Andreas M. Kraxberger
Should we take political satire seriously? What at first glance sounds like a frivolous question no longer seems entirely far-fetched against the backdrop of increasing news avoidance, a widespread loss of trust in traditional media, and the growing influence of algorithms in the dissemination of information. The weekly satirical programme “Last Week Tonight”, which is particularly well-known in the US, is often regarded by academics, commentators, staff and viewers as a form of opinion journalism and therefore plays a prominent role in the political discourse there. In an article recently published in the journal *Journalism*, I analysed whether and in what form this satirical-journalistic approach is also reflected in the German and Austrian interpretations of the format – using AI-supported content analysis. How does satire adapt to the changing media and social contexts of its target audience? What freedoms do satirical formats have, and how are these utilised? And finally: if AI finds its way into the social sciences, what changes might this bring about?
Paper Thoughts: Why Political Satire Should be Considered Journalism and How LLMs can Actually Support Social Science Research
A few weeks ago, my paper "Same Same But Different: Communicative Functions and Journalistic Role Performance in Weekly News Satire Across Cultural Contexts" was published in Journalism. With a bit of necessary distance, I finally took the time to collect my thoughts about this paper, and three aspects stand out in particular - one about international political satire, one about the changing nature of journalism, and one about how LLMs might be revolutionizing research.
1. Political humour is no longer just an American export but a genuinely international phenomenon taking on a life of its own outside of the U.S.
We all (should) know John Oliver and his brainchild Last Week Tonight. It’s won truly so many Emmys, has been studied by academics and regularly drives both professional and cultural conversations. His German and Austrian counterparts are Jan Böhmermann and Peter Klien with their respective programmes ZDF Magazin Royale (Germany), and Gute Nacht Österreich (Austria). Though they are clearly inspired by one-another, they are not doing the same thing. All three programmes set out to talk about a topic most people have already scrolled past, or never come across in the first place, spend 20-30 minutes on explaining it, poking fun at it, and ultimately make you care. If and how exactly the German-language approach to weekly political satire differs from its U.S. rolemodel has remained unexplored so far. My study compared all three programmes, coding over 6,000 sentences across 30 segments for both their communicative strategy and their journalistic role performance.
What I found is that while all three share a common genre DNA - they all mix humour, information, and opinion, and none of them attempt to justify or support their respective governments (something called “Loyalist journalism” which you see a lot with access journalism) - they are approaching their comedic-journalistic role in meaningfully different ways. American satire leans hard into civic mobilization, actively telling audiences what they can do and why it matters. On a recent episode, Oliver called himself out for always ending with a “What can YOU do about it?”-segment. In contrast, German satire concentrates its energy around a central core of information (often sourced from cooperations with investigative journalists or citizen advocate groups) and subjects it to sustained humorous and critical commentary. Austrian satire, operating inside a heavily centralized public broadcaster with a fraction of the production resources of the other two programmes, delivers comparatively more straight information, often recycled news footage from its public broadcaster home, with less tendency towards explicitly voiced opinion.
In other words: these programmes go beyond adapting their jokes to local audiences, instead altering their entire communicative strategy. The balance of information, humour, critique, and civic call-to-action is adapted to fit the expectations, political cultures, and institutional constraints of the media systems they inhabit. Political satire, it turns out, is more than an American cultural export with local flavouring. I was reminded of my paper’s findings when I saw (clips on YouTube of, I simply refuse to pay for SkyTV) the inaugural episode of Saturday Night Live UK. While I am sure they will find their footing in time, the bones of what could be an exceptionally funny experiment in translating a comedic institution’s traditions, oddities and possibilities to a significantly different context were there. Their success will hinge on the performers, writers and producers finding their audience’s preferred balance of humour, information and opinion and doubling down on it, regardless of how far away from the American original it gets them.
2. The absence of journalistic constraints is not the same as the absence of journalistic standards. Fight me.
I am far from the only communication scholar, let alone media consumer that treats political satire as a trusted source of information and opinion. More specifically, I view political satire as a locally shaped journalistic practice: a form of opinionated journalism that happens to be funny. However, there is a persistent and slightly condescending assumption that political satire occupies a lower tier of the information ecosystem. Often derided as “Infotainment” or “mere” entertainment dressed up as news, the format’s discussions of serious and underreported topics as well as its rigorous sourcing and commitment to informational accuracy – in short, the format’s journalistic role – gets brushed aside and minimized. Using my findings, I want to push back against this.
Across all three programmes, the role of “Loyalist Journalist” – which comprises of such behaviours as endorsing governmental leaders, signalling reverence for authority or calling for national unity – was virtually absent (to be expected, given the inherently critical nature of satire as an art form). Meanwhile, performance of watchdog journalism, i.e. holding political actors, private institutions, governments, and media organizations to account, was the dominant journalistic posture, especially in the U.S. and German programmes. In the vocabulary of journalistic role performance theory, the satirists are functioning as adversarial, advocacy-driven reporters who have simply traded the convention of (often inaccurate) balance at all costs for something more honest: clearly voiced, fact-based and well-researched opinion delivered under the cover of comedic licence and with the necessary emotional depth.
