Fake Scholarship on “Fake News”
When I first heard this new “fake-news” charge my impression of it was that it was simply the latest version of the cry originated by the desperate conventional news media that “you can’t trust anything you see on the Internet.” That dodge clearly hasn’t been working, as their declining circulation and advertisement revenue indicate, and, furthermore, they have had to resort to the Internet themselves to try to salvage their bottom lines, so they had to refine their attack on the Net.
But it didn’t take long for those defenders of the alternative media, particularly supporters of Donald Trump, to take over the fake-news charge and turn it on its originators when they found examples of what they deemed to be less-than-accurate accounts of events in conventional news media, which everyone knows are out there in abundance. That seemed to have neutralized the charge to a considerable degree.
The Professors’ Paper
It was with some surprise then when I ran across a monograph published by the prestigious National Bureau of Economic Research (NBER), of all places, entitled “Social Media and Fake News in the 2016 Election.” It will cost you $5 if you want to get it from them, so if you want to check what I have to say about the work you need to go to the Stanford University site where one of the paper’s two authors, Matthew Gentzkow, is employed as a professor of economics. The other author, Hunt Allcott, is an economics professor at New York University.
Their paper is long on scholarly trappings, methodological razzle dazzle, and especially on innuendo, but it is very short on good scholarship. Without the insidious, establishment-supporting innuendo, in fact, it really amounts to little more than a showy belaboring of the obvious.
When you start off, you think you might get a bit more than the paper ends up delivering. After an opening paragraph summarizing the evolution of the news media and its effect on the American practice of democracy, we have this:
Following the 2016 presidential election, the focus of concern has shifted to social media. Social media platforms, such as Facebook, have a dramatically different structure than any previous media technology. Content can be relayed among users with no significant third party filtering, fact checking, or editorial judgement, and an individual user with no track record or reputation can in some cases reach as many readers as Fox News, CNN or the New York Times.
Among the most prominent concerns has been the impact of false or misleading information – “fake news,” as it has been dubbed in the public discussion. Recent evidence shows that: (i) 62 percent of U.S. adults get news on social media (Pew 2016a); (ii) the most popular fake news stories were more widely shared on Facebook than the most popular mainstream news stories (Silverman 2016); (iii) many people who see fake news stories report that they believe them (Silverman and Singer-Vine 2016); and (iv) the most discussed fake news stories tended to favor Donald Trump over Hillary Clinton (Silverman 2016). Putting these facts together, a number of analysts and commentators have suggested that Donald Trump would not have been elected president were it not for the influence of fake news spread through social media.
At the end of the second paragraph there is this footnote:
In an interview with the Washington Post, fake news writer and promoter Paul Horner said that “I think Trump is in the White House because of me. His followers don’t fact-check anything – they’ll post everything, believe anything. His campaign manager posted my story about a protester getting paid $3,500 as fact. Like, I made that up. I posted a fake ad on Craigslist”
Net News Shortcomings?
Concerning the first paragraph, the claim that the conventional press is more reliable than what one can get on the Internet is an old one, put forward with those with an interest in maintaining the status quo, and it is obvious balderdash. On its face the Internet is far superior as a source of information because of the much greater opportunity that it provides for learning what is true, which greatly overrides the danger of being deceived by what is false.
Showcasing Dubious “Facts”
Now let’s take a closer look at the most extraordinary, even shocking, claim that the authors in that second paragraph characterize as a “fact,” that the most popular fake news stories were more widely shared on Facebook than the most popular mainstream news stories.” Really? For that claim, they have a single source, an article by the founding editor, Craig Silverman, of the very partisan left-wing online publication, BuzzFeed. A line chart dominates the article with numbers on it purporting to show how the total number of “engagements,” that is, the “shares, reactions, and comments” on Facebook on the top “fake” news stories came to eclipse the same number of those for “mainstream” news stories in the run-up to the election.
