Recently, Kommersant drew attention to the advertising of a dubious financial laboratory , which a number of Russian media missed. Signs of fraud were found easily, but some media did not give up the money - the most cynical of them said that they advertised not the financial pyramid itself, but only friendly meetings with its founders and employees. So to say advertised not contributions to MMM, but meetings with employees of MMM.

Observers were especially annoyed by the format of “native” advertising that was used, that is, the design decision that hides the differences between the format of advertising and the format of editorial articles.


Just below the link with the date you can see this whole post, and in the tape of Yuri there is also its continuation - he found several more examples of “crafty” advertisements from famous publications.

And the whole story is not unique. Here is a post by another author - also about mixing high-profile reputations and dubious ads. Lying in advertising and public relations is only a part of violations that are caught by advertising systems, starting with Google and Yandex and right up to social networks led by Facebook.
')
Google's 2016 report mentions such threats monitored on the AdSense network as:

1. banner ads for infecting computers,
2. intrusive adware,
3. disguise of prohibited advertising and news under the allowed (the Synodov’s cases fall into this category),
4. Bright news that hides links to go to allowed or prohibited ads
5. made-up news like black PR

Unfortunately, there are not so many numbers in the review, only one tenth of the cases out of 1.7 billion deleted ads, punished websites and blocked advertisers' accounts in the past year are described in total.

Here are some interesting touches:

• about 6.5% of pseudo advertisements refer to virus infiltrations through system alerts;
• the most active in advertising are short-term credit sellers. 1 site has 625 ads;
• Slimming sites are punished 3 times more than spreading malware and 6 times more than websites issuing loans or sites selling fakes for branded brands.

The report for 2016 was not the first and usually they did not cause much excitement. But last year, elections in the United States made of fake news just the same bogey, presented to the public as a universal evil. Media moguls and politicians were outraged that the media monopoly to form the agenda was bypassed by secondary publications through social networks. Calling all the news that flowed to the facebook audience in an uncontrolled stream - “fake”, the largest media outlets demanded that Zuckerberg and other publishers close all the gaps.

To block dubious news, they remembered the fact-checking technology. This is how the fact-checking is called through the additional editorial department of control on hints and readers' complaints. The department of verification can be both within one media company, and with several. Facebook has hired four companies Snopes, Politifact, ABC News and FactCheck.org, which will "handles" to check the accuracy of the news appearing in its feed. If at least two of these four hired companies find the content of the news questionable, users in the feed will see a pop-up window with a warning about this and a question if they want to repost this message.

The speeds of Google’s analytical system are dozens of times higher than a person’s ability to analyze content and track questionable ads (the authors write that it would take a person 50 years to block a high-speed round-the-clock blocking of ads caught in 2016). Dozens of times is not so much. Trading robots process signals tens of thousands of times faster than humans. Obviously, content recognition and questionable responses can be improved.



In this case, the fight against fake news itself looks like a Sisyphean task. First, its share is insignificant. Especially if this mechanism is piled up just for the sake of politics. Political advertising does not fall into the top ten advertising categories on facebook, for example, even in the election year. Secondly, the struggle for viewing has already led to the fact that previously almost yellow press - for example, the online edition of Buzzfeed - was not much different from the once-respected newspapers and vice versa.

The English daily newspaper, the second after The Sun in terms of circulation, received the status of “unreliable source” from Wikipedia. Editors Wiki believe that the newspaper is looking for sensations and publishes fakes . And this statement is justified. When the Daily Mail published a text about the Australian criminal gang Apex, the newspaper took a photo of a popular rap group as an illustration of a street gang. From the high-profile fakes, it can still be recalled that the Daily Mail is suing Melania Trump because of an article stating that the latter worked in the escort service in her youth. Or recall the news that the late Russian ex-press minister Lesin may be alive and testifies against Vladimir Putin at the FBI.

So the line between tabloids, the yellow press and serious media today is different only for those who order music there - media moguls, professional publishers or the public. The credibility of the seriousness of the publication now has little to do. By the way, “Men in Black” knew everything about the problem of the quality of news and where to look for facts 20 years ago.

How fake news is recognized today

Facebook, in addition to attracted people-controllers, there is also an automation of the understanding of the content. The fastText library analyzes the text, highlighting its essence. More precisely, fastText allows you to organize automatic assignment of categories for arbitrary text using machine learning methods. When testing a model of 1 billion words, the neural network learned to put them into 300 thousand categories in 10 minutes, and then it took less than 5 minutes to process a sample of 500 thousand posts.

But this is a theory. How so far in practice? Assigning a category to understand the essence of the news is not enough. To recognize the fake news, technologies automatically need to understand the context, that is, relate different events, determining their priority and the essence of what is happening. As an example, let's see how it works in a much simpler example - in advertising targeting. Everything is very formal there. They look at the user's data, look at the history of searches and views, and then “sell the viewer” to the advertiser whose banner is being displayed. What is the result?

