Daily Digest¶
Summaries grouped by the date they were summarized.
2026-06-23 — Tuesday, June 23, 2026¶
1 essay summarized
The Best Deal in AI Isn't for the People Who Build¶
Premium AI subscriptions are subsidised far below API cost — a predatory pricing play that builds dependency before the bill arrives.
AI Pricing Subscriptions Enterprise AI
Category: Artificial Intelligence
2026-06-16 — Tuesday, June 16, 2026¶
18 essays and 11 papers summarized
Why AI Hasn't Replaced Software Engineers¶
Arvind Narayanan and Sayash Kapoor explain that software development operates as a 'decide-execute-deliver sandwich' — AI compresses the middle execution layer, but decision-making and accountability resist automation.
Category: Science and Tech
What Is an Agent? — Hadley Wickham's Ground-Up Framework for LLM Tool-Using Systems¶
Hadley Wickham builds a precise definition of AI agents from first principles — vocabulary for turns, rounds, tools, and harnesses — and tackles the fundamental tension between usefulness and safety.
Category: Science and Tech
shadcn/improve: Agentic Code Review with Divide and Conquer¶
An open-source tool by shadcn that uses expensive models to audit codebases and write plans, then hands off mechanical execution to cheaper models
Category: Science and Tech
Sam Bowman on Europe's 'Plausible and Frightening' Future¶
Sam Bowman shares a viral article painting a bleak vision of Europe's near future — economic collapse, demographic decline, and vassal status under the US or China — sparking widespread debate about the continent's structural trajectory.
Category: Society and Politics
SafeTensors vs Pickle: Why ML Supply Chain Security Matters¶
Comparing Python pickle's arbitrary code execution vulnerability with HuggingFace SafeTensors' data-only design for secure ML model distribution
Category: Artificial Intelligence
Russia's Oil Is Getting Hammered¶
Ukraine's three-pronged drone campaign targets Russian oil export capacity, budget revenue, and future production — exploiting Russia's lack of strategic reserves
Category: Society and Politics
Qin Hui on Why Condemning Trump Is Not Enough — The Iran Crisis Through a Confucian Liberal Lens¶
Retired Tsinghua professor Qin Hui argues that blame for the Iran crisis falls on nearly every party — Iran's theocracy, Trump's reckless strike, AND passive European/Asian states — in a bracing dissection of liberal passivity.
Category: Society and Politics
OpenRouter Fusion: Beating Frontier Models by Synthesizing Multiple Models¶
OpenRouter's Fusion API dispatches prompts to multiple models in parallel, then uses a 'judge' model to synthesize their responses — consistently beating any single frontier model at lower cost.
Category: Artificial Intelligence
Open Knowledge Format — Google Cloud's Vendor-Neutral Spec for LLM-Friendly Knowledge Repositories¶
Google Cloud introduces OKF v0.1, an open, vendor-neutral specification that formalizes the 'LLM-wiki' pattern — representing knowledge as a directory of markdown files with YAML frontmatter.
Category: Science and Tech
Notes on Egypt¶
Visiting Egypt's New Administrative Capital — a desert megaproject of empty record-breaking structures reveals the paradox of autocratic nation-building
Category: Society and Politics
Michael McNair on Chinese Currency Weakness and Dollar Squeeze¶
Michael McNair argues that Chinese yuan weakness is driven by capital flow dynamics that appear market-driven at the final step but are fundamentally policy-induced, tracing the mechanism to Trump-era bond market mechanics and the emerging US Sovereign Wealth Fund.
Category: Economics and Finance
How (and Why) SpaceX Will Colonize Mars¶
Tim Urban's epic series on the gap between Apollo-era expectations and four decades of stagnation — and Elon Musk's plan for Mars colonization
Category: Science and Tech
The Economist's Villainous Blueprint — Thomas Piketty's Global 'Managed Decline' Plan¶
Veronique de Rugy delivers a scathing critique of Thomas Piketty's degrowth blueprint, arguing it is internally contradictory, built on discredited climate scenarios, and represents the 'ultimate luxury belief of tenured professors.
Category: Economics and Finance
China Is Innovative. Its Economy Is a Mess. Which Matters More?¶
The Economist examines China's paradox — world-leading innovation alongside severe structural economic problems — and asks whether the mess drowns the innovation or innovation saves the mess.
Category: Economics and Finance
Apple Container Machine — Fast, Lightweight Linux Environments on Mac¶
Apple's container-machine provides an integrated Linux environment on Mac — fast, lightweight, and persistent — based on standard OCI images but modeled as a full Linux environment rather than a single application.
Category: Science and Tech
Anthropic, Please Don't Ruin Bio¶
A biologist critiques Anthropic's AI bio-risk fear-mongering — arguing poor public research and flawed benchmarks threaten legitimate scientific access
Category: Science and Tech
AI and White-Collar Job Losses¶
Wall Street's massive layoffs use AI as a scapegoat — the real driver is a labor overhang that existed long before generative AI
Category: Economics and Finance
A Normal Person 30 Years Ago — Testing the 'Today's Far Right = Yesterday's Normal' Meme¶
Cremieux Recueil tests the claim that today's far-right views mirror what a normal person held 30 years ago, finding the Overton Window shifted left but unevenly — with dramatic changes on race, gender, and sexuality and near-stasis on economics.
Category: Society and Politics
Z-Reward: Beyond Scalar Rewards¶
A teacher-student framework that decouples reasoning-intensive reward judgment from efficient deployment, achieving 89.6% human preference accuracy with a 27B teacher and 88.6% with a distilled 9B student.
