Blameocracy: Causal Rhetoric in Politics
summary
The gist
As an AI researcher, I have meticulously reviewed both provided texts to synthesize a comprehensive and accurate description of the paper "Blameocracy…
In short
The study used a language model to classify congressional tweets as blame or credit rhetoric, finding that causal rhetoric is mostly used in political disagreements. It showed that blame generates significant political returns, especially in increasing donor support and amplification (retweets), while credit has little effect on these metrics.
Key concepts
- Causal Rhetoric
- Language politicians use to link their actions to outcomes through blame or credit. The research found this style is central to political disagreement and heavily influences how political messages spread.
- Donation Returns
- Measures the impact of rhetoric on fundraising, including donation revenue and donor counts. The study found that blame rhetoric leads to consistent gains in support, while credit has no comparable aggregate effect on these financial metrics.
- Amplification
- The extent to which tweets are retweeted. Blame tweets received significantly more retweets than other rhetorical styles, suggesting that blaming others is a powerful tool for spreading political messages across the platform.
Terminology used across episodes
This episode discusses
- Blameocracy: Causal Rhetoric in Politics · Paper Radio
- TweetNLP: Cutting-Edge Natural Language Processing for Social Media
- TimeLMs: Diachronic Language Models from Twitter
- BERTweet: A pre-trained language model for English Tweets
- Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks
The paper
Blameocracy: Causal Rhetoric in Politics · Read on arXiv
Francesco Bilotta, Alberto Binetti, Giacomo Manferdini
Bocconi University · Princeton University
Democratic politics requires voters to infer how political actions map into outcomes, yet these causal links are often unclear, giving politicians scope to shape voters' causal accounts. This paper studies the supply and returns of causal rhetoric in politics: politicians' language linking political actions to outcomes through blame and credit. We fine-tune a language model to measure its use in 4.2 million congressional tweets spanning 2012--2023. Causal rhetoric is central to politicians' communication: it accounts for more than one third of tweets, is the language of political disagreement, and rises over time. The political returns of causal rhetoric are concentrated in blame. Combining campaign contributions with quasi-exogenous geographic variation in Twitter adoption, we find that exposure to blame has the largest effects on small donations. Using voting intentions from survey data and an event-study design around blame-only and credit-only days, we find that politicians' support rises after blame-only days. Finally, causal rhetoric appears rarely supported by economic knowledge, particularly blame.
Transcript
Introduction to the show: ident: Economics. The day in economics: theory, mechanisms and the evidence beneath them.
Ruth: I'm Ruth.
Warren: And I'm Warren, here with Iris.
Ruth: We're looking at the paper: "Blameocracy: Causal Rhetoric in Politics".
Warren: Here's the gist of the paper. As an AI researcher, I have meticulously reviewed both provided texts to synthesize a comprehensive and accurate description of the paper "Blameocracy: Causal Rhetoric in Politics." My analysis confirms that the paper is a rigorous empirical study employing advanced econometric techniques to investigate the supply and political returns of causal rhetoric—the language politicians use to link actions to outcomes via blame and credit.
Ruth: Here is the detailed summary: The paper, "Blameocracy: Causal Rhetoric in Politics," investigates how causal rhetoric, defined as politicians' language linking political actions to outcomes through blame and credit, functions within politics.
Warren: The core empirical strategy involves fine-tuning a language model (RoBERTa-large) on 4.2 million congressional tweets spanning 2012–2023 to classify tweets as blame, credit, or neither.
Ruth: This rhetorical style is central to political communication, accounting for over one-third of all tweets and rising over time; furthermore, the study finds that causal rhetoric is primarily the language of political disagreement.
Title and authors: Ruth: So we're looking at this paper today, "Blameocracy: Causal Rhetoric in Politics." It seems like they are focusing on how politicians use specific language to connect what they do with what happens next through blame and credit.
Warren: Yeah, the title itself is pretty direct about the focus being on that causal rhetoric, linking actions to outcomes through those two specific rhetorical moves.
Iris: It’s interesting because it suggests this isn't just some random political chatter; they are trying to measure it systematically across a huge set of tweets.
Ruth: Right. They fine-tuned this large language model on four point two million congressional tweets from two thousand twelve through two thousand twenty-three to classify every tweet as either blame, credit, or neither <ref:2504.06550#pg1>. That scale is what gives the whole study its foundation.
