown alongside videos posted by extremists including David Duke, former leader of the Klu Klux Klan.

Numerous other racists, holocaust deniers and rape apologists have received payouts from Google for YouTube commercials.

Taxpayer-funded ads for various branches of the British Government were appearing alongside Isis propaganda videos and other offensive content.

[independent.co.uk...]

Current companies that have pulled advertising from YouTube:

The Guardian
Channel 4
BBC
UK Government
HSBC
Lloyds
Royal Bank of Scotland (RBS)
Marks and Spencer
McDonald’s
L’Oreal
Audi
French advertising group (Havas) that has clients including O2, EDF and Royal Mail
Dominos
Transport for London
Financial Conduct Authority

Sky, Barclays and Vodafone are understood to be considering whether to cancel their campaigns unless Google is able to resolve the problem rapidly.

[londrum]

what's their motto again?

[NickMNS]

@ChanandlerBong it is very easy to filter most adult content, the images are quite distinct from non-adult content, they tend show a lot of bare skin. This is relatively easy for image recognition software to pick out. One often hears reports about complaints about false positives for people posting videos on topics such arts or cancers.

A person talking about some subject as opposed to another person talking about a different subject are pretty much the same when it comes to image recognition. Even if the audio could be reliably converted to text and then analyzed, you still have the problem of differentiating between say a video that is critical of terrorism vs one that is advocating for it.

@londrum

they don't employ a single person to look through all the dubious videos

That is not true. Google's Eric Schmidt yesterday on the Brian Lehrer show (NPR) stated that filtering of video and fake news was being done both algorithmically and by humans. I posted a link to the interview yesterday here [webmasterworld.com...]

[ChanandlerBong]

@NickMNS, yes, I know how they do it, I'm guessing there are also certain sounds/words they can easily scan for, which they can also do for the types of videos that are being discussed here. Yes, there would be false positives, but it just means those that get flagged would have to be manually flagged as being 'ok'. They choose not to do that.

The problem is, G doesn't really see an issue with these extremist/hate videos, they earn good $$$ off them, so they don't bother to scan for and delete them. With advertisers now unhappy, there's a problem and they will come up with a quick and dirty fix and wait for the media frenzy to move onto the next target.

[blend27]

--what's their motto again? --

"We're Gorg"?

-- Major advertisers on YouTube are suspending their Ads because of extremism concerns. --

Will extremists get to enjoy Ad-Free YouTube now?

Jokes aside, I think providing a platform for extremist propaganda is plain wrong.

[nonstop]

@NickMNS

they don't employ a single person to look through all the dubious videos

That is not true.

it is true, Google's Peter Barron said to the UK Home Affairs Committee that they do not employ a single human being to pro-actively search through YouTube looking for extremist videos.

They are being reckless by not even looking for the extremist content that is being uploaded. It takes only 2 seconds to find these videos, but Google do not care.

Google relies on it's own users reporting videos which is a flawed method of policing content.

[nonstop]

It looks like extremists will always be paid by Google? and Google will always make a profit from the worst kind of content?

Google have admitted that they can't stop adverts appearing on extremist material

Eric Schmidt, executive chairman of Alphabet which owns Google which, in turn, owns YouTube, says the internet giant cannot guarantee that adverts will not appear next to offensive material online.

[engine]

nonstop, ES didn't say they can't stop adverts appearing, they said they cannot guarantee they won't appear. There's a difference.

I'm pretty certain Google will put a stop to it as soon as identified. Part of the problem is the sheer volume of material uploaded every day, and then identifying it.

Google really has to step up and use all the tools at its disposal to solve this. If it doesn't, it will continue to lose advertisers on YouTube.

[nonstop]

what's the difference?

[londrum]

they should start arresting a few of the top bosses for funding terrorists. i'm guessing they'd miraculously find a fix within a few days.

[NickMNS]

Here is how I see it.

This is making headlines now, so big brands are jumping on the band-wagon and calling Google out because that is what their customers want to hear. Google has received the message they will work harder to improve, but mathematically it is simply impossible to stop this completely.

The news cycle will churn, something else will grab people's attention and then slowly but actually quickly the advertisers will begin buying ads again seduced by the allure of the Youtube demographic. If you doubt the seductive allure of the 18 to 25 demographic (or younger), think Snap.inc IPO.