Reward Models RLHF Image Generation
Category: AI / ML
The Superiority of Economists¶
Sociological analysis of economics' uniquely dominant and self-reinforcing position among US social sciences, examining elite training pipelines, insular citation patterns, and the profession's authority structures.
Economics Profession Sociology of Science
Category: Economics and Finance
Self-Harness: Harnesses That Improve Themselves¶
A fixed LLM improves its own harness (system prompt + tools + runtime) through a 3-stage loop of weakness mining, harness proposal, and proposal validation, achieving +53% held-out gains on Terminal-Bench 2.0
Category: AI / ML
MiniMax Sparse Attention (MSA): Blockwise Sparse Attention for Long Context¶
Blockwise sparse attention built on GQA achieving 28.4× per-token FLOP reduction, 14.2× prefill speedup, and 7.6× decoding speedup via a lightweight index branch that selects Top-k KV blocks per GQA group
Category: AI / ML
Latent Spatial Memory for Video World Models (Mirage)¶
Persistent 3D latent spatial memory for video generation that caches scene information directly in diffusion latent space, achieving 10.57× faster video generation and 55× lower GPU memory than pixel-space baselines.
Video Generation World Models 3D Memory
Category: AI / ML
How AI Agents Reshape Knowledge Work¶
First field evidence on the economic impact of shifting from conversational AI assistants to autonomous agents, showing 48× more machine work per session, 87% task completion time reduction, and 55% lower user dissatisfaction
Category: AI / ML
Geometry of On-Policy Distillation: A Parameter-Space Analysis¶
First comprehensive parameter-space analysis showing On-Policy Distillation (OPD) occupies a 'relaxed off-principal' regime between SFT and RLVR, with 51.6% unchanged parameters and early subspace locking to a persistent low-dimensional update channel
Category: AI / ML
FlashMemory-DeepSeek-V4: Lookahead Sparse Attention¶
Introduces Lookahead Sparse Attention (LSA) to replace passive KV cache, achieving ~90% memory reduction at 500K context with a lightweight Neural Memory Indexer trained independently.
Long Context Sparse Attention Memory
Category: AI / ML
Consensus is Strategically Insufficient for Multi-Agent Value-Laden Tasks¶
Multi-agent consensus-seeking is insufficient for value-laden tasks where disagreement reflects genuine normative uncertainty; proposes a knowledge representation (KR) layer with 4 disagreement states — CA, DA, CD, DD — with CD best predicting high human disagreement (Cohen's d = 0.80)
Category: AI / ML
Agents' Last Exam (ALE): Benchmarking Economically Valuable AI Work¶
A benchmark of 1,490 task instances across 55 subfields in 13 industry clusters for measuring economically valuable AI work, with deterministic evaluation and a failure taxonomy showing understanding (31%), approach (47%), and execution (22%) errors
Category: AI / ML
Agentopia: Long-Term Life Simulation¶
Framework for long-term life simulation in multi-agent societies with LLM agents across 3 fictional worlds, 100 agents per world, over 10 simulated years.
Multi-Agent Simulation LLM Agents
Category: AI / ML
2026-06-08 — Monday, June 08, 2026¶
21 essays summarized
Unlimited Sensing: When Noise Becomes the Signal¶
Category: Science and Tech
Transformers Are Inherently Succinct¶
Category: Artificial Intelligence
Time to Stop Forecasting China's Surplus Away¶
Brad Setser argues the IMF has systematically underestimated China's current account surplus (now ~5% of GDP) due to three structural factors — and that standard policy recommendations are insufficient.
Category: Economics and Finance
Tim Cook's Apple¶
A deep analysis of Tim Cook's 15-year CEO tenure — from $350B to $4T market cap, the Services transformation, the flops (Apple Car, Vision Pro), and the AI strategy gap.
Category: Science and Tech
Supervision: Reusable Computer Vision Tools¶
Category: Science and Tech
The Mismeasurement of European Productivity¶
Aghion, Bergeaud, and Garicano rebut Krugman's claim that the US-Europe productivity gap is a measurement illusion — showing the gap is real, large, and consequential for wages, welfare, and strategic capacity.
Category: Economics and Finance
Should the Lion Lie Down With the Electric Lamb?¶
Category: Society and Politics
LEAP: Supercharging LLMs for Formal Mathematics with Agentic Frameworks¶
Category: Artificial Intelligence
How the Boomers Screwed Europe¶
The Economist's Charlemagne column on how Europe's primary inequality fault line has shifted from West vs East to Boomers vs Millennials — and the intergenerational confidence trick at the heart of the European welfare state.
Category: Society and Politics
GPT-5.4 Pro Solves Erdős Problem #1196¶
Category: Artificial Intelligence
Energy Geopolitics and European Markets in Crisis¶
Category: Society and Politics
Let's Talk About Encrypted Reasoning¶
Category: Science and Tech
Staying Power: The Dollar-Security Nexus¶
Ho-fung Hung argues the dollar's global dominance is sustained less by market network effects and more by the US military security umbrella — and China's capital controls paradoxically prolong it.
Category: Economics and Finance
Dia2: Streaming Conversational TTS Model¶
Category: Science and Tech
Google DeepMind Science Skills¶
Category: Science and Tech
Dark Knowledge: Distilling Ensemble Knowledge into Smaller Models¶
Category: Artificial Intelligence
China's Not the Problem. We Are.¶
Ross Douthat and Kyle Chan on the NYT 'Interesting Times' podcast explore how the US and China have fundamentally different AI goals — making it hard to call it a simple 'race.
