arXiv

Evidence (17)

• 8Hacker News·10/2/2026research
  AI beats Stratego under hidden-information conditions

  Researchers report an AI system that can play Stratego effectively despite the game's hidden information, a setting that had long challenged computer agents. The work, published in Nature and mirrored on arXiv, suggests new progress in handling imperfect-information games and decision-making under uncertainty.

• 4Hacker News·10/1/2026research
  Context Language Models Paper

  This arXiv paper introduces Context Language Models, a research proposal related to language modeling. The Hacker News post highlights discussion around the paper rather than a product release or finalized system.

• 6Hacker News·9/23/2026funding
  arXiv secures multiyear support to remain independent

  arXiv announced multiyear commitments that will help fund its operations as an independent nonprofit. The support is intended to strengthen the platform’s long-term sustainability for hosting and disseminating research papers.

• 4Hacker News·9/19/2026research
  Open Source Software as a Foundation for Science

  The article argues that modern computational science depends on open, reproducible software and that software should be treated as part of the scientific method. It says scientific results must be testable, systematized, and shareable so others can reproduce, adapt, or reject them.

• 4Hacker News·9/18/2026research
  Empirical study examines harness design for coding agents

  This arXiv paper studies how harness design affects the performance and evaluation of coding agents. It compares design choices empirically to identify which setups lead to more reliable results. The work is mainly relevant to developers and researchers building agentic coding systems.

• 6Hacker News·9/16/2026research
  Research Breaks 1.58-Bit Barrier for Ternary LLMs

  This arXiv paper reports a method for ternary large language models that exceeds the long-cited 1.58-bit per-weight barrier. The work is presented as a research advance in low-bit model representation and compression, with potential implications for more efficient inference and deployment.

• 6Hacker News·9/16/2026research
  Dream-RSI: recursive self-improvement in evolving worlds

  This arXiv paper proposes Dream-RSI, a framework for recursive self-improvement using evolving simulated worlds. It explores how agents can improve by interacting with and adapting to changing environments, with discussion amplified by Hacker News interest.

• 7Hacker News·9/5/2026security
  Trusting-Trust Attack on a Linux Distribution

  This arXiv paper describes a trusting-trust style attack targeting an entire Linux distribution, highlighting how build and supply-chain trust can be subverted. The work is framed as a security research result and underscores risks in compiler and distribution-level integrity assumptions.

• 6Hacker News·8/21/2026open_source
  Kobo e-readers get open-source app platform Cobalt

  Cobalt is an open-source application platform for Kobo e-readers that adds a launcher, signed app store, Rust SDK, and sandboxed runtime. Apps are installed over USB once, then managed on-device over Wi‑Fi, with signatures verified and a reboot restoring the stock Kobo reader.

• 3Hacker News·8/15/2026research
  Simplifying Introductory Calculus

  This arXiv paper proposes a refactoring of introductory calculus to make the subject more conceptually direct and less notation-heavy. It argues for simplifying core definitions and teaching approaches while preserving mathematical rigor.

• 4Hacker News·8/5/2026research
  Study Compares Zigbee and Matter over Thread in Real-World IoT

  This arXiv paper examines the practical performance of Zigbee and Matter over Thread for Internet of Things deployments. It focuses on how the two protocols compare under real conditions, rather than in idealized specifications.

• 6Hacker News·7/31/2026research
  Maxwell Conjecture Disproved

  An arXiv paper reports a counterexample that disproves the Maxwell Conjecture. The post drew strong Hacker News attention, with discussion centered on the mathematical result and its implications for the underlying theory.

• 7Hacker News·7/20/2026research
  Study Finds AI-Style Writing Has Risen Across arXiv Papers

  A study scored the full text of 12,750 arXiv papers and found that roughly a third of recent submissions read as machine-written under its calibrated detector. The effect varies widely by field, with computer science highest and mathematics near zero, while the authors emphasize false-positive limits and other methodological caveats.

• 3Hacker News·7/16/2026research
  Mathematics of Data Science

  This arXiv paper appears to survey or formalize mathematical foundations relevant to data science. The Hacker News discussion is brief, with interest focused on the paper itself rather than a product or release event.

• 4Hacker News·7/12/2026research
  Paper on Automation Without Understanding

  This arXiv paper examines the risks and limits of automation systems that can produce outputs without genuine understanding. It argues that apparent performance can hide brittle reasoning, making such systems unreliable in settings where correctness and interpretability matter. The Hacker News discussion reflects interest in the broader implications for AI deployment and evaluation.

• 6Hacker News·7/3/2026research
  Paper claims markets are competitive only if P ≠ NP

  This arXiv paper argues a theoretical connection between market competitiveness and the classic computational complexity question P versus NP. The post drew significant Hacker News discussion, but it is presented as a research claim rather than a confirmed practical result.

• 6Hacker News·7/1/2026framework_update
  arXiv Announces Its Next Phase

  arXiv outlines plans for its next chapter, focusing on sustaining and modernizing the preprint platform that serves the research community. The update emphasizes governance, long-term stability, and the infrastructure needed to support continued growth.