How to influence what millions of people read in Finland
In summer 2016, when the American presidential election campaigns were in its main phase, I noticed strange behaviour in the Ilta-Sanomat “most readed articles”-list. All news articles that were talking negatively about Russia or Donald Trump were falling very fast. These news articles could drop over 30 points in 10-20 minutes while some nonsense articles took their places. For me it seemed that someone was influencing what people in Finland were seeing in this specific news media.
My suspicions were raised even more when Cambridge Analytica released a presentation of “The Power of Big Data and Psychographics” at Concordia Summit in September 2016. In this presentation Cambridge Analytica’s spokesperson Alexander Nix plainly explains how they are manipulating election behaviour based on data they are gathering from American people. I thought that this was outrageously dangerous and surely just a tip of an iceberg. If the manipulation campaigns were so easy and cheap to do in the States then surely there are huge ongoing influence campaigns all over the world.
First I told my father about these suspicions of news article manipulation, but he of course didn’t believe me… Luckily after a few days, his curiosity won and he tried what would happen if he would select a specific news article and open it multiple times in a browser with incognito mode on. The finding was striking. He was able to raise a news article’s ranking from 70 to 50 in 30 minutes just by manually opening and closing it.
After this I started to build two scrapers that would scrape every 10 minutes all top 100 rankings from the “most readed”-list on Ilta-Sanomat and Helsingin-Sanomat web pages. I let the scrapers run for a year and my hope was that from the scraped data I could pinpoint a manipulation activity of any article.
However, this task was harder than I had anticipated. It is hard to say if some news article is influenced or is there some other reason why it’s dropping fast. Therefore, I had to invent something else to prove my point. Nevertheless, I could find categories that contained quite a lot of fast-dropping articles: Politics and Diginews.
My second way to prove this vulnerability was to create my own manipulation script. The script’s idea is the same as what my father had done earlier. Open a specific news article multiple times in a browser with incognito mode on. My script can open a specific article around 500 times in a minute. With this ability, I have managed to raise any article from the place of 100 to place of 14 with only one computer. I believe that with a few more computers it would be possible to raise any news article to the first place in “most readed”-list. I also assume that in order to gain full control of the “most readed”-list I would need only around a few dozen computers.
Final words. When the war started in Ukraine I told about this vulnerability to the Sanoma Media and Alma Media’s top management. These are the two biggest media houses in Finland and both of those have had this same vulnerability in their news web pages. I got answers from both media houses and they said that this issue will be notified inside the companies. I hope that by now they have fixed this issue.