[goodroi]

Mods Note: Please make sure to stay on-topic and follow the community rules. This includes not posting personal attacks, flames, or calls to action. Let's be careful to keep the criticism on a professional level. Thanks.

[tangor]

It can be stopped 100% of the time, but g will not do it for many reasons, and most have nothing to do with ad dollars!

Human review of content BEFORE release on yt is one place to start. Human review of content (and ads) before putting the two together.

G is relying on "ignorance of content" to operate yt and ugc "exceptions". While these concepts of law (particularly copyright law) are very real we do know it is outright fiction that g doesn't or cannot know what is published at yt until "notified". If that fiction ever fails then yt is toast, g's advert dollars there will disappear, and copyright holders will make g pay and pay and pay.

g doesn't want to know, hence these convoluted and tortured explanations on why they "can't"

But I suspect a way will be found if enough big advertising dollars is withdrawn. You bet your bippie!

[NickMNS]

@tangor even if it was feasible to human review all the video uploaded to yt, human review in it of it self is not infallible. So again 100% is impossible, can they improve, sure. But they face the laws of diminishing returns.

[tangor]

Doesn't have to be infallible, it has to be ACCOUNTABLE ... and that, dear friends, is the crux of the matter.

[NickMNS]

So what are your proposing. You hire a moderator, if the moderator accidentally at no fault of his/her own misses a bad video, then the person should be fired. How has that stopped the video from being flagged?

[tangor]

Deliberately being obtuse? Or just missing the big picture?

G does NOT want to use mods. Never has. Rather blame the machine (which is programmed by humans) to absolve itself of liability.

IF a human mod is put in place then these "failures" will diminish dramatically, and when there is a failure there will be a PERSON that can be held accountable.

FROM A LEGAL standpoint g does not want to do this .... and that's the whole reason this problem exists in the first place.

[mosxu]

@NickMNS

Sadly advertisers do not understand what is really happening behind the scenes. The world does not have to change for Google but the opposite: Google needs to change for the world

[Robert Charlton]

From one of many answers on Quora, which make basically the same point...

How long would it take to watch all the videos on YouTube?
[quora.com...]

It's physically impossible to watch all videos on YouTube, because people upload 100 hours of video every minute. By the time you've finished one video, there will be 1,000 more added to the website.

If you assume the same growth rate for the past 10 years and assume that no more video would be uploaded until you stop watching, it would take you 60,000 years of non-stop watching to watch each and every video on YouTube.

This answer was from a couple of years back, and the volume of new videos is higher now. So Google must rely on algorithmic solutions, with algos undoubtedly seeded by human quality raters... as well as on user feedback.

[tangor]

Oddly enough, it only takes five or six seconds to determine a video for CONTENT purposes without watching the whole video. But that's a different proposition ... and one that a robot AI might not be able to do but HUMANS can.

But image matching is something that robots/AI can do and I'm pretty sure g has at least the same chops as my graphics program of choice which can find and match images (all 3 million of them).

Capabilities are there, it's the desire to use them, or be bound by them, which is at play. Admitting such capability opens one up to RESPONSIBILITY and that's a completely different (legal) kettle of fish.

g survives now by claiming not to be able to programmatically do this, while insisting that programmatically it can. Match ads and content or claim it can't match ads with content. Win Win (or Lose Lose), either way advertisers are spending money and the coffers are filled.

The magic of g is being questioned at this time and those questions are going to become more intense as larger groups (brands at the moment) morph to governmental agencies. Meanwhile .... make hay while the sun still shines.

There will come an advertising reformation on the web similar to the one that hit the print industry back in the 1930s and 1940s. Just not here ... yet.

[londrum]

And they'd only have to check the ones that are getting views. Half the videos on YouTube probably only get a handful

[goodroi]

Ok, that sounds like a reasonable idea. Wait till a video gets 10 or 100 views or more and then have it manually reviewed. Let's do the math.

There is at least 400 hours of video uploaded every minute of every hour of every day. Instead of assuming half the videos, let's be optimistic and say 90% of that never gets enough views to qualify for review. We are now talking 40 hours of video every minute. So we are going to need (40 hours x 60 minutes = 2,400 reviewers). But video is being uploaded around the clock so we need 3 shifts of 2,400 reviewers = 7,200 reviewers. Also let's remember Google is charging advertisers around $0.002 per view. So if a video has 100 views, Google earns $0.20 and has to pay someone to watch that video.