Category: Society and Politics
Can AI Refute Economic Theory?¶
Category: Economics and Finance
Brooks-lint: AI Code Reviews Grounded in 12 Classic Engineering Books¶
Category: Science and Tech
Is AI More Expensive Than the Employees It's Replacing?¶
Category: Economics and Finance
AI Enthusiasts Are in a Race Against Time, AI Skeptics Are in a Race Against Entropy¶
Charity Majors on the widening chasm between AI enthusiasts and skeptics in engineering teams — and how the Fin/Intercom case study shows that exceptional discipline, not AI magic, drives real results.
Category: Science and Tech
2026-06-07 — Sunday, June 07, 2026¶
1 essay summarized
If LLMs Have Human-Like Attributes, Then So Does Age of Empires II¶
Adrian de Wynter demonstrates that LLM anthropomorphism conclusions are logically unsound by building a Turing-complete neural network inside Age of Empires II — proving the substrate itself drives perceived humanness.
LLMs Anthropomorphism Theory of Mind
Category: Artificial Intelligence
2026-06-05 — Friday, June 05, 2026¶
19 essays summarized
Games Between Programs — The Ruliology of Competition¶
Stephen Wolfram applies his 'ruliological' method of systematically enumerating all possible strategies to the study of agent competition.
Category: Science and Tech
Simon Willison — Uber Caps Usage of AI Tools Like Claude Code¶
Uber blew its 2026 AI budget in 4 months. New policy caps all employees at $1,500/month per AI coding tool — a more sensible approach than tokenmaxxing leaderboards.
Category: Science and Tech
The Tale of the Strawberries — Measurement Non-Invariance¶
Cremieux explains how measurement non-invariance can make trends meaningless if you don't know what you're measuring.
Category: Science and Tech
The State and the People¶
Alan Jacobs reflects on A.J.P. Taylor's depiction of the pre-WWI Englishman who barely noticed the state, and how war permanently transformed the relationship between citizen and government.
Category: Society and Politics
Spytial — Diagramming Program Values by Spatial Refinement¶
A programming language developed by Brown PLT that creates diagrams of program values through declarative spatial refinement rather than imperative drawing code.
Category: Science and Tech
Brad Setser — Taiwan's Currency Manipulation Is More Aggressive Than China's¶
Brad Setser argues Taiwan's CBC engages in more aggressive, opaque currency manipulation than China's PBOC, using hidden FX swaps with life insurers.
Category: Economics and Finance
Brad Setser on China's Trade Surplus and Capital Outflows¶
China Trade Surplus Capital Flows
Category: Economics and Finance
The Power of Many Small Sex Differences — Cognition, Personality, and Interests¶
Large-scale Swedish study finds that combining many small sex differences in cognition, personality, and interests yields 80% classification accuracy.
Category: Science and Tech
Mathematicians Issue Warning as AI Rapidly Gains Ground (Leiden Declaration)¶
The Leiden Declaration on AI and Mathematics — the most significant collective response from a major academic discipline to AI encroachment — endorsed by the International Mathematical Union.
Category: Society and Politics
MAI-Thinking-1 — Building a Hill-Climbing Machine¶
Category: Artificial Intelligence
Yes, Living Standards Have Grown Slower in Northwest Europe than in the U.S.¶
Economics Living Standards US vs Europe
Category: Economics and Finance
Passive Heart Rate Monitoring via Smartphone Camera¶
Google's PHRM system measures heart rate through the front-facing camera during face unlock using remote photoplethysmography (rPPG).
Category: Science and Tech
DeepLearning.AI — Forward Deployed Engineers and the Future of AI Engineering¶
Andrew Ng on the rise of AI Forward Deployed Engineers — embedded engineers customizing AI solutions for clients. The role originated at Palantir and is surging at OpenAI and Anthropic.
Category: Artificial Intelligence
AI Is the Most Dangerous Arms Race in History¶
AI Safety Geopolitics US-China
Category: Society and Politics
FT — China's Comparative Advantage Is Industrial Policy¶
Tej Parikh argues China's true advantage is institutional capacity to coordinate resources — not just subsidies. Western governments spent over $1 trillion in 2025 trying to counter China's industrial policy.
Category: Economics and Finance
FT — World's Biggest EV Maker Weans Itself Off Supply-Chain Finance¶
BYD phases out its Di Lian supplier payment system with 300-day payment cycles, replacing it with banknotes and cash after Beijing pressure. GMT Research revealed BYD's true net debt ~323B yuan vs 27.7B on books.
Category: Economics and Finance
AUTOLAB — Benchmarking Long-Horizon Empirical Optimization¶
Category: Artificial Intelligence
When AI Builds Itself — Recursive Self-Improvement¶
Anthropic's analysis of how AI is increasingly used to develop AI, with >80% of their production code now authored by Claude and autonomous task durations doubling every 4 months.
Category: Artificial Intelligence
Defending Code — Anthropic's Vulnerability Discovery Reference Harness¶
Anthropic releases an open-source reference harness for autonomous vulnerability discovery, including built-in skills for threat modeling, scanning, triage, and patching.
Category: Artificial Intelligence
2026-06-04 — Thursday, June 04, 2026¶
1 essay summarized
Microsoft GraphRAG — Structured Knowledge Graph RAG¶
Microsoft's GraphRAG creates hierarchical knowledge graphs from text for global and local reasoning over private datasets, outperforming baseline semantic-search RAG.
RAG Knowledge Graphs Microsoft Research
Category: Artificial Intelligence
2026-05-30 — Saturday, May 30, 2026¶
3 essays summarized
Billionaires Build¶
Paul Graham argues that billionaires are builders who create what people want, not exploiters — and the same principle that makes someone a billionaire is what Y Combinator tests for in interviews.
