AI Insights

• The authors also conduct ablations to investigate the effect of ordering bias and verbosity bias on the results, and find that randomizing the response order and using soft weights can mitigate these biases. (ML: 0.98)👍👎
• Listwise scores: Metrics used to evaluate the performance of language models, including Borda, Copeland, Kemeny, PL, and Mallows. (ML: 0.98)👍👎
• Sortition-weighted reinforcement learning from human feedback (RLHF): A method for training language models to align their preferences with those of a diverse and representative group of humans. (ML: 0.98)👍👎
• The authors use the PRISM dataset, which contains interactions between users and a language model, to create a pre-processing pipeline that generates DPO preference pairs for training. (ML: 0.98)👍👎
• The paper presents a novel method for aligning language models with human preferences and values using sortition-weighted RLHF. (ML: 0.97)👍👎
• The paper concludes by highlighting the potential of sortition-weighted RLHF for aligning language models with human preferences and values. (ML: 0.97)👍👎
• The paper presents a method for training language models using sortition-weighted reinforcement learning from human feedback (RLHF), which aims to align the model's preferences with those of a diverse and representative group of humans. (ML: 0.97)👍👎
• DPO preference pairs: Data points generated by the PRISM pre-processing pipeline that represent pairwise comparisons between responses. (ML: 0.97)👍👎
• The paper evaluates the performance of two models, Llama-3.2-1B and Llama-3.2-3B, using a variety of metrics, including listwise scores (Borda, Copeland, Kemeny, PL, Mallows) and pairwise preferences derived from the listwise rankings. (ML: 0.96)👍👎
• Pairwise preferences: Derived from listwise rankings, these metrics measure the probability of one response being preferred over another. (ML: 0.96)👍👎

Abstract
Whose values should AI systems learn? Preference based alignment methods like RLHF derive their training signal from human raters, yet these rater pools are typically convenience samples that systematically over represent some demographics and under represent others. We introduce Democratic Preference Optimization, or DemPO, a framework that applies algorithmic sortition, the same mechanism used to construct citizen assemblies, to preference based fine tuning. DemPO offers two training schemes. Hard Panel trains exclusively on preferences from a quota satisfying mini public sampled via sortition. Soft Panel retains all data but reweights each rater by their inclusion probability under the sortition lottery. We prove that Soft Panel weighting recovers the expected Hard Panel objective in closed form. Using a public preference dataset that pairs human judgments with rater demographics and a seventy five clause constitution independently elicited from a representative United States panel, we evaluate Llama models from one billion to eight billion parameters fine tuned under each scheme. Across six aggregation methods, the Hard Panel consistently ranks first and the Soft Panel consistently outperforms the unweighted baseline, with effect sizes growing as model capacity increases. These results demonstrate that enforcing demographic representativeness at the preference collection stage, rather than post hoc correction, yields models whose behavior better reflects values elicited from representative publics.

Why we are recommending this paper?
Due to your Interest in Democratic Processes

This paper directly addresses concerns about democratic processes and AI alignment, aligning with your interests in democratic systems and political philosophy. The focus on sortition offers a potentially valuable approach to ensuring fairer preference aggregation.

AI Insights

• Memory: The ability of individuals to recall past opinions or experiences, which affects their current voting decisions. (ML: 0.97)👍👎
• It highlights how short-term volatility and long-term inertia coexist in shaping collective decisions. (ML: 0.97)👍👎
• The model suggests that resilience increases when voters rely more strongly on their personal history of opinions. (ML: 0.97)👍👎
• The interplay between memory and shocks determines whether opinion shifts persist or dissipate. (ML: 0.96)👍👎
• The model suggests that resilience increases when voters rely more strongly on their personal history of opinions, underscoring the role of memory in buffering societies against abrupt perturbations. (ML: 0.95)👍👎
• The model assumes a fully connected network, which may not accurately represent real-world social connectivity. (ML: 0.95)👍👎
• Voter-like model: A mathematical framework for simulating the behavior of voters in an election, incorporating individual memory and external influences. (ML: 0.94)👍👎
• The model provides a quantitative framework for understanding the interplay between persistence and sudden perturbations in electoral opinion dynamics. (ML: 0.94)👍👎
• The model captures the core phenomenology of electoral dynamics, including persistence and inertia, as well as sudden realignments triggered by external shocks. (ML: 0.93)👍👎
• Electoral shocks: Abrupt external events or news that can influence voter opinions and alter election outcomes. (ML: 0.91)👍👎