The first thing we notice is that BuzzFeed’s definition of fake news seems nowhere to be specified, as it is by the authors in their own original research, that is, “news stories that have no factual basis but are presented as facts.” Of the “Top 5 Fake Election Stories by Facebook Engagement” that BuzzFeed lists, only the first and the last definitely fit the Allcott/Gentzkow definition. The apparent main fault of the other three is that they have misleading headlines. From what I know of the subjects covered, one could not say categorically that they have absolutely no basis in fact. One suspects that the number of “fake” news story engagements is greatly exaggerated because of an overbroad definition of what constitutes fake news.
Our second observation is that Allcott and Gentzkow, themselves, have falsely stated what BuzzFeed claims, when they say that the “fake” news items were more “widely shared” on Facebook. Engagements also includes reactions and comments as well as shares. Everyone weighing in who might say in so many words, “This is obviously phony and not worth anyone’s time” would be counted as an engagement.
Finally, one has to read all the way down to the end of the BuzzFeed article to get this statement, which one might take as a sort of disclaimer:
It’s important to note that Facebook engagement does not necessarily translate into traffic. This analysis was focused on how the best-performing fake news about the election compared with real news from major outlets on Facebook. It’s entirely possible — and likely — that the mainstream sites received more traffic to their top-performing Facebook content than the fake news sites did. As as [sic] the Facebook spokesman noted, large news sites overall see more engagement on Facebook than fake news sites.
Of course they do, as anyone with an ounce of gumption knows, so never mind this very misleading article with its misleading title that we’ve just put over on you, Silverman might just as well have written in conclusion. Then to further mislead us, whether intentionally or not, Allcott and Gentzkow misrepresented and oversold Silverman’s misleading analysis.
Now let’s have a look at that spectacular quote spotlighted by our scholars in footnote 2, the one in which a guy claiming to have found a way to make a handsome living by planting hoax stories on the Internet also claims inadvertent credit for Donald Trump’s election. I might as well talk about him because it looks like almost everyone in the mainstream news media, and then some, has talked about him. Try doing an Internet search for “Paul Horner fake news” and look at the pages of hits that come up, and where they come from.
There’s also a Wikipedia page on the guy, which began in late 2014, when he was described as an “internet news satirist and article writer.” Most recently the site begins, “Paul Horner is a contributor to fake news sites.”
Whatever you want to call him or whoever he really is, it’s pretty clear that he has been something of a godsend for the press, led by The Washington Post, looking for any straw they can grasp to explain how they could have been so wrong about the election. His big hoax score was getting Trump’s campaign manager to retweet a claim by a fictitious anti-Trump protestor that Hillary Clinton’s people paid him $3,500 to cause trouble at a Trump rally. Then, in the Post interview he proceeds to destroy what tiny bit of credibility a confessed professional hoaxer might have by making this statement:
Just ’cause his supporters were under the belief that people were getting paid to protest at their rallies, and that’s just insane. I’ve gone to Trump protests — trust me, no one needs to get paid to protest Trump. I just wanted to make fun of that insane belief, but it took off. They actually believed it.
He has the audacity to make such a statement—“trust me,” the professional liar—after it has come to light that one Zulema Rodriguez was indeed paid $1,610 to be disruptive at a Trump rally, as conceded by Politifact, well before the liar Horner gave his interview to The Post.
With this example we can see why almost any phony negative stories about Hillary Clinton that anyone might put out might be readily believed and circulated. It’s the verisimilitude of them. The Clintons have done so much that is wrong that we know about and seemed to have gotten by with it that almost anything seems plausible. Furthermore, concerning the “insane” notion that anti-Trump protestors might have been paid, the Project Veritas folks might have caught only one instance of it but I challenge anyone to watch the words and actions of Rakeem Jones, the recipient of that phony “sucker punch” at the Trump rally in Fayetteville, NC, and explain to me what, besides money, would have motivated someone like him to go to a Trump rally and stand up and shout insults when Trump began to speak.