Now, I am not arguing that political satire can replace traditional journalism – nor should it – but it has become a genuine and impactful contributor to the (political) discourse around certain topics, engaging in attribute agenda setting: it doesn’t tell people what to think, but rather how to think about a certain topic. This matters because we are living through a period of deep distrust in legacy news media, rising news avoidance, and a growing appetite for journalism that is willing to have a point of view, as evidenced by e.g. the rise of podcasts and partisan news networks. In that context, we need to view weekly news satire as more than a curiosity at the margins of political communication research, more than mere entertainment, more than cheap punchlines and easy laughs. I would argue, for many, it has become increasingly central to how they engage with political information – a valuable addition to, if not outright substitute for their news media diet – and understanding how it works across different cultural contexts is therefore not just academically interesting, it is urgent. Especially since, though this may be a topic for a future article, there are things that professional communicators can (and should) learn from the way political satire packages its often difficult topics.
3. LLMs, used responsibly, may be good for social science?!
The methodological contribution of this paper is one I am genuinely excited about, because I think it points toward something important for the kind of research I have dedicated my life to for the past ten years.
Coding 6,356 sentences across three languages for both communicative function and journalistic role performance would, using traditional methods, require either significant funding to hire and train multiple coders, or an uncomfortable compromise on either sample size or inter-coder reliability. I used to be one of these coders, in fact, my first job at the University of Innsbruck was coding the tonality of press releases for a large dataset. I did not enjoy that job very much, I did not learn a lot in that job, and it really wasn’t paid well. I considered it a necessary evil on the “career ladder” inside the university, as a foot in the door for later employment. In that sense, it worked out for me. The project supervisor became my supervisor and later hired me for a larger research project they received funding for.
However, that outcome is the exception. Most untrained coders leave academia and take with them only bad memories of needlessly repetitive tasks and awful pay. Meanwhile principal researchers need to beg, steal and borrow to scrape together enough funding to throw at these untrained coders, in the hopes that they produce reliable results – something not guaranteed by any stretch of the imagination. I am a pre-doc researcher with a small Early Career grant, I, like so many others in my position, do not have the means to even apply for the kind of funding I would need to employ a sufficient number of coders. Under normal circumstances, I simply would not be able to conduct research like this in a reliable and ethical way.
Instead of giving up and moving on, though, I used the tools that were at my disposal to conduct a Large Language Model-Assisted Content Analysis (LACA). In practice, this meant customizing the, at the time, frontier model (GPT-4o) to act as a trained coder and benchmarking its performance against data coded previously by a human expert using Cohen's Kappa.
The results: a Kappa of 0.81 for communicative functions and 0.85 for journalistic role performance. To validate the performance of the LLM-coder, I also tested three other LLMs – Claude 3.7, DeepSeek, Gemini 2.0 – and all performed at substantial to moderate reliability, though none as impressive as GPT-4o. Just like I would have done with a human coder, I made sure to prompt the LLM for coding justifications so I could supervise the output. Also, 50 items per prompt seemed to be the limit before hallucinations started happening, so I made sure to observe that. Overall, especially compared to my experience coding similar data before, the LLM allowed me to metaphorically speed through my sample data and finish the coding in no-time.
What this means in practice goes beyond time-saving though. First, because the LLM's instructions are fully documentable and the model itself can be made publicly accessible, the coding process becomes replicable in a way that human coding – which depends on the training, attention, and idiosyncratic judgment of individuals – simply is not (having the LLM code the same sample twice yielded an ever higher intra-coder reliability than what one would expect when giving the same task to a human coder). But perhaps more importantly, this approach substantially lowers the resource barrier to quantitative content analysis. As I said, you used to need a large grant and a team of research assistants to conduct rigorous, large-scale coding. I know, because I used to be one of those research assistants just before ChatGPT was released. The ability for anyone familiar with the general research process to become a supervisor to easy-to-use, resource-sparing and incredibly customizable artificial coders is a meaningful step toward a more egalitarian research landscape – one where the quality of the question and the creativity of the researcher matter more than the size of the budget.
The reliability crisis and funding scarcity in social science are both real issues and not going away any time soon. Replicability, transparency, and consistency in measurement are at the heart of it. I am not claiming LLMs are a complete solution – they require careful validation, human oversight, and honest reporting of their limitations, which I have done in the paper stemming from my analysis. But used responsibly, they represent one of the more promising methodological developments I have seen in the time I have been doing this work.
Über die Autorin
Andreas M. Kraxberger, MA, is a University Assistant at the Department of Media, Society and Communication. His research and teaching is focussed on political communication, science communication, the journalistic function of political satire, as well as innovative methods in the social sciences. He is the Student Representative at the Doctoral College "Politics, Power and Language".
Zitieren
Kraxberger, A. M. (2026). Same same but different: Communicative functions and journalistic role performance in weekly news satire across cultural contexts. Journalism, 1-23. https://doi.org/10.1177/14648849251413636
This article gives the views of the author(s), and not necessarily the position of the Department of Political Science.
The text is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).