In the summer of 2016, Google showed on Youtube an advertisement for impotence for viewers of a children's cartoon about Pepe Pig. The authors of the ad understood the hooligan nature of the slogan and accompanied the advertisement with a warning “Read until you have deleted”. There were many indignant viewers, and especially active deputies even demanded to block Youtube in our country. A clear example is that advertising is not analyzed for the adequacy of the neighborhood with the viewed materials. Viewing a cartoon about mumps can be very happening on papa's or mom's tablet, who previously searched for a similar drug. And then they watched something non-child. It is clear that at the device level, the user of a particular movie or article is not tracked. It is also clear that now there is no well-functioning analytics of the content of advertising and materials.



And since neither the advertisers themselves, nor the users know which advertising will work, and which will be annoying or ignored - in principle it is impossible to adjust the filters to search for the desired advertising. Such a filter, in theory, should show only what a person needs right now and need it in this particular configuration, design, price and delivery. That is, such a filter should, firstly, perform the above-mentioned task of precisely hitting an actual need, which none of the advertising systems have yet, and secondly, weeding out all non-working advertising means “to kill” 90% of the advertising market. It's clear that no one will do such a filter.

Incorrect advertising as a reflection of economic competition

Obsessive, hidden and just false advertising is much more common fake news. Every month, for every 4,000 faulty sites, only in one in the category - “advertising of dubious products for weight loss” - there are 250 sites punished for disguise under various news.

On the Internet, questionable advertising is much more than quality. The rule “it is necessary to gain a foothold in the subconscious as early as possible and as deep as possible” forces the advertiser to show an advertisement for motor oils and winter tires to a person when he has just begun to learn to drive. And show it again 50 until the moment the question becomes relevant. A classic of advertising Ogilvy said about this situation: “I know that only half of the advertising works, but alas, I don’t know which one.” It is extremely difficult to determine what argument and when it became decisive for the purchase, if we start the promotion six months before the choice of goods. But you have to start. Formats and promotion channels designed for a “hot” buyer are as expensive as possible (in industry terminology, the actual need is a “hot client”, a “warm” one - at the moment of awakening interest in a particular product, a “cold client” is a contact to determine the theoretically possible interest when -or future buyers).

Obviously, not everyone can give expensive advertising to the "hot buyer". Someone has low profitability, while others have a poor budget. For them, there are experts who promise a good result in other formats at low prices. So there is an advertisement that works poorly. For it is located in channels far from “hot buyers” or in unsuccessful places. As a result, there are 10 sellers per 1 buyer, of which 2 trade figuratively speaking on the “main street”, 3 “in the lanes” and another 5 “on the roofs of houses”.

But the customer still wants from advertising effectiveness. Among this temporarily meaningless advertising is fierce competition in obsession. The headlines are getting brighter, the priority in showing is more insistent, the gap is deeper from the point. We have to contend with other sellers, as well as with filters from advertising systems and banner networks. The main thing is that the display of 10,000 stupid advertisements gave at least 1 real customer and the whole snowball paid off. Therefore, the better it bypasses all the filters - the higher its demand. The customer may not even know that he is advertised through the aggregators of gossip and viral plugins and pop-ups. The intermediary loses 99.99% of traffic, but sells 0.01% at. It seems the format itself is slowly "dying." Screaming headlines, shamelessly inflating a minor detail in the article being sold, did their job. Their traffic buyers are less and less qualified, news providers are less and less respected. Passed its peak popularity MarketGid or Inforotor. But the sites are changing, but the business remains. New players came to replace them and again collect their traffic.



Google is struggling with fake ads in general, Facebook - with fake news. But how much will their technology change the balance between quality and incorrect advertising? Will the slots where the adventurous Internet user will wander and where will the opportunity to show cheap flashy ads? Of course there will be - no one killed and will not kill pornoresursy, forex trading and other services that avoid legal regulation. They have a lot of their audience. These resources will be advertised, create your websites and transfer to any loyal audience any costs. But with ordinary goods, where profitability must be considered more carefully, the situation is not much different. All advertising transferred to the sector "hot sale" is impossible. The inevitable reduction of sites and formats leads to higher prices and most advertisers will again go to slightly less accurate, but significantly less expensive formats and sales channels.

Sites of the second and third echelon are likely to be ready for a long time to put "yellow advertising and news" aggregators on their sites in an attempt to increase profitability. They can be killed only by a fundamentally different way of working with information on the Internet. Otherwise, there will always be leaders and there will always be laggards, ready for bad advertising. There are other models, but this is already a topic of conversation about startups.