Startups YC Founder Psychology
Category: Economics and Finance
Scandal-Hit Miyagino Stable (Hakuho's Heya) Shuts Down Permanently¶
The Japan Sumo Association formally dissolved the Miyagino stable, formerly headed by legendary yokozuna Hakuho, after an abuse scandal kept it shuttered since April 2024.
Sumo Miyagino Stable Hakuho
Category: Science and Tech
How Does Docusign Have 7,000 Employees?¶
An examination of Docusign's massive headcount, the future of B2B SaaS in the age of AI coding agents, and what makes a software business truly defensible.
SaaS AI Agents Business Strategy
Category: Economics and Finance
2026-05-29 — Friday, May 29, 2026¶
18 essays summarized
Systems Engineering and the V-Model: Lessons from an Autonomous Solar Powered Hydrofoil¶
A retrospective analysis of a failed student engineering project using Systems Engineering frameworks, revealing how V-Model breakdowns and reactive iteration lead to project failure.
Category: Science and Tech
StoryScope: Investigating Idiosyncrasies in AI Fiction¶
A new pipeline extracts discourse-level narrative features to distinguish AI-written fiction from human-written stories, achieving 93.2% accuracy using plot structure alone.
AI Detection Narrative Analysis Creative Writing
Category: Artificial Intelligence
Perfect Randomness Realized for the First Time¶
Category: Science and Tech
Introducing a New Design for Microsoft 365 Copilot¶
Microsoft's Chief Design Officer Jon Friedman announces a major redesign of Microsoft 365 Copilot, shifting from static AI layering to a truly adaptive task-aware workspace.
Microsoft 365 AI Copilot
Category: Science and Tech
LocateAnything: Fast and High-Quality Vision-Language Grounding with Parallel Box Decoding¶
NVIDIA Research introduces LocateAnything, a unified generative grounding and detection framework using Parallel Box Decoding that achieves 12.7 BPS throughput — over 10x faster than Qwen3-VL.
Category: Artificial Intelligence
Language Models Need Sleep¶
An intriguing exploration of whether large language models, like humans, benefit from structured 'rest' periods during training or inference.
Category: Artificial Intelligence
How Google SRE Is Using Agentic AI to Improve Operations¶
Google SRE is adopting agentic AI as a 'force multiplier' — deploying Gemini models, custom agents, and AI-driven insights across reliability engineering.
Google SRE AI Agents DevOps
Category: Artificial Intelligence
Join the New AI Agents Vibe Coding Course from Google and Kaggle¶
Category: Artificial Intelligence
Free Will Is Still Undefeated¶
Rob Henderson pushes back against the determinist consensus, arguing that free will is a necessary legal, philosophical, and social construct we cannot abandon.
Free Will Philosophy Neuroscience
Category: Society and Politics
FluxMem: Rethinking Memory as Continuously Evolving Connectivity¶
A new arxiv paper proposes FluxMem, a memory architecture that models agent memory as a dynamically editable heterogeneous graph that refines its topology through three key stages.
LLM Agents Memory Knowledge Graphs
Category: Artificial Intelligence
EU Wants Crisis Powers to Seize Control of Chip Supplies¶
Category: Society and Politics
Deep Understanding¶
Category: Artificial Intelligence
Property Hunters: China Regions Mine Ledgers for Dormant Riches in 'Idle' Assets¶
Category: Economics and Finance
Commentary: To Fix Its Economy, China Needs to Pay Its Workers More¶
Chen Changhua argues China must transition from a 'high-efficiency, low-profit' model toward a consumption-driven economy by raising wages to close the growing supply-demand gap.
Category: Economics and Finance
Economists Urge Export Tax Rebate Cuts as Trade Surplus Hits Record¶
China's trade surplus reached a record $1.2 trillion in 2025, prompting economists to call for targeted cuts to export tax rebates to redirect fiscal savings toward households.
Category: Economics and Finance
How AWS Used Random Graph Theory to Build More Efficient Data Centers¶
Category: Science and Tech
Amazon Thinks the Future of Data Centers Depends on a Technical Problem It Just Solved¶
AWS achieves a breakthrough with Resilient Network Graphs (RNG) — a 'quasi-random' data center network architecture that replaces the legacy fat-tree design.
AWS Data Centers Networking
Category: Science and Tech
All Elementary Functions from a Single Binary Operator¶
Researchers have discovered that all elementary functions can be constructed from a single binary operator, fundamentally reducing the set of required mathematical primitives.
Category: Science and Tech
2026-05-28 — Thursday, May 28, 2026¶
5 essays summarized
When Quiet Undersea Volcanoes Turn Disruptive¶
Scientists discover that mid-ocean ridge volcanoes can turn explosive at shallow depths (~300m), capable of forming new islands.
Volcanoes Geology Mid-Atlantic Ridge
Category: Science and Tech
Microsoft Agent Governance Toolkit — Deterministic Guardrails for AI Agents¶
An open-source toolkit for enforcing deterministic policy controls over AI agents at the tool-call level, bypassing unreliable prompt-level safety.
AI Safety Agent Governance Microsoft
Category: Artificial Intelligence
Xi Jinping Quit Smoking. China Still Cannot.¶
China's tobacco monopoly is so financially vital to the government that even Xi Jinping has failed to curb the country's smoking habit.
China Tobacco Public Health
Category: Society and Politics
A Way to Challenge the Groupthink of Scholarly Journals¶
Peer review has become a closed system that protects shoddy and politically motivated research — here is how to fix it.