Abstract
We propose a computational framework for modeling opinion dynamics in electoral competitions that combines two realistic features: voter memory and exogenous shocks. The population is represented by a fully-connected network of agents, each holding a binary opinion that reflects support for one of two candidates. First, inspired by the classical voter model, we introduce a memory-dependent opinion update: each agent's probability of adopting a neighbor's stance depends on how many times they agreed with that neighbor in the agent's past $m$ states, promoting inertia and resistance to change. Second, we define an electoral shock as an abrupt external influence acting uniformly over all agents during a finite interval $[t_0, t_0+Δt]$, favoring one candidate by switching opinions with probability $p_s$, representing the impact of extraordinary events such as political scandals, impactful speeches, or sudden news. We explore how the strength and duration of the shock, in conjunction with memory length, influence the transient and stationary properties of the model, as well as the candidates' advantage. Our findings reveal a rich dynamical behavior: memory slows down convergence and enhances system resilience, whereas shocks of sufficient intensity and duration can abruptly realign collective preferences, particularly when occurring close to the election date. Conversely, for long memory lengths or large election horizons, shock effects are dampened or delayed, depending on their timing. These results offer insights into why some sudden political events reshape electoral outcomes while others fade under strong individual inertia. Finally, a qualitative comparison with real electoral shocks reported in opinion polls illustrates how the model captures the competition between voter inertia and abrupt external events observed in actual elections.

Why we are recommending this paper?
Due to your Interest in Political Economy

This research investigates opinion dynamics within electoral campaigns, a key area of interest given your focus on political movements and democratic systems. The use of a computational framework to model voter behavior is highly relevant.

AI Insights

• The output losses translate directly into lower living standards. (ML: 0.91)👍👎
• Previous studies have shown that sanctions can have significant negative effects on a country's macroeconomic performance. (ML: 0.90)👍👎
• The effects were cumulative and compounding over time, indicating that the confrontation operated through channels such as reduced investment, lower productivity growth, and sustained external constraints. (ML: 0.90)👍👎
• Iran's real GDP was approximately 24% lower than it would have been absent the confrontation, on average over the post-shock period. (ML: 0.88)👍👎
• Estimand: The parameter of interest being estimated, in this case, the average post-confrontation gap between Iran's realized outcome path and the outcome path of a weighted combination of non-treated countries. (ML: 0.88)👍👎
• The study relies on the synthetic control method, which may not accurately capture the complex interactions between Iran and the global economy. (ML: 0.86)👍👎
• The estimated per capita GDP gaps between Iran and its synthetic non-confrontation peer showed a 15% decline in income per person relative to the counterfactual. (ML: 0.86)👍👎
• The confrontation with the West had a significant impact on Iran's macroeconomic performance, resulting in a 24% decline in real GDP and a 15% decline in per capita GDP. (ML: 0.84)👍👎
• Synthetic control method (SCM): A statistical technique used to estimate the causal effect of an event or policy on a particular outcome by creating a synthetic series that mimics the pre-treatment path of the treatment unit. (ML: 0.82)👍👎
• The analysis focuses on aggregate output and does not examine other important economic indicators such as employment or trade. (ML: 0.82)👍👎

Abstract
This paper studies the long-run economic and institutional consequences of Iran's confrontation with the West, treating the 2006-2007 strategic shift as the onset of a sustained confrontation regime rather than a discrete sanctions episode. Using synthetic control and generalized synthetic control methods, I construct transparent counterfactuals for Iran's post-confrontation trajectory from a donor pool of countries with continuously normalized relations with the West. I find large, persistent losses in real GDP and GDP per capita, accompanied by sharp declines in foreign direct investment, trade integration, and non-oil exports. These economic effects coincide with substantial and durable deterioration in political stability, rule of law, and control of corruption. Magnitude calculations imply cumulative output losses comparable to civil-war settings, despite the absence of internal armed conflict. The results highlight confrontation as a deep and persistent economic and institutional shock, extending the literature beyond short-run sanctions effects to sustained geopolitical isolation.

