Most criticisms of the media – especially those popularized during the “Enemy of the People” phase of the Donald Trump presidency – have focused on assertions of its bias. Bias certainly exists, but we’re not here to adjudicate that. We think identifying bias is very hard to do so systematically. It is even harder to do without injecting our own biases into the creation of whatever system we might use. Fortunately, the concept of fiat news we so often write about isn’t really about bias – or at least, it’s about far more than bias. Fiat news is about the press telling you how to think about issues. Fiat news is about the presentation of opinions as facts, regardless of whether they consistently favor one group or another. If you want a bit more of a primer, including why we call this fiat news, the original piece Ben wrote in 2017 is located here.
We think there are some ways to measure this, so we’re going to try. And we’re going to do it in the open. Let me introduce you to the Fiat News Index.
I’ve selected 20 of the largest and most prominent US-based news and commentary organizations. Using the tools and database from our friends at Quid, the Index measures the proportion of articles from a media outlet which use one of a range of words or expressions I selected. These words and expressions fall roughly into three categories: words that convey a causal link between two statements (Causal Expressions), words that seek to communicate the Common Knowledge element of a narrative (Common Knowledge Expressions), and words that communicate explicit value judgements (Value Expressions). These concepts will be familiar to readers of the recent In Brief, The Tells of Fiat News.
The basic idea behind this framework is that writers, when using Causal Expressions, are communicating how you should perceive the relationships between facts and other facts, or between facts and certain conclusions and analysis. This conflation is a common way to present a judgment or opinion as objective fact. It is a writer coaching you on the logical path they wish you to follow. Sometimes that is innocuous, because sometimes the relationship between two ideas, two facts or two statements really is incontrovertible. Often it is not. When using Common Knowledge Expressions, the writer is encouraging you to think less critically about an assertion or argument. It is, after all, obvious to everyone else. Value Expressions are more straightforward and easily understood. They also look a bit more like an analysis of bias, although these words may just as easily be used to tell you how to think about what is good and what is bad without any element of structural favoring of one point of view.
I suspect you could come up with many more such expressions. The danger to adding too many is that you end up with Type 1 errors, where we catch more innocent uses. News articles include quotes from subjects that include these terms, for example. And it’s not as if every use of these words in a news article can or ought to be avoided. In addition, the preferred style of different venues will be more or less likely to lean on these expressions. For this reason, the absolute levels are much less instructive than the relative levels. For me, I understand this index to mean, “If I open the pages of this publication, how much more likely is it than in another publication that I will read a story that is telling me how to think?”
Here is the Fiat News Index for the last 12 months ended November 10:
A few words. First, the Index includes four media companies that are not news outlets. This is by design. The unit of the Fiat News Index is the Vox, not because there’s necessarily anything bad or dishonest about what Vox does, but because Vox’s stated mission is to explain the news. Approximately 91% of its articles in the last year included one of these explainer words. Nothing necessarily wrong with that in a commentary or analysis publication (like Vox, The Atlantic, National Review or The New Yorker), but potentially a matter of concern when it takes place in a news outlet. Each other source is scaled to express how many Voxes of explaining their articles have engaged in over the last year.
The poles are instructive. On the one hand, we have Vox, and on the other, Reuters. In between, there is a meaningful range. While I don’t have the data to give Reuters a completely clean bill of health, for our purposes I think it is useful to think of their level as a baseline of the innocuous usage of these terms. From there, Voxes will rise with the (1) use of these terms to explain topics in news articles and (2) the relative proportion of opinion and commentary to pure news coverage. The first is our primary focus, but the second isn’t irrelevant, and we don’t consider it a false positive. You should read this as an attempt to proxy the following question: “If I open this publication, how likely is it that I will be told how to think about world events instead of being given simple information about world events?”
It is possible, as noted previously, that some publications have adopted or prefer a simpler style that is less likely to use these terms in coverage. If you want to see that as a potential flaw, go ahead. I don’t want to tell you how to think about it. But I will anyway. I believe that the conscious choice by some venues to use less complicated language and sentence construction is a similarly conscious choice to keep reporting confined to explicit facts. As with so very many things, we might reasonably consider the New York Post the exception which proves the rule.
