Ricardo Baeza-Yates: “The majority of internet users are voyeurs who do nothing” | Technology



Everything on the internet is the work of a few. A "few" that are tens of millions of people, but that are only a small percentage of total users. The web is an infinite ocean that covers the entire earth, but the people who feed it fit rather on a large island.

Less than 1% of internet users create more than 50% of the content. "If you take any specific time segment, the percentage of active people on the internet is sure to be less than 10%," explains Chilean professor Ricardo Baeza-Yates. "I have seen it in places where I have worked. Most people on the Internet, especially on social networks, are looking at it, doing nothing. They don't even do a like. It does not generate data for the internet, which is not the same as being active. The people who contribute, who make a like, it could be 10%, but those who make a tweet or a post or post a photo, will be less, "he adds.

4% of users write reviews on Amazon and 2% write half of tweets on Twitter

In different works, Baeza-Yates has found that 4% of active users write reviews on Amazon ("and that one month after publishing the article Amazon began to pursue the paid reviews, with what the real number is smaller "), 2% of users write half of tweets on Twitter and that the first version of half of the Wikipedia entries in English was created by 0.04% of its users registered, about 2,000 people. "And it was because they were paid, because who participates in something that is empty," says Baeza-Yates.

On the internet there is also a human law that Spanish popular wisdom knows well: one does, many look. "It is the activity bias: few work and many do nothing. In any human activity, this bias is seen. Few pull, the others follow," Baeza explains. It is such an established phenomenon that it has a name: Zipf's law.

"The most important consequence of this phenomenon is that mass wisdom is an illusion," says Baeza-Yates in a conversation with EL PAÍS in Madrid, where he went to participate in the inauguration of the headquarters of the BBVA artificial intelligence factory . Baeza-Yates is too National prize of Informatics in industrial research and technology transfer 2018, director of graduate programs in data science at the Silicon Valley campus of Northeastern University, professor at Pompeu Fabra University and chief of technology at NTENT, a search company in Silicon Valley, where he lives.

The wise mass of the Internet is therefore not all humans who use the network, but who generate content. "It's the wisdom of people who are doing things. If there are groups that want to manipulate on Twitter in the United States, on Facebook in the Philippines or WhatsApp in Brazil they have a lot of power because there are many who do nothing, they just look, "he explains.

Baeza-Yates believes that an often common feature of the most active users is that they are "bad", as in the cases of the Philippines, Brazil or the United States. "People who want to manipulate are more active than those who don't want to manipulate," he says, "for one simple reason: people who want to manipulate have a goal and therefore a motivation to act, while most people don't have not even the motivation to participate. " The desire to annoy is a wonderful incentive in a world where the majority only appears.

These findings are part of a long investigation of Baeza-Yates on biases on the web. As the main channel of human communication, the biases produced by the internet are destined to mark many of our future decisions. The few active users are a habitual bias we have when we believe that what we read on Twitter is what "people believe". It is activity bias. But there is more.

Internet depth

What is in the bottom of the internet? The real feeling that everything is on the internet is not good for things we don't know exist. How does Netflix know that a movie doesn't like it if it hasn't even been taught to a sufficient group of users to decide if they want to watch it? Something similar happens with search results, especially now when Google tries not to leave its page to see what they are looking for. Who bothers to look at the result 35 of a search? A page can go up in the results thanks to links and other criteria, but the competition is growing.

"This presentation or exposure bias is the most serious. It is impossible to teach everyone to decide," says Baeza-Yates. Nor is it trivial to calculate: who is taught what nobody sees? For example, a Polish movie on Netflix. If 5% of North American users are randomly shown, are they representative and then show it to more? "And why in that case only 5%?" Asks Baeza-Yates. "Because I lose money," he replies. If Facebook risks teaching more than 5% of the posts that nobody allegedly wants to see, it risks users having less interest and end up seeing fewer ads to click on.

The long internet queue, where everyone had what they wanted because hanging a book or movie was almost free, has become the "digital desert." "The web has become almost infinite and has grown much faster than the number of people connected to the Internet," says Baeza-Yates. In a 2015 article, he tried to find approximate thresholds for that desert. 1.1% of tweets are written by people without followers, said Baeza-Yates, and 31% of Wikipedia articles modified in May 2014 were never visited in June. "The size of that digital desert is probably at the bottom of that 1% -31% range," says Baeza-Yates. And growing.

. (tagsToTranslate) most (t) user (t) internet (t) be (t) voyeur (t) make (t) wisdom (t) mass (t) false (t) false (t) people (t) active ( t) active (t) network (t) minority (t) explain (t) expert



Source link