Mainstream Media Claims Its Own Coverage Is Worse for Biden Than for Trump
In a truly stunning piece, a columnist in the Washington Post has tried to claim that “statistics” show a shocking fact about media coverage. The argument is that coverage of president Biden in 2021 has at times been more negative than coverage of Trump was four years ago.
The article even comes with an impressive-looking infographic:
The claim in the piece should indeed startle anyone who has paid attention to politics news over the last four years, because it really isn’t true.
In fact, the thing I find most concerning about this article — otherwise, I wouldn’t bother responding to it, even though its warm press reception seems to contradict its thesis — is that it really is an impressive use of shoddy and out-of-context information. Wrapped up in a shiny graphic, it’s been persuasive among the pundit class precisely because of the religion of data worship endemic to that class.
To start with, we should ask, how exactly does one calculate negativity? Trying to prove with a dataset that news coverage treats one politician worse than another is kind of like thinking you’ve proven with a graph that music has gotten progressively shittier since the 1960s. (That graph basically exists by the way.)
The distinction we forget is that in cases like these, the graphs don’t reveal qualitative abstractions in themselves (music quality as such, news negativity as such, and so on); such a revelation is in fact impossible. What we’re really showing with graphs is a set of quantitative values that are arranged, and in some cases generated, by means of subjective and arbitrary determinations. (Hence the scenario of racist AI.)
With that in mind, let’s look now at the claim from Dana Milbank in The Washington Post.
The big story to unpack is that Milbank hired a data firm to perform “sentiment analysis” on a dataset of 200,000-plus articles. These articles were culled from 65 news websites and run through “algorithms that give weight to certain adjectives based on their placement in the story.” Essentially, the team handed over some news to a bot, which can’t understand news but can categorize it according to simple rules.
It’s the lack of good qualitative analysis that turns this piece into cut-rate data science. For now, robots can’t interpret, so we have to. Milbank doesn’t, or at least doesn’t do it well.
For one thing, weighting adjectives and their ordinal positions doesn’t even come close to approximating linguistic meaning, but that’s the kind of messy detail that Milbank finds it best to just leave out. This becomes a problem because the number and position of adjectives in an article cannot capture a kind of negativity that’s intrinsic to subject matter. We’re not comparing Trump articles to Biden ones in some kind of neutral, politics-free context — the gray utopia of the data nerd. Instead we’re comparing coverage of Biden’s limp-handshake legislative ability to coverage insinuating that Trump was a Manchurian candidate installed in Russia-rigged election, that he was a living, breathing embodiment of falsehood itself, that he was a white supremacist fascist on the order of Hitler.
Examples that Milbank actually provides for negative coverage of Biden include headlines such as “Dems start to face the hard questions,” and “No BIF [bipartisan infrastructure] bump for Biden.” But how does one even begin to compare these stories to ones such as “Here’s what we learned about Team Trump’s ties to Russian interests,” published in the Post in 2017, which contain web-like graphics of “connections” between Trump staff and very Russian-sounding individuals — basically ripped straight from a TV police procedural, and not even a very good one — and an absolutely sprawling list of data points that all correspond to individual articles, breathlessly detailing moments in a saga of unproven (and eventually falsified) collusion with the Russian state?
How do we compare a headline saying “Biden dithers,” another Milbank citation, to coverage that directly references a supposed “mountain of evidence piling up that Donald Trump is a puppet … of Vladimir Putin?”
The New York Times actually promised its readership to use the word “lie” in an unprecedented manner, specifically for political stories connected to Donald Trump. They collected his lies (not fibs or falsehoods!) in exhaustive lists.
The cultural impact of this coverage was profound. It essentially normalized a form of liberal Russophobia that’s still playing out. Pundit Keith Olbermann recently described living his entire life “with the stain of Russian heritage in my family.” And U.S. appetite for war with Cold War foes has steadily increased since 2016.
But even with all these things considered, there’s actually an even dumber element to Milbank’s coverage comparison and the argument it entails.
Another place where the term “gray utopia” fails to apply is the media landscape itself. To produce an “average sentiment” measure for the media, the Milbank-style analysis has to take in data from a range of overtly partisan and ideological outlets. Milbank makes no effort to exclude even fringe pages such as Breitbart News. So although he has made a plea to all media, ostensibly, to take a more positive stance toward the Biden administration, in reality there are many conservative outlets which simply won’t do so, because they are conservative, and many leftist outlets that won’t because Biden is conservative (to them).
The only option this leaves, in terms of pulling Biden’s coverage back to Milbank’s idea of neutral, is for the more balanced mainstream outlets to modulate their coverage into positive territory — simply to counterbalance outlets that oppose Biden’s ideology. This is more than a simple request to abandon the idea of ideological neutrality, which might admittedly be flawed: it’s an ask for outlets to make all reportorial decisions with the express purpose of glamorizing an administration, regardless of its activity. In fact, Milbank almost explicitly tries to frame his plea for Biden hagiography as post-ideological:
We need a skeptical, independent press. But how about being partisans for democracy? The country is in an existential struggle between self-governance and an authoritarian alternative. …
Too many journalists are caught in a mindless neutrality between democracy and its saboteurs, between fact and fiction. It’s time to take a stand.
This reasoning runs right into the same pitfall I pointed out in my piece on Kamala Harris some months ago. Milbank wants the media to become Biden totalitarians, to agree with Biden publicly even when their hearts disagree. But what should happen when Biden’s actions actually leave us vulnerable to what Milbank fears most, to a Trumpian rise to power from the right?
People need to know Biden is struggling in order to do anything about it. But Milbank cares more about the appearance of fighting fascism than about actually stopping the people he believes to be fascist.