You will note two very glaring omissions: two massive publications, Fox News and the Wall Street Journal. This is a data issue on our end. We are working to resolve it. But a point of emphasis that is worth highlighting: this Index is not a measure of political bias. This Index is not intended to be a measure of political bias. Transparently, I still rather strongly expect to find a higher than average Index for Fox News once we can update our database. But in the meantime, we are hopeful that this proves useful to you in making up your own damn mind about what the news means. Doing so will play a critical role in our conscious efforts to resist both narratives and the exploitable memes which infect each of our minds.
We will continue to update this over time. We plan to research topic-related tendencies and add additional venues.
Author’s Note: Since publishing, we have resolved the sourcing issue noted above in part and added three more sources to the chart below: FoxNews.com, Breitbart and Huffington Post. We agree with many readers that WSJ and the Financial Times would be worthwhile additions but still cannot resolve paywall issues there.
Thank you Rusty, this looks like it will help me manage through this time and age as I’m quite unhappy with fiat news.
Please keep updating
As always a lot to study, absorb and digest in several rereads before I really “own” what Rusty wrote. But one thought jumped out at me from a simpler time when I was “taught” how to read a newspaper in middle school in a “social studies” class. We were told that a newspaper’s news pages reports facts – the who, what, when, where, why – without bias; whereas, the editorial page gives an opinion / a view and that papers keep a strict divide – an impenetrable Chinese Wall – between the two.
Okay, stop laughing. The thing I had to learn on my own was that the news pages had plenty of commentary, now not only in the “straight” news articles, but also in all the news “analysis” and similar articles that sit right in the middle of newspapers’ news pages (the NYT loves to do this as does the WSJ, for example, with its “Capital Journal” pieces). Conversely, occasionally, you’ll learn facts – straight news – in an editorial that somehow (hmm) have been missed by the five news articles you read in three different papers on the same subject.
Rusty, as he and Ben always do, provides a deeper dive and more structured analysis of this issue while also offering up a rigorous process to help you navigate all of this without being “nudged” into this or that narrative or view. From the first day I started reading newspapers, I recognized that something was off from the idealized view that I had been taught. It’s a challenge everyday for every active newspaper reader to disaggregate opinion from news / the nudge from the facts – glad to have some tools and a process to help. Thank you Rusty.
Test
This is great! Thank you for doing this. I grew up in Asia and one time I was visiting Taiwan and my friend was telling me how the cab drivers there (at last back then) were the most knowledgeable and passionate about politics and news. It stuck with me because in retrospect their jobs allow them the time to do their job while taking in and processing / judging /perhaps critically thinking through information, back then, transmitted from the radio; a critical, informed mind requires the space, energy and time to do this type of triage, no? Is it not highly energy intensive for individuals to assess the fiat nature of the news they consume? Don’t most individuals’ jobs interfere with this process and the more general role of citizen from a time/energy/space perspective?
I’ve thought about this as well. I’m lucky in that my job allows me a lot of free time in front of a PC screen, so I’m able to read and digest content in a somewhat critical context. Can’t say this about most of my social circle though. It’s exhausting and time-consuming to maintain clear eyes!
I’ve heard it asserted that communication is 60% body language, 30% tone and only 10% the uttered or written words. If there’s any truth to that, this type of analysis will fall short. I don’t have any suggestions for capturing the other 90% (can’t even vouch for the accuracy of the assertion that body language and tone are that significant), just think it’s important to consider.
Isidore, I don’t just think you’re right, I know you’re right. While you should decide for yourself, my personal point of view is that this is probably only useful in the aggregate. For the reasons you describe, I don’t think it’s particularly useful for any single article. But that 10% can add up across many articles to tell you something, I think.
Rusty, great topic. Sources and their methods
People should figure out what this index truly tells them. To me it is an “obviousness” index.