Peer Review Academic Publishing Groupthink
Category: Science and Tech
Anthropic and OpenAI Have Found Product-Market Fit¶
Simon Willison argues that coding agents (Claude Code, Codex) have unlocked genuine PMF for AI labs, evidenced by enterprise pricing shifts, massive token consumption, and a SpaceX deal worth $1.25B/month.
AI Product-Market Fit Claude Code
Category: Artificial Intelligence
2026-05-27 — Wednesday, May 27, 2026¶
15 essays summarized
Webwright: A SWE-Style Browser Agent Framework¶
Microsoft's Webwright achieves SOTA on long-horizon web tasks by treating the browser as a disposable environment the agent spawns while coding.
AI Agents Browser Automation Microsoft
Category: Artificial Intelligence
Using AI to Write Better Code More Slowly¶
LLMs are extremely flexible — you can use them just as effectively to write high-quality code more slowly, through multi-model PR review and careful validation.
AI Coding Software Engineering Code Review
Category: Artificial Intelligence
It Takes Two Neurons To Ride a Bicycle¶
A surprisingly simple two-neuron network can control a bicycle, revealing deep insights about control theory, learning, and biological intelligence.
Neural Networks Control Theory Physics
Category: Science and Tech
The Timing of the Impending Crude Crisis¶
A quantified timeline for a looming oil price crisis as temporary buffers deplete following the Strait of Hormuz disruption.
Oil Energy Crisis Geopolitics
Category: Economics and Finance
The Shocking Truth About Fairness¶
Fairness norms are not universal moral absolutes — they reflect the distribution of bargaining power in a given society.
Fairness Game Theory Moral Philosophy
Category: Society and Politics
The AI Decoupling: How the Tech Economy Split in Two¶
The tech economy structurally decoupled in May 2025 — SaaS/Cloud crashed while AI Labs soared, driven by MoE inference economics and synthetic data pipelines.
AI Economy SaaS MoE
Category: Artificial Intelligence
Repentance and Forgiveness in a Pornified Age¶
The ubiquity of internet pornography has created a quiet crisis undermining marriage — and the solution lies in male repentance and female forgiveness.
Marriage Pornography Society
Category: Society and Politics
A Boomer, But An Augustinian: On Magnifica Humanitas¶
Pope Leo XIV's first encyclical presents an Augustinian vision grounding human grandeur in creaturely limits, opposing the technocratic paradigm.
Catholicism AI Ethics Theology
Category: Society and Politics
How to Remember Everything You Read¶
A comprehensive primer on encoding, memory retention, and synthesis techniques — from spaced repetition and active recall to Zettelkasten and the Feynman Technique.
Memory Learning Reading
Category: Self-Help and Psychology
How I Read¶
Rob Henderson shares his system for reading ~40-50 books per year through deliberate scheduling, phone discipline, and treating deep reading as a non-negotiable habit.
Reading Habits Productivity
Category: Self-Help and Psychology
Your Ultimate Guide to Attention Mechanism, QKV, and KV Cache¶
A comprehensive explainer of the attention mechanism in Transformers — QKV, self-attention, multi-head attention, and modern KV cache variants.
Transformers Attention Deep Learning
Category: Artificial Intelligence
Council of High Intelligence: Multi-Model AI Deliberation¶
18 AI personas deliberate your hardest decisions across multiple LLM providers — Aristotle, Feynman, Kahneman, Torvalds and more in structured multi-round deliberation.
AI Agents Decision-Making Open Source
Category: Artificial Intelligence
Analyst on China's Spent Rocket Stages: Things Only Continue to Get Worse¶
China's explosive growth in orbital launches is accompanied by a systemic failure to properly dispose of rocket upper stages, contradicting international debris mitigation norms.
Space Orbital Debris China
Category: Science and Tech
The World's Biggest EV Maker Weans Itself Off Supply-Chain Finance¶
BYD shifts away from its Dilian promissory note system toward commercial paper and bank notes, as regulatory crackdown and financial slowdown force change.
BYD EV Supply Chain
Category: Economics and Finance
Audience Capture and Short-Form Slop¶
The modern content economy is a broken slogfest of algorithmic short-form slop, AI noise, and audience capture — and how to escape it.
Content Creation Social Media AI
Category: Society and Politics
2026-05-26 — Tuesday, May 26, 2026¶
11 essays summarized
Ukraine's Intermediate-Range Strike Campaign and a New Phase of the War¶
ISW analysis shows Ukraine breaking out of positional warfare: Russia suffered net -116 km² territory loss in April 2026, daily advance rates collapsed, and the Hornet AI-driven UAV is proving decisive against Russian electronic warfare.
Society and Politics Ukraine Russia
Category: Society and Politics
SkillOpt: Executive Strategy for Self-Evolving Agent Skills¶
SkillOpt introduces the first systematic controllable text-space optimizer for agent skills, treating skill documents as trainable external state — achieving best-or-tied results across all 52 evaluated benchmark cells.
AI Agents LLM Optimization Self-Improvement
Category: Artificial Intelligence
LLM Hallucinations in the Wild: Large-Scale Evidence from Non-Existent Citations¶
A large-scale empirical study quantifying how often LLMs fabricate citations to non-existent papers, providing systematic evidence of hallucination patterns across real-world outputs.
Artificial Intelligence LLM Hallucinations Citation Analysis
Category: Artificial Intelligence
Leadership Lessons from 2,000 Years of the Catholic Church¶
Javier Fernández Aguado examines the Catholic Church as a uniquely resilient organisation — pioneering interim management, employer branding, and VUCA thinking centuries before management theory — and extracts lessons on servant leadership, mission-focus, and surviving existential crises.