To understand where Horner and his sort might well fit into the news picture we turn to my Seventeen Techniques for Truth Suppression. The making up of false stories is part of #4, the knocking down of straw men. The false stories that seem to originate with the opposition, and could well be very close to the truth, help discredit the true charges when the hoax accusations are exposed. Horner even tells his interviewer that that was exactly what he was up to, but it backfired on him. He is reminiscent of most of the mainstream news commentators who say regretfully that they promoted Trump by giving him so much publicity, when in fact it was nothing but negative publicity that they gave him.
Who is Paul Horner in reality? Is there any reason that we should take his word for anything? Why do all the mainstream news media people, starting with The Washington Post, treat the man with such credulity, and why should anyone who would even pretend to be a scholar display his nutty opinion so prominently and with such little skepticism? As a final question, how does he get by with using the web address of abcnews.com.co, making money on it he tells us, and flaunt an official looking web site without being sued by the Disney-ABC Television group?
“Fake News” Shot Down, Hidden
As narrowly as they define “fake news” the authors probably do no harm to themselves in using Snopes.com along with Politifact to determine what fits for the purpose of their own original research. When it comes to any matter of political controversy, anyone with any serious political Net sense knows that Snopes is something of a joke, a bigger joke even than the authors should have known at the time they wrote their paper, what with the revelations of their moral depravity that has come to light.
When the authors conducted their own post-election survey of the influence of fake news stories they had the good sense to include what they call, without the use of quotation marks, “placebo” questions. With the use of this little device, the authors were able to shoot down the feverish speculation about the influence of fake news on the election, but it takes a very careful reader of their article to realize it. You can find the key explanation at the top of page 12:
However, figure 4 also shows that Placebo fake news articles, which never actually circulated, are approximately equally likely to be recalled and believed as the Fake news articles that did actually circulate. This clearly shows that there is a meaningful rate of false recall. If this false recall rate is similar for Fake and Placebo articles, this suggests that the raw responses significantly overstate the circulation of Fake news articles, and that the true circulation of Fake news articles was quite low.
Quite low, indeed! Eyeballing Figure 4 we see that something on the order of 7% of their respondents said they had heard and believed actual fake news headlines that had circulated, but a slightly higher percentage reported hearing and believing fake news headlines that the authors simply made up. That means that it is possible that virtually all of what little fake news that people thought they heard they didn’t really hear, let alone were so impressed by or influenced by that they passed it on to other people.
These findings, I don’t think I need to tell you, are wholly inconsistent with the BuzzFeed story implying that fake news was getting more attention than real news on Facebook before the election.
The professors’ article, to say the least, is not easy for the layman to understand, so one would hope that this one revealing discovery in the authors’ original research would figure prominently in the conclusion. Instead, what we get there is primarily a blast of gratuitous, irresponsible, and very unscholarly innuendo, with only the last paragraph hinting obliquely, with epic fake precision, at the fact that the effect of false news stories on the election was trivial. Here it is in its entirety:
As a concluding note, we observe that rumors, conspiracy theories, and other cousins of fake news are not new to the social media era. Figure 7 considers 14 conspiracy theories with political implications that have circulated over the past half-century. Using polling data compiled by the American Enterprise Institute (2013), we plot the share of people who believed each statement is true, from polls conducted in the listed year. These conspiracy theories are slightly different than most of the fake news we study, in the sense that many fake news articles can be traced back to a single person who invented the article without any facts to back it up, whereas some conspiracy theories could in principle be true and often have no unique origin. Notwithstanding, they are an interesting historical benchmark.
For example, during the 2016 campaign, Donald Trump and various online media outlets reopened speculation that Bill Clinton aide Vince Foster had been murdered, whereas the five investigations into his death had concluded that it was a suicide (Kessler 2016). Four official investigations were completed between 1993 and 1995, all of which concluded that his death was a suicide (Kessler 2016). In 1997, independent counsel Kenneth Starr released a fifth report on the matter, concluding that “In sum, based on all of the available evidence, which is considerable, the [Office of Independent Counsel] agrees with the conclusion reached by every official entity that has examined the issue: Mr. Foster committed suicide” (Kessler 2016). Figure 7 illustrates how divergent conclusions on factual issues predates the social media era: as of 1995, 20 percent of Americans reported believing that Foster had been murdered.