I strongly disagree about sweeping Reuters and BBG use to “innocent baseline”. They are powerful narrative sources with a lot of fiat smell to them – perhaps more powerful than others as ubiquity of non-loaded news makes them look neutral so they seek to pass you loaded topics as facts when your guard is down. Their use of “due to” or “following” in their highlighted news or numbers reports is a powerful missionary statement and an obvious fiat index candidate. They also have a way of loading meaning into seemingly facts by promoting things into Top page for no obvious reason, omitting true top news, highlighting news, repeating them with “updated” prefix when the update is non-critical. All those things spike my narrative barometer readings. I personally do not think there is all that much difference between any sources for amount of fiat once you mentally correct them for their typical methods and once your ears are tuned – it can be the absence of things, it can be the flow of the way facts are presented where you are invited to read between the lines and made to think it was your own conclusion. I think the index is about how obvious the attempt to create the narrative is.
Do you correct your narrative maps for source’s breadth of coverage (reuters would have every news possible whilst Vox would only have the bits they want to push)? I would assign higher weight to something appearing on WSJ or FT as an equivalent on BBG would be to see it promoted to Top News page and flash up at you several times a day with “Updated:” or being highlighted red.
Presumably you should have typical targets audience collections of sources when creating your narrative maps and weight them differently depending on whose map you are trying to see. Institutional investors map on gold will look very different from that of a prepper gold bug that has only ZH to read. For politics it would be interesting to see you supplement your all encompassing map with two polar maps weighted differently for typical source collections for those camps so we know whether they are being pushed towards centre of more radical.
I would love to see you develop a missionary index (including the news sources) – an objective way to see whose statements carry more weight and really shape the narrative – perhaps citations or name mentions combined with a lagging indicator of narrative change.
There’s a lot to chew on here, and I think most of your criticisms / ideas are fair and interesting for further examination. For example, part of the reason I kicked this analysis off in the first place was because of Reuters and Bloomberg articles that seemed very off to me. A baseline is a relative thing these days, I suppose.
I’ll give all this some thought as we continue our little research project.
now I reread my comment I have to apologise for my brain incontinence. This is just too exciting a topic
Very interesting, but somehow disturbing. The words in the list are, as your earlier stuff says, excellent hints or clues of fiatness (fiatism, fiatedness) but they don’t characterize it. Aren’t you at risk for creating an index that creates for the reader a Narrative of the type you and Ben discuss, a derivative or abstraction based on incomplete data with a particular goal in mind? Many other techniques are used to nudge opinion – story selection and rejection, loaded terms like “radical” and “progressive” and even “liberal” and “conservative”, names of persons not directly related to the story, and the like. Your index is just one corner of the vast expanse of nudge techniques in use, but it could be seen as a valid index of fiat characteristics – which in fact seems to be what you are attempting. Narrative thrives on such simplified pictures, graphs, tables.
Hints and clues are helpful, but this index or any attempt to reduce the notion of fiat news to a number or graph goes too far, methinks.
If someone was purporting to have a complete understanding of all the nature of bias and abstraction, then yes, I think I’d agree very strongly with you that reducing it to a number isn’t just foolhardy, but definitionally an abstraction itself!
But I think we’re comfortable that our readership understand that we are attempting to highlight one narrow portion, and that our objective is to be probably approximately correct. There’s zero question it incorporates our own biases (how could it not?) and that it presents an incomplete picture. Can we do more and better? Yes, and we’ll try. It is better than nothing? Well, on that, we may have to disagree! I think “yes!”
Since you have 3 types of words (Causal, Common Knowledge, and Value), it would be interesting to see the charts of the different types – maybe you already did this and it just looks the same as the aggregated chart…
The normalized “Fiat News Index” is my favorite ET thus far (not counting the personal ones)! I hope you will apply this methodology to alternative media sites as well, e.g. The Intercept, Consortium, Truthdig, Lew Rockwell, Weekly Standard, etc. etc.
What I’m sure I need going forward is not investment brilliancy but a way to sort out the cacophony of (non-financial) opinion and content not generated by the current main streamers featured above. VK