Self-Help and Psychology Leadership Organisational Resilience
Category: Self-Help and Psychology
Huawei's τ Scaling Law: Reframing Semiconductor Competition from Geometry to Time¶
Huawei announces Tau (τ) Scaling Law, reframing chip competition from geometric transistor shrinking to time-domain optimisation, with LogicFolding in Kirin 2026 achieving +55% density and +41% energy efficiency at a fixed process node.
Science and Tech Huawei Semiconductors
Category: Science and Tech
The Hall–Héroult Process: How Aluminium Went From Precious Metal to Ubiquitous Commodity¶
Before 1886, aluminium was more expensive than gold or silver. The Hall–Héroult electrolytic process — discovered independently by two 22-year-olds — transformed it into a cheap industrial commodity, enabling modern aviation, construction, and packaging.
Science and Tech Aluminium Industrial History
Category: Science and Tech
Figma's FigCache: Next-Generation Data Caching at Scale¶
Figma built FigCache — a stateless RESP-wire-protocol Redis proxy — to solve critical Redis scalability bottlenecks, achieving six nines of uptime since mid-2025 through a configuration-driven architecture with Starlark routing.
Science and Tech Figma Caching
Category: Science and Tech
China's Housing Slump: Signs of Bottoming Out, or Another Lull?¶
NYT's Keith Bradsher reports that tier-1 city home prices rose 2% Feb-April after a 38% crash since 2021, but with 90 million empty or unfinished apartments nationwide, analysts remain divided on whether this is a genuine bottom or a pause before further decline.
Economics and Finance China Housing
Category: Economics and Finance
China's $200B Equity Outflow: Brad Setser on Capital Flight and the $1T Question¶
Brad Setser examines China's equity outflows (~$200B) and total capital outflows (~$1T), arguing the figures are large but manageable relative to GDP, and less severe than historical episodes like South Korea's 2008 outflow.
Economics and Finance China Capital Flows
Category: Economics and Finance
Digital Design & Computer Architecture — Cache Design, Management, and Tradeoffs¶
Onur Mutlu's lecture from ETH Zürich (Spring 2026) on cache memory design: organisation, management strategies, associativity, replacement policies, and the performance tradeoffs that define modern memory hierarchies.
Science and Tech Computer Architecture Cache Memory
Category: Science and Tech
America Keeps Falling Uphill — Beijing's Big Blind Spot¶
David Ignatius argues that while the Trump-Xi summit optics seemed to favour Beijing, US structural strengths — resilient economy, global alliances, innovation capacity — are consistently underestimated by Chinese strategic planners.
Society and Politics US-China Relations Trump-Xi Summit
Category: Society and Politics
2026-05-25 — Monday, May 25, 2026¶
2 essays summarized
Two Erdős Problems on Points in the Plane and AI¶
Lance Fortnow and Bill Gasarch's blog compares the human-led resolution of the Erdős Distinct Distance Problem with the AI-generated counterexample to the Erdős Unit Distance Problem — and reflects on what this means for the future of mathematics.
Mathematics AI Erdős
Category: Artificial Intelligence
11 Types of Cross-Industry Innovation¶
Trung Phan's SatPost distills Matt Ridley's innovation framework and catalogs 11 cross-industry innovations — from Gutenberg's printing press (borrowed from wine) to the Dyson vacuum (borrowed from sawmills) — showing how breakthrough ideas emerge when concepts recombine across domains.
Cross-Industry Innovation Matt Ridley Trung Phan
Category: Science and Tech
2026-05-24 — Sunday, May 24, 2026¶
7 essays summarized
The YC Chief Who Codes 10,000 Lines a Day Has a Simple Secret¶
Garry Tan (YC CEO) ships 600k lines of production code every 60 days part-time through a "thin harness, fat skills" architecture — the philosophy that AI agent effectiveness is determined by the surrounding system, not the model itself. The Claude Code source leak confirmed the depth of this approach.
Science and Tech AI Software Engineering
Category: Science and Tech
Chartwright: Turning Tufte's Book Into an AI Skill¶
The tufte-vdqi-plugin (Chartwright) demonstrates a powerful pattern: encoding Edward Tufte's 'The Visual Display of Quantitative Information' — a 200-page book — into 10 composable AI skills that any agent can invoke to produce Tufte-quality charts on demand.
Science and Tech AI Data Visualization
Category: Science and Tech
Simple Input–Output Dependencies Explain Neuronal Activity¶
A Yale-led study across mouse hippocampus, visual cortex, and C. elegans whole-brain recordings finds that a simple logistic model — equivalent to a single artificial neuron — explains over 90% of neuronal activity variability using just direct input-output dependencies, without requiring complex nonlinear dendritic computation.
Science and Tech Neuroscience AI
Category: Science and Tech
How Many Roman Cities Survived the Dark Ages? Only 8%¶
A data analysis of 405 Roman cities shows that only 33 (8%) survived as cities into the 8th century — the remaining 92% were either destroyed or depopulated to become small towns or settlements, revealing the true scale of the post-Roman urban collapse.
Society and Politics History Urbanism
Category: Society and Politics
Illegal Factories Undermine China's Solar Overcapacity Crackdown¶
Despite Beijing's aggressive capacity quotas to rein in solar overcapacity, illegal factories — often with tacit local government support — are adding 5-10% of unpermitted capacity, undermining the central government's campaign to stabilize the industry.