In summary, our data suggest that social media were not the most important source of election news, and even the most widely circulated fake news stories were seen by only a small fraction of Americans. For fake news to have changed the outcome of the election, a single fake news story would need to have convinced about 0.7 percent of Clinton voters and non-voters who saw it to shift their votes to Trump, a persuasion rate equivalent to seeing 36 television campaign ads.
How about that? The reader is hoping for something that might at least begin to disentangle the foregoing thicket of exposition, but what he is presented with, instead, is the spectacle of the authors wandering off into an entirely new topic, one in which they are clearly as lost as Hansel and Gretel in the forest. They might as well have brought in Igbo storytelling or Polynesian navigation for all the knowledge they show of this new subject and for all the relevance it has. (See what you can get by with when you don’t have editors.) It is so bad, so unhelpful, so, yes, puerile, that one gets the impression that it could only have been the product of academically licensed economists, and I say that as one of them. I was going to include “modern” in the adjectives, but then I thought of H.L Mencken’s great essay, “The Dismal Science,” written almost a century ago:
One of [my poisons], following hard after theology, is political economy. What! Political economy, that dismal science? Well, why not? Its dismalness is largely a delusion, due to the fact that its chief ornaments, at least in our own day, are university professors. The professor must be an obscurantist or he is nothing; he has a special and unmatchable talent for dullness; his central aim is not to expose the truth clearly, but to exhibit his profundity, his esotericity—in brief, to stagger sophomores and other professors. The notion that German is a gnarled and unintelligible language arises out of the circumstance that it is so much written by professors….
In wandering off into the topic of the Foster death, one in which they are so conspicuously ignorant that they think that a May 25, 2016, article written by Glenn Kessler of The Washington Post, of all publications, is some kind of definitive, convincing source, the authors might have revealed a great deal more than they intended to. One of the most disturbing things, to me, about this whole phony “fake news” mantra and how it originated is that it conveys the impression that one has to make up things to make Bill and Hillary Clinton look bad. The fact of the matter is that all it takes is simply to report only a little bit of the truth, and the Foster death presents probably the best example one can find. Almost as soon as Kenneth Starr’s team produced the results of its long, unnecessarily drawn-out inquiry, The Washington Post posted it on its web site. What it did not say for the longest time was that it left off the best, most revealing part, that it was not the entire report, though the newspaper represented it as such. At some point, The Post added, no doubt because of what independent critics like me had written, the following bland disclaimer: “This file does not contain the report's footnotes or appendix.”
But why would The Post not include the appendix, one might ask. They would not include it for the same reason that they and the rest of America’s mainstream news media did not even tell you of its existence when the report was released and still have not reported it as news. And it is very, very big news, indeed! In Part 3 of my “America’s Dreyfus Affair,” I called it “The Great Suppression of ’97. The appendix contains the letter and exhibits by John H. Clarke, the lawyer for the key witness in the case, Patrick Knowlton, that completely destroys the government’s case for suicide. Almost as important, the letter was included in the report over Starr’s strenuous written objections. Starr made a very good legal case not to accept Clarke’s motion to include his letter, and the lead judge of the three-judge panel that appointed Starr, David Sentelle, was initially inclined to go along with Starr. He and the third judge, Peter Fay, though, were persuaded by Judge John Butzner, primarily with these words by Butzner, I believe: "I suspect if we deny the motion we will be charged as conspirators in the cover-up."
