News & Digests
29 issues · 29 keepers · 2 tier-5 · 27 tier-4
1 tier-5 · 5 tier-4
The oldest stratum of the archive (2018–2019) is also its most canonical: a tight cluster of foundational crypto essays that still function as reference points years later. Boyapati's store-of-value argument and Taleb's terse Bitcoin notes anchor the investment-and-philosophy side; Dixon's "Why Decentralization Matters" and Kasireddy's Ethereum explainer cover the why and the how; and the NYT's Puerto Rico reporting supplies the era's cautionary counter-narrative. Read together, they map the intellectual scaffolding the crypto debate was built on.
Taleb's short but durable Bitcoin essay ("It may fail but we now know how to do it"), a frequently-referenced touchstone in crypto discourse and a precursor to his later writing on the subject.
The consolidated version of Boyapati's canonical Bitcoin store-of-value essay — one of the most-cited foundational crypto-investment arguments, later expanded into a widely read book.
NYT reporting on the crypto-millionaire migration to Puerto Rico to dodge taxes — a documented and much-debated episode of the 2018 crypto era that remains a reference point for the 'crypto colonialism' critique.
Landmark crypto/web3 essay by Chris Dixon that became a foundational reference for the decentralization argument.
Taleb's durable, frequently-cited statement of his Bitcoin thesis from a canonical author.
Canonical evergreen technical explainer of Ethereum internals, still widely cited as a reference long after publication.
1 tier-5 · 7 tier-4
The densest theme in the archive captures the pandemic in near-real time: the Atlantic's early-warning coverage from the week US lockdowns began, first-person dispatches and epidemiology from Italy, contrasting leadership case studies (Ardern's New Zealand vs. Georgia's reopening), and Ed Yong's canonical "How the Pandemic Will End." The mRNA origin-story interview with Moderna's Noubar Afeyan closes the arc, reframing the era's defining technology as a 33-year overnight success. Together these read as a contemporaneous documentary record of 2020.
Landmark Atlantic early-warning piece from the week US lockdowns began, documenting the preparedness gap as the pandemic hit.
Early, influential Atlantic reporting on the catastrophic US COVID-19 testing failure — a defining policy failure of the pandemic's first weeks.
First-person primary-source dispatch from inside Italy's lockdown weeks before most countries locked down — a durable contemporaneous record of the pandemic's onset.
Durable early demographic/epidemiological explainer of Italy's high COVID-19 fatality rate that served as a warning template for other countries.
Ed Yong's March 2020 Atlantic piece is the canonical, widely-cited early-pandemic essay that correctly framed how COVID-19 would unfold in the US — durable reference reporting.
Durable Atlantic profile of Jacinda Ardern's empathy-driven COVID leadership, a touchstone of the pandemic-era leadership conversation.
Landmark Atlantic reporting on Georgia's early COVID-19 reopening, a defining and widely-cited document of the 2020 pandemic-vs-economy debate.
Landmark long-form Conversations with Tyler interview with Moderna co-founder Noubar Afeyan on how the mRNA vaccine's 'overnight' success was 33 years in the making — durable origin story of a defining COVID-era technology.
0 tier-5 · 4 tier-4
The substantive-science wing of the recent AI coverage: pieces that explain mechanisms rather than chase hype. A 27-minute essay maps biological neurodevelopment onto artificial neural networks; a From-Narrow-to-General-AI piece probes how machines build abstractions from concrete perception; Alberto Romero unpacks OpenAI's research on why language models hallucinate; and Jim the AI Whisperer documents a perverse evaluation bias where AI reviewers reward AI-rewritten papers. Read these to understand what the models are doing, not just how to prompt them.
Lead is a 27-min essay genuinely comparing biological neurodevelopment to artificial neural-network construction — substantive science writing, not a listicle. The slate reinforces it with Francisco Rodrigues on Shannon information theory (art to entropy) and a custom AI-PDF-assistant build. The strongest digest in this batch on intellectual depth, dragged down only by the usual 'made $3,000 with AI' filler.
Featured piece reports that AI reviewers score AI-rewritten papers higher than human-written ones, creating a perverse incentive to hide AI use—a genuinely interesting finding about AI-evaluation bias and academic integrity, the most substantive lead in this batch. Digest also includes Kartscrut on convenience eroding skill and a classic software-architecture-patterns reference.
