Web of Lies: A Commentary on Power, PageRank, and Fake News
The term ‘fake news’ has been thrown around a lot lately. Maybe the term’s growing use is due to the president’s utilization of it. But, upon inspection, the fact that there is more ‘fake news’ abound than there was before makes plenty of sense for a few seemingly simple reasons. We have more easily accessible technology around us, and it’s easier to share information with a wide network of people. If a group of people believe that an article is giving proper information, and share it with their friends, and their friends read that information, and they may subsequently ‘share’ or upvote/like that link. So what happens if the article being shared is actually fake news?
Going back to our lectures on power in a network, let’s consider a network of friends, A, B, C, and D. Let’s say that all of these nodes are connected to all other nodes, but except for A, none of the other nodes have any other connections. A has a separate friend, E, who doesn’t know A’s other friends. B, C, and D like to stick to their own crew.
Now, imagine that E gives A a piece of unique information. Even though A doesn’t have to believe E’s information, it’s the only exterior source of information that the node has on this specific topic. A now holds information that B, C, and D do not have. Thus, A is in a power position within their own friend network. B, C, and D must rely on A for information, and it’s A’s choice to distribute this information. If the information that B, C, and D receive is fake, they wouldn’t know.
Now imagine E as the distributor of a fake news article, A as a site like Facebook, and B, C, and D as you and two other friends. Once the article is distributed, and then posted, how do you know that it’s not real? If it’s something that’s easy to label as false, or easy to check on another site, then maybe you can avoid it. But if it’s harder to discern, you may click on it. If you like what you see, you may give it a like. Your friends will see that you liked it. The news may then be available in your friends news feeds. They may see it. In a large network of friends, this process will keep propagating for a bit until the article’s popularity plateaus. But now what happens? Now a whole bunch of people have just read a fake article, maybe posting it to their own blogs, posting it to their pages, mentioning the link, etc. With more sites pointing to the fake article, it’s PageRank score will increase, making the article easier to view for people who didn’t even want to see it.
In the article that I read, the author discusses how Facebook, Google, and other companies are using A_ algorithms to detect fake news sites. The algorithm has a lower rate of success detecting fake news sites than real people do, but it is cheaper to implement. However, the Chief Security Officer at Facebook, Alex Stamos, believes that “in any situation where billions of items are sorted, algorithms have to be used”. While he does make a very sensical point, as the algorithm fails on detecting fake news, and reinforces those failures (thinking that they’re successes), more fake news will thus be spread. Over time, the AI-algorithm will grow to not detect as many fake news articles as it would have in its earlier stages. But, being that big name companies are behind this, I’m sure they’ll find a way to minimize the fake news detection failures as much as possible.
So if people may choose to believe fake news due to a lack of knowledge(which in this case translates to a lack of power), and if algorithms, trained by these humans, fail more often than the humans do, what’s the solution? What’s the point of even talking about this if I’m not going to give a way to solve the problem? Those are good questions. I write about fake news because I think that most of what we receive isn’t actually the truth. Truthful articles may be labeled as ‘fake news’ because certain companies don’t want information getting out, just as fake news/science articles may be endorsed by brand names to support a product that their marketing. It’s a real issue. I believe that the topics that we’ve gone over in class, specifically web search and the power dynamics of a network, are very helpful in determining how to solve the problem of fake news.
Link: http://www.computerworld.in/feature/can-algorithms-really-tackle-fake-news-fiasco