Can algorithms really tackle the fake news fiasco?
http://www.cio.in/feature/can-algorithms-really-tackle-fake-news-fiasco
This article discusses how Google’s PageRank algorithm is affecting the battle against fake news. There have been many instances lately of fake news, or social media manipulation, being extremely prominent and altering public opinion. Google’s PageRank, much like what we talked about in class, is described as ranking “search results based on the pages that are linked to the publication,” and it works across multiple platforms, which is what makes it so effective. Google also has a reverse image search that can pinpoint the source of an image from its upload. The ability to trace this link back contributes to the fight against fake news. Facebook also has algorithms to identify fake news, but started with and still uses some human intervention. Algorithms can be manipulated and can lack judgment in complex decisions, so sometimes humans are more reliable, yet human intervention is not always a feasible solution with such grandiose amounts of data.
I think the difference between human and algorithm analysis of pages is evident in the way we examined hubs and authorities in class. We had to choose a number of rounds for which to update, or we could update infinitely many times. For page rank, updates eventually stabilized to equilibrium values of page rank, but it is easy to see how with large number of pages it is nearly impossible to account for every necessary update without some type of computer algorithm. Especially considering the speed and high volume with which new content is added. Yet, a human mind is also valuable in determining when to “stop” calculating. Some pages may be so obviously bad, yet could be linked to other equally bad pages, and receive much higher scores than they deserve. An algorithm may not be able to identify this situation from a newly published, valid page. Another problem I see becoming possible is with the momentum of news. A fake story may be published, and reputable sources may pick it up before properly analyzing, and soon it has spread like wildfire. Algorithms as we have them now, can’t quite account for all of these high complex edge cases, but they are certainly a necessary start.