The good professors Allcott and Gentzkow, you can be quite certain, don’t know the first thing about any of this because you can be almost as sure that they don’t really know anything about the Foster case, itself. They couldn’t know it and still cite, with any kind of honesty, a Washington Post article as though it were a reliable authority. Even what The Post says about the four official investigations before the Starr report is not true, as I point out in my 1998 article, “The Post’s Sloppy Cover-up.” The Senate Whitewater Committee, whose report is included in the list, was not tasked to examine the question of whether Foster committed suicide or not. As Clarke, Knowlton, and researcher Hugh Turley maintain with ample evidence to support them, the Foster death “investigation” has really been the work of one FBI team and hardly anyone else all the way through from the very first day through the completion of the Starr Report. Anyone reading only a small part of what I have written on the Foster case will quickly conclude that if the FBI was primarily responsible for the Foster murder cover-up, The Washington Post is running a pretty close second.
As profoundly ignorant of the Foster case as they indisputably are, though, these two “scholars” recklessly characterize the notion that Foster was murdered as a “cousin” to the sort of rascality that a reprobate like Paul Horner puts out on the Internet for profit (according to Horner’s unreliable word). You can almost see the sneer on their faces as they write, “…as of 1995, 20 percent of Americans reported believing that Foster had been murdered.”
The Trouble with Economists
“Where do these guys get off with such quack scholarship?” you might be asking yourself at this point. With their background and training, unfortunately, they probably have hardly any notion of what their scholarly shortcomings are. Digging into messy source documents, evaluating the quality of sources, and doing tons of reading is not what they do. That’s for people in the “softer” fields, like history. “If you can’t count it, it doesn’t count,” is the glib expression I heard a number of times in economics graduate school. My feeling at the time, and still is, was that when it comes to endlessly complicated human affairs, if you can count it, it doesn’t count. That is to say, most of the really important and interesting things in human endeavor don’t lend themselves very well to quantification.
But one gets ahead in Allcott and Gentzkow’s chosen field by wielding the tools that they have been trained to use in an impressive fashion (staggering sophomores and other professors) and they are determined to use them, even if they’re not very good at uncovering important truth. It’s where their comparative advantage lies.
This pair’s manifest scholarly failings are of an entirely different stripe from that of Matthew McNiece, the Texas professor who got a history Ph.D. at Texas Christian University partly on the strength of a dissertation that attacked my original research on the death of Secretary of Defense James Forrestal in a spectacularly bumbling way. They are not dim bulbs. It was in economics graduate school, in fact, that I encountered the sort of intellectual snobbery in which about the highest praise one could heap on a person was to say that he or she was “bright.” To be bright was to have the mental facility to pick up quickly upon the concepts and to master the analytical techniques that were required of us.
The attitude engendered by the experience is one of a sort of unwarranted intellectual smugness. And the sort of brightness we are talking about is more on the order of an LED bulb or a mercury vapor lamp, as opposed to natural sunlight. It reflects only a small part of the color spectrum and ends up shedding very little real light on anything.
I also recall being impressed in graduate school by an article by the British development economist, Dudley Seers, on the teaching of economics. It was one that a fellow graduate student had discovered, not one assigned to me. I’m pretty sure that it’s the first listed on his Wikipedia page, “Twenty Leading Questions on the Teaching of Economics in The Teaching of Development Economics.” He contrasted the teaching of economists and the teaching of medical doctors. The closer the latter gets to his degree and actually practicing medicine, the more practical his education becomes. Generally, the higher one goes in his education in economics and the closer he gets to obtaining a Ph.D., the more theoretical his education gets. Seers suggested that before anyone is licensed to practice economics he should have to spend at least a year or more working for an organization involved in the actual collection of data so that he can have a better appreciation of the value of the numbers he will be dealing with.
So, as it is, these facile wielders of the economist’s tools turn out not to be as wise to reality as one might expect them to be. Rather, the average person, accustomed to employing a broader range of his intellectual faculties, and certainly a more widely read and educated person, might find them to be surprising simpletons. Such types are very different from dim bulbs, but, alas, no more effectual in the pursuit of truth, as we see from this most recent work by Harvard educated Professors Hunt Allcott and Matthew Gentzkow.
February 17, 2017