weekly-papers-2026-06-27 on Sparse Notes https://sparsenotes.com/series/weekly-papers-2026-06-27/ Recent content in weekly-papers-2026-06-27 on Sparse Notes https://sparsenotes.com/images/og-default.png https://sparsenotes.com/images/og-default.png Hugo -- gohugo.io Sat, 27 Jun 2026 00:00:00 +0000 Bitcoin: A Peer-to-Peer Electronic Cash System https://sparsenotes.com/posts/2026/06/papers/2026-06-27-bitcoin-whitepaper/ Sat, 27 Jun 2026 00:00:00 +0000 https://sparsenotes.com/posts/2026/06/papers/2026-06-27-bitcoin-whitepaper/ Nakamoto's 2008 nine-page whitepaper that solved double-spend without a trusted third party using proof-of-work, hash-linked blocks, and the longest-chain rule. Hallucination in World Models is Predictable and Preventable https://sparsenotes.com/posts/2026/06/papers/2026-06-27-world-model-hallucination/ Sat, 27 Jun 2026 00:00:00 +0000 https://sparsenotes.com/posts/2026/06/papers/2026-06-27-world-model-hallucination/ Generative world models hallucinate in low-coverage regions of state-action space; three data-centric signals both detect failures and guide mitigation. Weekly CS Paper Digest — Jun 20 – Jun 27, 2026 https://sparsenotes.com/posts/2026/06/papers/ Sat, 27 Jun 2026 00:00:00 +0000 https://sparsenotes.com/posts/2026/06/papers/ Three blog-length entries from this week's CS paper digest: Nakamoto's Bitcoin whitepaper (2008) as Seminal pick, plus top picks on the co-failure ceiling for multi-model LLM ensembles and predictable, preventable hallucination in generative world models. When Does Combining Language Models Help? A Co-Failure Ceiling on Routing, Voting, and Mixture-of-Agents Across 67 Frontier Models https://sparsenotes.com/posts/2026/06/papers/2026-06-27-co-failure-ceiling/ Sat, 27 Jun 2026 00:00:00 +0000 https://sparsenotes.com/posts/2026/06/papers/2026-06-27-co-failure-ceiling/ An information-theoretic ceiling on ensemble gains: accuracy cannot exceed 1 − β, where β is the rate at which every model is wrong on the same query.