Ever since last winter I am a bit cautious when articles like this pop up. I was on a conference where one of the handful tracks was fact checking, and a keynot

Ever since last winter I am a bit cautious when articles like this pop up. I was on a conference where one of the handful tracks was fact checking, and a keynote speaker used images like the one in the article to illustrate how news media has failed during the 2016 election. One image that specially caught my attention was a word cloud made from polls about what voters knew about each candidate, and the speaker made a big point that all the major scandals was missing from Trumps side but the Clinton side was full of it. This rang all kind of warning bells, so I went and looked up the source and suddenly it made all good sense. The polls was done between the start of the preliminaries (which happen to be around the time wikileaks started publishing their stuff) and ended a couple months before the election day, and all those "missing" scandals about Trump that made international news are all more or less during those last couple of months. In that context the word cloud perfectly matched more or less everything what international news had been writing about the two candidates during the preliminaries and shortly afterward before the election campaign had really started. The researchers even noted this in their original paper, pointing out that if the polls had been done closer to election date then the data would look very different. I could have used many of the same material to hold my own keynote about how news has successfully informed the public correctly. I notified the conference organizers, but I doubt it had any effect since the message followed the political tides right now. People want to believe that fake news caused the election results, and that is to me a very worrying. First we need to establish if it did, and it seems that everyone is skipping that step.