News Aggregators and the Overton Window

Having spent over half a decade in university and across various positions focused on policy, I’ve come across many theories that try to taxonomise the ideas that comprise the political “left” and political “right”. Most are familiar with the wide variety of political compasses. But there are a handful of others that have gained some degree of mainstream appeal due to their accessibility and simple premises:

Horseshoe theory: probably the one I come across the most. This theory states that the so-called ‘far left’ and ‘far right’ of a political ecosystem actually loop around and become nearly synonymous with one another. I often see the regimes of Stalin in the Soviet Union (the ‘far left’ example) and Hitler in WWII-era Germany (the ‘far right’ example) invoked to substantiate this theory. Proponents cite similar tactics used to enforce authoritarianism, like the jailing/expelling of academics and suppression of religion, as evidence that going far enough in either direction eventually loops one back around to the other side.
Fishhook theory: is, as far as I understand, a response to or satire of the horseshoe. It argues that centrist politics and policies align more with the ‘far right’ than the ‘far left’, either through rejecting leftist policies or by ceding ground to the right by allowing right-coded policies to encroach upon the centre in the first place.

I’m not one to argue for or against these theories one way or another; to me they’re closer to thought experiments than to true representations of a state’s policy environment.

The Overton Window

The Overton Window: proposed in the 90s by Joseph P. Overton, a prior executive for the Mackinac Center for Public Policy. Per the Center’s own attribution, the idea was introduced less as an academic contribution than as a tool for strategic planning. It argues that there is an ‘acceptable window’ within which the policies a state (or other body) can realistically adopt must fall. The farther you travel outside the bounds of the window, the more unacceptable or radical a proposed policy becomes. This is important, Overton argued, in knowing what a legislature can and can’t realistically pass. Over time, the window can contract, expand, or shift to either side based on what constituents view as acceptable.

It doesn’t serve as an all-encompassing view of a policy environment. Indeed, one of its main critiques is that it simplifies the complex and ever-evolving nature of the policy process. Another is that a single window is insufficient, and that a ‘collection of windows’ is more accurate to explain the views and actions of different sections of a state’s policy environment. However, I believe it’s accurate in its foundational belief that some policies may lose merit based on the perception thereof rather than the substance of their measures.

News aggregators

The next concept to introduce is news aggregators. Simply enough, they’re services that compile a large collection of news stories from various sources and present them in a feed. They’re not a new concept — Apple News and Google News are aggregators, and therefore nearly every modern phone comes with one preinstalled. Newer aggregators include, among others, the main subject of this article.

What differentiates this newer cut of aggregators from those of old is that they take an active role in the classification of stories and media providers. Ground News has gained popularity for its “Media Bias Bar”, a visualisation of how many news sources from across the aisle are reporting on a particular story. It relies on ratings from independent media bias groups like Ad Fontes, AllSides, and Media Bias Fact Check, who classify media providers by “left”, “lean left”, “centre”, “lean right” and “right”. Ground News’ bar groups “left” and “lean left”, and accordingly so for “lean right” and “right”. This sacrifices a more nuanced view with five stacked bars for a more readable presentation with three stacked bars. Ground News has a detailed page on their methodology on their site.

Before diving into the exploration of how exactly a service like Ground News may affect the Overton Window, I’ll first present my own biases. I find Ground News genuinely useful and rely on it. A core belief of mine is that those wishing to limit their personal biases while consuming media must expose themselves often to sources from across the aisle, even those they find unacceptable, repulsive, or radical. This being said, the factor of credibility must be weighed. This is not to say that established news sources closer to the political centre are infallible, nor that those to the extremes are less credible by construction. However, trust is built over time, and those with more extreme views may tend to rely on more opinionated sources that lie outside what is or has been studied in an academic setting. It is also generally true that, despite the ability to separate fact and bias, sources biased more heavily one way or another may become dishonest insofar as the omission of contrary facts is concerned.

This all being said, I’ve become more critical of Ground News over time. Their marketing strategy, while seemingly effective, relies heavily on sponsoring YouTube creators for mid-roll ads. Many are likely familiar with Honey and BetterHelp, two other companies which relied on this strategy. A few years ago, Honey was exposed for short-changing those they sponsored, while BetterHelp became wrapped up in a lawsuit alleging improper release of user health data. While these companies’ actions should by no means bias one against Ground News, they’ve made me a bit more skeptical. It’s also worth citing the artificial intelligence tools that make aggregating such a large number of daily news stories possible, though I’ll leave discussion of this for another time in the interest of staying on topic.

Resizing the window

I approach this section by once again stressing that acceptance of and research on the Overton Window is very split, and I encourage everyone to make their own assessments per an individual reading. I’ll use the Window as a ‘framing device’ (pun intended) and proceed with it in its simpler form.

As mentioned by Johannes Kleske in his “primer on the Overton Window”, media framing has an influence on what is considered acceptable in the way of policy. How parties are presented in an article, the language used to describe different facets of an issue, and whether an issue is actually covered in the first place all weigh on the frame in their own right. At first glance, this shouldn’t be an issue for a news aggregator. As long as their method faithfully cites (or distills) article headlines and descriptions, they should be a non-factor, right? In theory, yes, in practice, it’s more complicated.

Ground News also provides source “factuality” ratings, which are drawn from Ad Fontes and Media Bias Fact Check. At first, I was surprised at this. The quick cards they post on various social media sites don’t include this, and I didn’t remember seeing it in their app. There’s a good reason for this: it’s paywalled, as is their indexing of source ownership influence. This is a decision that I’m not criticising for its margin-driven nature. However, I believe the unintended effect is clear.

Per Ad Fontes’ own charting of “media bias” on “news value & reliability”, the distribution of a mix of legacy news sites and more alternative sources (like talk shows and YouTube channels) is fairly symmetric about the centre; it resembles $y = -x^2$. Where I believe a shifting effect may enter is the presentation of sources from across the aisle being comparable in terms of factuality and reliability. Given the absence of these attributions from social media content and the website (absent a paid plan), it’s not a stretch to hypothesise that the everyday user may consider highlighted sources to be of equal reliability. As a media source in and of itself, Ground News has formidable reach — 1.2m followers on Instagram, 67.6k on X/Twitter and 120k on Facebook.

The potential effect is subtle. Then again, so are most other media-induced effects on agenda setting. Were there a clear imbalance in mean reliability of sources from either side, presenting a veil of equality would likely work in the favour of the side of the aisle with comparatively less of said reliability. An extension that I may make to this article will include an analysis of said ratings from Ad Fontes and Media Bias Fact Check, looking for a statistically significant difference in mean reliability across 7-way, 5-way and 3-way spectrums.

Final notes

This is not meant to be a take down of news aggregators or Ground News itself. As mentioned, these are tools I use and rely on often. Without proper recognition of their limitations, however, I see it as entirely possible that a false balance may emerge: with the reliability axis hidden behind the paywall, the bias bar flattens a two-dimensional space into one, and sources separated by a wide margin in factuality appear as neighbours. Questionable sources under this veil of equality gain traction faster than they otherwise would, carrying fringe ideas into the window of acceptable policy.