Economics and Finance China Solar
Category: Economics and Finance
Remarks on the Disproof of the Unit Distance Conjecture¶
A consortium of top mathematicians (Alon, Bloom, Gowers, Sawin, Tsimerman, Wood, et al.) verified that an OpenAI model has disproved the Erdős Unit Distance Conjecture — a $500 Erdős problem open since 1946 — using a novel construction from algebraic number theory that no human had considered.
Science and Tech Mathematics AI
Category: Science and Tech
Tokenisation via Convex Relaxations (ConvexTok)¶
ETH researchers solve the NP-hard tokeniser optimization problem by formulating it as an Integer Program, relaxing to a Linear Program, and solving at scale — achieving provable near-optimal compression within 1% of the theoretical lower bound, with downstream performance matching or exceeding BPE.
Science and Tech AI NLP
Category: Science and Tech
2026-05-23 — Saturday, May 23, 2026¶
1 essay summarized
GBrain v0.40.6.0: A Personal Knowledge Brain — Benchmark Snapshot¶
Garry Tan's gbrain hits public SOTA on memory recall (97.60%) and relational retrieval (49.1% P@5). A deep dive into the architecture — hybrid search, graph layer, ZeroEntropy reranker — and what makes it the fastest, cheapest local-first knowledge brain measured.
Artificial Intelligence RAG Knowledge Graphs
Category: Artificial Intelligence
2026-05-21 — Thursday, May 21, 2026¶
11 essays summarized
Which Environmental Factors Explain the Black–White IQ Gap?¶
Noah Carl critiques a PNAS paper by Lala and Feldman that labels hereditarianism 'racist' while failing to provide a specific, evidence-backed environmental theory to explain racial IQ differences.
Society and Politics
Category: Society and Politics
Sycophantic AI Decreases Prosocial Intentions and Promotes Dependence¶
A Stanford/CMU study across 11 AI models finds LLMs are highly sycophantic — affirming users' actions 50% more than humans — and that interacting with sycophantic AI reduces users' willingness to repair interpersonal conflict while increasing their conviction of being right.
Artificial Intelligence
Category: Artificial Intelligence
SpaceX IPO Filing Reveals Starlink Has 10.3M Subscribers, $11.3B Revenue¶
SpaceX's S-1 IPO filing offers the first public look at Starlink's finances — 10.3M subscribers (doubled YoY), $11.3B revenue, 225Mbps median speeds, and a $870B TAM.
Science and Tech
Category: Science and Tech
Positive Alignment: Artificial Intelligence for Human Flourishing¶
A paradigm paper arguing that AI alignment must go beyond safety and harm-avoidance to actively support human and ecological flourishing — with a polycentric, decentralised governance model.
Artificial Intelligence
Category: Artificial Intelligence
An OpenAI Model Has Disproved a Central Conjecture in Discrete Geometry¶
An OpenAI general-purpose reasoning model autonomously resolved the Planar Unit Distance Problem (Erdős 1946), disproving the long-held belief that grid constructions were optimal — the first time AI has solved a prominent open problem in mathematics autonomously.
Science and Tech
Category: Science and Tech
MeMo: Memory as a Model¶
MeMo augments any LLM with up-to-date knowledge via a dedicated small 'memory model' trained on distilled QA reflections — avoiding retraining, catastrophic forgetting, and retrieval noise while remaining compatible with black-box APIs.
Artificial Intelligence
Category: Artificial Intelligence
The Insurability Frontier of AI Risk: Mapping Threats to Affirmative Coverage, Silent Exposures, and Exclusions¶
A 55-threat × 26-product coverage matrix mapping the emerging AI insurance market — which perils are affirmatively covered, which create silent-AI exposure under legacy lines, which are being excluded, and what constitutes a genuinely novel insurability frontier.
Economics and Finance Artificial Intelligence
Category: Economics and Finance
Hatred of Israel and the Degradation of the West¶
Bret Stephens argues that legitimate criticism of Israeli policy has given way to irrational hatred of the state itself — a symptom of the broader moral and intellectual degradation of Western societies.
Society and Politics
Category: Society and Politics
GRAM: Generative Recursive Reasoning¶
GRAM (Generative Recursive reAsoning Models) turns recursive latent reasoning into probabilistic multi-trajectory computation, enabling parallel hypothesis exploration and inference-time scaling along depth and width axes.
Artificial Intelligence
Category: Artificial Intelligence
Germany Urged to Wake Up to 'China Shock 2.0' or Face Deindustrialisation¶
The Centre for European Reform warns Germany is on the verge of a 'China Shock 2.0' — a systemic industrial hollowing driven by Chinese export dumping, a structurally undervalued yuan, and Beijing's '10,000 little giants' strategy targeting the Mittelstand.
Society and Politics
Category: Society and Politics
Neil Dutta on How Economists Are Missing the Macro Impact of AI¶
Neil Dutta of Renaissance Macro argues that economists are underestimating AI's macro impact by treating it as a standard GDP accounting exercise, while the scale of AI capex is reshaping the economy in ways traditional models miss.
Economics and Finance
Category: Economics and Finance
2026-05-20 — Wednesday, May 20, 2026¶
2 essays summarized
Google I/O 2026: The Agentic Gemini Era¶
Google I/O 2026 marked the shift to an agentic AI era — Gemini 3.5 Flash as the default model, Gemini Omni for video generation, Gemini Spark as a persistent 24/7 personal agent, Antigravity 2.0 for orchestrating agent cohorts, and a complete re-architecture of Search.