That does not seem like a trivial task to manage.

[londrum]

Why do people stick up for Google when they do wrong? I don't get it. They're showing videos that aid or support terrorism (you know what terrorism is: that thing where innocent people die). And not only that, they're making money from them as well! They're actually profiting from them. And this is a company that already makes bazillions of pounds every year. It's indefensible.

[nonstop]

If Google can't run businesses at scale safely for everyone maybe they shouldn't be running them.

100 hours = 6,000 minutes uploaded per minute
6,000 minutes = 360,000 minutes uploaded per hour
360,000 minutes = 8,640,000 minutes uploaded per 24 hours

if the average video length is 5 minutes then there are 1,728,000 videos uploaded per day

I think a human reviewer could check 100 videos per day

1,728,000 / 100 = 17,280 staff needed

17,280 x $30,000 = $518 million cost to review all incoming videos per year.

Google made a net profit in 2016 of $24 billion

even if it cost Google $1 Billion a year to check all the uploaded videos they could still afford to do it, and they could easily do it. Even if it cost $2 billion and each video was checked twice or three times they could do it :)

[NickMNS]

The cost of the staff is not merely the cost of the employees doing the work but also the required support staff and infrastructure. 17,280 is in the order of 1/3 to 1/4 the current size of Google. Increasing the head count of any corporation by that magnitude is a challenge.

This doesn't address the issue of finding qualified employees nor the issue of training, and retention, and then turnover.

Then there is performance, where there is no doubt that a few to possibly thousands of these employees could perform perfectly, but others, most will not and my guess would be that the overall performance of human moderators would be far less than that of a machine.

Finally there is accountability, which is what this is supposed to solve. How does that work with employees, "Dear advertiser we are really sorry this video was offensive. We have dismissed the employee that let your ad appear next to it.", What did that solve? The ad still appeared where should not have, except now some person needs a new job. Google's situation hasn't changed, but now instead of scapegoating their algos they are scapegoating low-level employees. The issue remains fundamentally the same whether you a human or a machine is doing the work. Human just cost more, and could be spending their time contributing to society in a much more productive fashion than spending there time watching cat videos.

[nonstop]

$500 million figure was just a ball park number, but even at $3 billion Google could afford it.

they choose not to do it.

I'm happy for them to use an algo but at the moment they don't have one and they can't just produce one over night. They've had 10 years to make an aglo but they haven't, and the algo needs to be just as good as a human being which I doubt is really possible.

[tangor]

Do not forget that real penalties (failure to perform) have real consequences. You can be sure that if g is penalized by a third party (be it an ad group or a governmental agency) the "magic" would start working pretty darn quick.

[toidi]

Why not just review the channels. 99.9999% of the yt channels are harmless and do not need to be monitored.

[nonstop]

Advertisers are demanding prime space at discounted prices from Google

Advertisers are demanding from Google prime space at discounted prices, after this month’s revelation that many brands had appeared next to extremist content on YouTube, the Google-owned site.

Rob Norman, chief digital officer for Group M, one of the world’s biggest media buyers and part of the WPP Group, said he has been in “constant dialogue with Google” suggesting “a range of actions they may take to give greater comfort and security” to advertisers.

Mr Norman said this included reduced rates for premium advertising inventory in order to give brands greater protection and avoid them appearing next to inappropriate, or extremist, content on YouTube.

[ft.com...]

[engine]

Prime space at discounted rates over this! That's just an excuse to, imho, but not a reason.

Programmatic advertising will tend to cause issues.
If, as an advertiser, If don't know where my ads are going, that's my fault. I'd want to control the ads to know which ones are in my market.Some just want to get to as many people as possible.

It's up to both the advertisers and Google, Yahoo, and others to better control their programmatic advertising. If it's done properly there will be better performing ad campaigns.

It's not easy to do, but it has to be done. The Washington Posts describes it as "whack-a-mole."
[washingtonpost.com...]

[nonstop]

Google's stock is down $41 since this all started

if anyone is interested you can view the UK home affairs committee grilling Google here:

[parliamentlive.tv...]