Unusual lead: a 33-minute essay on how AI builds abstractions from concrete perceptions, a substantive AGI/cognition piece rather than the typical hype. Reinforced by Devansh's inference-scaling deep dive and Linda Caroll's investigative piece on nudify sites making $36M; some macOS-tips and micro-SaaS filler around them.
Lead is Alberto Romero's review of OpenAI's research paper on why LLMs hallucinate (the training/eval incentives that reward confident guessing) — a substantive, well-sourced explainer of a genuinely important mechanism. The digest also carries strong technical secondaries (RAG failures fixed with knowledge graphs, enterprise RAG design).
0 tier-5 · 4 tier-4
The practitioner's shelf: reference-grade material for people actually shipping AI systems. A survey of nine RAG architectures beyond naive retrieval, a guardrails-and-evals piece on making agents reliable, a candid $47K production field report on the real costs of A2A and MCP at scale, and a curated roundup of 20 AI GitHub projects to watch. Where Theme 3 explains the science, this theme is the engineering counterpart.
The strongest technical digest in this batch: lead on use-case-specific validation tests and guardrails for AI agents, joined by Ida Silfverskiold's 'Working with Evals', Mark Riedl's intuition-behind-LLMs explainer, an HNSW vector-DB matching deep dive, and a LangChain knowledge-graph extraction with BAML fuzzy parsing piece. A coherent, production-minded cluster on agent reliability and evaluation worth reading for builders.
Lead is a concrete production-economics writeup on running multi-agent systems at scale ($47K spent), covering the real costs and gotchas of A2A and MCP that tutorials skip — the kind of field report worth reading. The digest otherwise leans on labor-market doom and lifestyle pieces.
Lead is a substantive, example-driven survey of nine RAG architectures beyond naive RAG — a genuinely useful reference for anyone building retrieval systems. Slate reinforces the technical thread with the 8-MCP-servers roundup and a Claude Code piece, though it also carries the heavier psychology/wealth filler. The lead is worth reading and bookmarking.
Lead is a reference-grade roundup of 20 AI GitHub projects to watch, and the slate adds LangChain Deep Agents, a 'CLIs beat MCP for agents' argument, and Claude Code source-leak analysis. Higher-signal than most digests for someone tracking the AI tooling landscape.
0 tier-5 · 5 tier-4
The archive's durable humanities cluster — viral-but-substantive essays that became cultural touchstones or books. Taleb coins rhetorical concepts for spotting bad-faith argument; Gabrielle Blair reframes the abortion debate (later the book *Ejaculate Responsibly*); Bianca Bosker's Atlantic longform shows how scientific consensus is actually contested; Devon Price's "Laziness Does Not Exist" reframes unproductivity as barriers rather than character; and Julie Zhuo distills management craft into ten sketches. Each rewards reading on its own terms, independent of the digest churn around it.
Taleb INCERTO chapter coining durable rhetorical concepts (pedophrasty, bigoteering) still referenced in argumentation debates.
Viral cultural-touchstone essay that later became the book Ejaculate Responsibly; durable reframing of the abortion debate.
Landmark Atlantic longform on geologist Gerta Keller's decades-long fight over whether volcanism, not the Chicxulub asteroid, caused the dinosaur extinction — durable, still-cited science journalism on how scientific consensus is contested.
Anchored by Devon Price's landmark essay 'Laziness Does Not Exist' (357K claps), a widely cited, genuinely substantive psychology piece arguing barriers — not character — explain unproductivity. The slate also surfaces 'RAG is Just Bayesian Inference' and a US-debt-ownership explainer; the strong, durable lead lifts this above the batch norm.
Stands out because the lead is Julie Zhuo's well-regarded 'Average Manager vs. Great Manager' (explained in 10 sketches, 48K claps), a genuinely good management-craft piece from a credible author. Also carries Max Petrusenko's '30 days building an AI dev team' field report and a BitNet.cpp 100B-on-a-laptop item; the rest is Claude-Code/n8n churn.
0 tier-5 · 2 tier-4
Two outliers that resist the AI/crypto/pandemic gravity of the rest of the archive but are durable on their own. Foreign Policy names the US-China contest that became the decade's defining geopolitical thesis; the Conversations with Tyler interview with Nobel laureate Michael Kremer argues for economists building real-world institutions rather than only publishing research.
Foreign Policy framing of the multi-decade US-China contest that became the defining geopolitical thesis of the 2020s.
Substantive Conversations with Tyler interview with Nobel laureate economist Michael Kremer on economists building real-world institutions rather than only research — durable economics content.