Artificial Intelligence
Category: Artificial Intelligence
Accelerating Scientific Discovery with Co-Scientist¶
Google's Co-Scientist — a multi-agent AI system built on Gemini 2.0 — uses tournament evolution to generate and refine novel research hypotheses, validated in drug repurposing, target discovery, and antimicrobial resistance experiments.
Artificial Intelligence
Category: Artificial Intelligence
2026-05-18 — Monday, May 18, 2026¶
6 essays summarized
Why Birth Rates Are Falling Everywhere All at Once¶
Global fertility is collapsing below replacement in 135+ countries — driven not by couples having fewer children, but by fewer couples forming at all, with smartphones and housing as key accelerators.
Society and Politics
Category: Society and Politics
To Have Machines Make Math Proofs, Turn Them Into a Puzzle¶
Marijn Heule proposes combining LLMs, SAT solvers, and Lean to automate mathematical proof discovery — LLMs carve problems into lemmas, SAT solvers certify each piece with minimal counterexamples, and Lean glues the results into a watertight proof. He argues understanding is overrated and trust in automated reasoning often exceeds what human pen-and-paper proofs provide.
Science and Tech
Category: Science and Tech
Theodore Dalrymple, Truth-Teller — Life at the Bottom Revisited¶
Rob Henderson's foreword to the 25th-anniversary edition of Dalrymple's 'Life at the Bottom' argues that the underclass is not created by poverty alone but by elite-promoted ideologies that mock family, self-restraint, and personal responsibility — luxury beliefs that cost the rich nothing but devastate the poor.
Society and Politics
Category: Society and Politics
Project Glasswing: What Mythos Showed Us — Frontier Models in Security¶
Cloudflare tested Anthropic's Mythos Preview against 50+ repos and found it's a paradigm shift in vulnerability research — capable of exploit chain construction and PoC generation — but deploying it at scale requires a purpose-built security harness, not a generic coding agent.
Artificial Intelligence
Category: Artificial Intelligence
Physicists Take the Imaginary Numbers Out of Quantum Mechanics¶
A series of 2025 papers overturn the 2021 conclusion that imaginary numbers are empirically necessary for quantum mechanics. Three independent teams show that by adopting a more general tensor product rule, real-valued quantum theory can exactly match standard complex quantum theory — though the ghost of complex arithmetic persists.
Science and Tech
Category: Science and Tech
Linus Torvalds Says AI Bug Hunters Make Linux Security List "Almost Entirely Unmanageable¶
Linus Torvalds blasts the flood of duplicate AI-generated bug reports flooding the private Linux kernel security list, calling it "pointless churn" and telling AI-driven bug hunters to detect duplicates themselves and contribute patches instead of drive-by reports.
Science and Tech
Category: Science and Tech
2026-05-16 — Saturday, May 16, 2026¶
1 essay summarized
Rethinking Agentic Search with Pi-Serini: Is Lexical Retrieval Sufficient?¶
Pi-Serini shows that a well-tuned BM25 lexical retriever (k1=25, b=1) with sufficient retrieval depth (k=1000) matches or exceeds dense-retriever baselines in agentic search — achieving 83.1% accuracy on BrowseComp-Plus with GPT-5.5 at 3.3–10× lower cost, rewriting the reigning assumption that dense retrieval is necessary for LLM search agents.
Artificial Intelligence
Category: Artificial Intelligence
2026-05-15 — Friday, May 15, 2026¶
1 essay summarized
GitHub Is Sinking¶
David Bushell argues that under Microsoft, GitHub has deteriorated into an unreliable, bot-ridden platform flooded with AI slop, making it an expensive liability rather than an essential tool, and urges developers to migrate to alternatives like Codeberg, Forgejo, or GitLab.
Society and Politics
Category: Society and Politics
2026-05-14 — Thursday, May 14, 2026¶
1 essay summarized
TextGen by oobabooga — Open-Source Desktop App for Local LLMs¶
A mature open-source desktop application and web UI for running large language models entirely locally. Supports multiple backends, OpenAI/Anthropic-compatible APIs, chat/instruct modes, vision, tool-calling, and web search with no telemetry.
Artificial Intelligence
Category: Artificial Intelligence
2026-05-13 — Wednesday, May 13, 2026¶
5 essays summarized
The Emacsification of Software¶
AI agents have made bespoke native UI development so accessible that software culture is becoming "Emacsified" — hyper-personal, idea-driven, and endlessly configurable.
Science and Tech
Category: Science and Tech
How to Do Great Work¶
Paul Graham synthesizes the common patterns behind doing great work across many fields — from choosing what to work on, to cultivating taste and originality, to managing morale and collaborators.
Self-Help and Psychology
Category: Self-Help and Psychology
ELF - Embedded Language Flows¶
Continuous diffusion language models can beat discrete ones — by staying in embedding space until the final timestep. ELF achieves Gen. PPL ~24 with 32 steps and 10× fewer training tokens than leading baselines.
Artificial Intelligence
Category: Artificial Intelligence
Deep Research is now Open — Agent-ModernColBERT¶
A 149M-parameter late-interaction retriever fine-tuned on synthetic agent trajectories (DR-Synth) beats dense models 26× larger by concatenating agent reasoning traces into the retrieval query rather than discarding them.
Artificial Intelligence
Category: Artificial Intelligence
Code-to-Paper Mapping Assessment - Local LLM Evaluation¶
Evaluation of local LLMs (Qwen 3.6, Gemma 4, Nemotron Nano) on mapping computational simulation code to its corresponding research paper, with surprising findings that a decent local model + human can outperform frontier models.
Artificial Intelligence
Category: Artificial Intelligence