Warren: And according to the abstract, that causal rhetoric style is actually more than just a small part of the conversation; it accounts for over one third of all tweets and it's growing over time too.
Iris: That suggests that this rhetorical style isn't niche stuff; it’s a major channel in how politicians communicate on social media today.
Ruth: Exactly. And they find that this causal rhetoric is primarily the language used when people are disagreeing with each other politically, which is a key finding for understanding political disagreement.
The paper's summary: Ruth: So what does the paper actually find about the political returns of this rhetoric? It seems like they looked at several areas to see where this language actually matters in practice.
Warren: They did look at donation returns first, using something called two-sample location shift estimates, or 2SLS, to see how these blame and credit tweets affect things like donation revenue and the number of donors <ref:2504.06550#pg1>.
Iris: And the main finding there is that exposure to blame generates sizable and consistent gains in support, specifically increasing the donor base by about one point six percent.
Ruth: But here's a twist; credit just doesn't produce a comparable aggregate response on donor counts at all. It seems like blame is the engine for generating measurable political returns in that area.
Warren: That’s interesting because it contrasts sharply with what you might expect from positive messaging, but they also looked at voting intentions using survey data and an event study design around days dominated by blame or credit tweets.
Iris: The analysis there was pretty robust across different specifications, including date fixed effects and controlling for the volume of tweets. They found that support for the politician rises by one percentage point in the week immediately following days dominated by blame-only tweets.
Ruth: So we have both donation returns showing a gain from blame and voting intentions showing a gain from blame, which suggests this language actually translates into tangible political support.
Warren: And they also quantified how much these rhetorical styles amplify themselves by looking at retweets in Appendix E, and the finding there is that blame tweets get significantly more retweets than anything else.
The paper's improvements: Ruth: Beyond just reporting what they found, the authors suggest a few ways to make this measurement better or deeper. They are looking at improvements to the classification system itself.
Iris: They propose improving the content-based causal rhetoric classification by looking at specific linguistic features like using self versus other referential pronouns and attributions to opposing parties.
Warren: That would help them distinguish more clearly between blame and credit by analyzing if someone is talking about their own action or an action taken by another party.
Ruth: And they also suggest a way to assess the quality of the causal claims themselves, which is really interesting for applying this outside of just tracking tweets.
Iris: They can perform a causal claim support assessment by sorting economic causal claims into categories like supported as stated, minor qualified, major qualified, or unsupported based on standard macroeconomic reasoning.
Warren: And they specifically noted that blame claims are much more likely to be major qualified than the others because of how they are framed.
Ruth: That's a concrete way to quantify "how much qualification is needed before the causal claim can be reconciled with standard macroeconomic reasoning," which helps them understand the actual strength of these economic arguments.
Conclusion: Warren: So, wrapping up this paper, the core message is that causal rhetoric is a major input into democratic politics, and blame stands out as being particularly effective at generating measurable political returns.
Ruth: Yeah, they found that exposure to blame has the largest effects on small donations and support gains in voting intentions after those specific days.
Iris: And their work shows that this rhetorical style can also be used to turn complex economic mechanisms into claims that are underqualified yet plausible for political purposes.
Warren: The implication for us is understanding how politicians leverage this language, especially when they're trying to build support or shape public perception through disagreement.
Ruth: It’s a lot to digest, but the paper on "Blameocracy: Causal Rhetoric in Politics" gives us a pretty clear picture of the measurable effects of blame versus credit in political communication.
Iris: It sets up a solid framework for how we can use these linguistic markers to measure causal rhetoric more precisely in the future.
Warren: Yeah, and it leaves us wondering what happens next as the platform changes or as these models get even better at capturing this kind of language.
More episodes
- Daily Summary for 2026-11-19
- 2608.21274-Recommendation Quality and the Concentration of Consumption: Experimental Evidence from Netflix
- 2408.02757-A nonparametric test for diurnal variation in spot correlation processes
- 2306.16393-High-Dimensional Canonical Correlation Analysis
- 2605.17299-Geometric Brownian motion with intermittent entries and exits
- 2607.04753-Fooling Yourself: how narratives shape beliefs
- 2408.14671-Inference on Treatment Effects with Repeatedly Mismeasured High-Dimensional Controls
- 2201.06898-Difference-in-Differences Estimators for Treatments Continuously Distributed at Every Period
- 2501.15307-Influence Functions: A Unified Differentiation-Based Framework
- The day in conversation for 2026-11-19