Why we are recommending this paper?
Due to your Interest in Political Economy

Coming from the University of Oxford, this paper examines the long-term economic and institutional impacts of geopolitical confrontations, aligning with your interest in political economy and political science. It offers a nuanced perspective on systemic change.

AI Insights

• A transição para uma jornada de trabalho mais curta pode ter um impacto significativo na produtividade e no desempenho econômico O ajuste de curto prazo tende a se concentrar em pequenas empresas, que exibem maior incentivo ao vazamento para a informalidade A transição depende criticamente do desenho regulatório e da forma como o custo privado de operar formalmente é alterado, sobretudo para pequenas empresas Areq: Custo agregado do teto de horas (medido pela perda de produtividade) σsub: Elasticidade de substituição formal–informal 1−τcapS/τS: Alívio temporário de encargos/compliance A transição para uma jornada de trabalho mais curta pode ter um impacto significativo na produtividade e no desempenho econômico O ajuste de curto prazo tende a se concentrar em pequenas empresas, que exibem maior incentivo ao vazamento para a informalidade A transição depende criticamente do desenho regulatório e da forma como o custo privado de operar formalmente é alterado, sobretudo para pequenas empresas O exercício é mais informativo sobre o custo macro de implementação e o canal de realocação formal–informal no horizonte relevante para a política do que sobre efeitos distributivos finos, trajetórias de bem-estar ao longo de anos, ou respostas endógenas de longo prazo. (ML: 0.88)👍👎

Abstract
This paper quantifies, within a short-run structural model with predetermined capital, the immediate effects of imposing a cap on formal working hours that reduces the weekly workweek from 44 to 36 hours. The central object is the total factor productivity required to preserve GDP at its baseline level, A_req, defined as the multiplicative factor applied to A_t that equates output under the policy to output in the baseline. In the baseline simulation, the 44 -> 36 transition implies A_req ~ 8.5%

Why we are recommending this paper?
Due to your Interest in Political Philosophy

This paper tackles the productivity question related to reduced working hours, directly relevant to your interest in democratic systems and political economy. The use of a structural model provides a quantitative approach to a key policy debate.

AI Insights

• The role of social media, particularly YouTube's recommendation algorithms, in facilitating radicalization parallels findings in other extremist contexts where users progress from mainstream to increasingly extreme content. (ML: 0.98)👍👎
• The dataset draws from a single forum, limiting generalizability. (ML: 0.97)👍👎
• The study's focus on thematic analysis may not capture the full complexity of the radicalization process. (ML: 0.97)👍👎
• The role of social media in facilitating radicalization is significant, highlighting the need for online platforms to take responsibility for promoting extremist content. (ML: 0.97)👍👎
• The study's findings suggest that incels follow a typical pattern of radicalization, with social isolation, difficulty coping, and blame-seeking playing key roles. (ML: 0.96)👍👎
• The study found that incels follow a typical psychological radicalization process, with the same social and psychological factors at play as in other extremist communities. (ML: 0.96)👍👎
• Previous research has identified social isolation and difficulty coping as key factors in radicalization. (ML: 0.96)👍👎
• Deradicalization efforts have shown promise in helping individuals leave extremist groups or ideologies behind. (ML: 0.95)👍👎
• Radicalization: The process by which individuals become more extreme in their views and behaviors, often leading to violence or terrorism. (ML: 0.92)👍👎
• Deradicalization: The process of reversing the effects of radicalization, helping individuals to leave extremist groups or ideologies behind. (ML: 0.86)👍👎

Abstract
Incels, or "involuntary celibates", are an extreme, misogynistic hate group that exists entirely online. Members of the community have been linked to acts of offline violence, including mass shootings. Previous research has engaged with the ideologies and beliefs of incels, but none has looked specifically at the radicalization process. In this paper, we perform a thematic analysis on social media posts where incels describe their own radicalization process. We identified six major themes grouped into four chronological steps: Pre-radicalization (themes of Appearance, Social Isolation, and Psychological issues), Searching for Blame, Radicalization, and Post Radicalization. These results align closely with existing work on radicalization among other extremist groups, bringing incel radicalization inline with a growing body of research on understanding and managing radicalization.

Why we are recommending this paper?
Due to your Interest in Political Movements

This paper investigates the radicalization process within a specific online community, aligning with your interest in political movements and activism. Understanding the dynamics of extremist groups is crucial for analyzing social movements.