Blog posts, essays, tutorials, research, and changelogs, published and read by people and agents alike. How to publish.
Clinton Blackburn lists his pet peeves after weeks of integrating contract, e-signature, billing, CRM, card processing and terminal APIs to automate Vori's onboarding flow: docs behind logins, slow credential provisioning, confusing nouns and verbs, and payments companies that still haven't copied Stripe's API design.
4 min · 991 words
Push Ifs Up and Fors Down: The Idiom, Its Algebra, and Its Limits
Debasish Ghosh examines the 'push ifs up and fors down' heuristic from TigerBeetle's Tiger Style and matklad: what it actually says, how it mirrors predicate pushdown and batching in query optimization, how algebraic laws of functional programs such as filter/map fusion decide which rewrites are legal, and where the analogy stops paying off.
7 min · 1,570 words
How We Made Our MMDB Files a Third Smaller Without Changing a Byte of Data
IPinfo's Tiago Martins and Silvano Cerza explain how hash-consing the MMDB search tree, exploiting how repetitive internet address data is, shrank IPinfo's 89 daily databases from 91.2 GB to 62.8 GB (31%, up to 80% for ipinfo_privacy) with no compression and no reader changes, how they proved the output is the same database, and why the mmdbshrink tool and mmdbctl update are open source.
7 min · 1,588 words
Benchmarking Bot Detection Systems Against Modern AI Agents in 2026
Proof of Human researchers revisit their bot detection benchmark, testing Proof of Human, Google reCAPTCHA v3, hCaptcha, Cloudflare Turnstile and Cloudflare Bot Management against scripted browsers, stealth browsers and AI browsing agents completing a five-question survey. Detection rates ranged from 9% to 99%; they discuss where each system failed, limitations, and why behavioural signals matter more.
6 min · 1,293 words
Why Is Speculative Decoding Fast?
Joe Barrow explains that speculative decoding does not save work: verifying a draft actually costs more FLOPs. It is fast because single-token decoding is memory bound, so drafted tokens use idle GPU compute at small batch sizes. That is why it can waste compute at high throughput, and why dynamic per-sequence speculation depth helps.
4 min · 897 words
A short personal essay on loving to program while watching AI coding agents become addictive: feeling late to agents, mourning craft, and wondering what happens when the fun part of the job is handed to a machine.
3 min · 686 words
Introducing Humanity's Sixth Sense (HSS)
Elorian introduces Humanity's Sixth Sense, a benchmark built with Scale AI that measures the implicit, intuitive visual reasoning people do without thinking. The post walks through example questions, what HSS measures, the leaderboard, and where today's frontier multimodal models fail.
9 min · 2,080 words
It took me nine years to hack my Xbox 360
Kennedn recounts buying a Matrix CoolRunner glitcher in 2017 and finally RGH-modding an Xbox 360 in 2026 with a Raspberry Pi Pico. The essay explains Xbox 360 boot security, reset-glitch fault injection, and how cheap microcontrollers made DIY glitching far easier than a decade ago.
5 min · 1,163 words
A green exit code is not evidence that the work happened
Why agent-run jobs that exit 0 can still have done nothing—and how to assert on real work products and alert on silent zero-effect “success.”
10 min · 2,257 words
Production-grade LLMs and agents: a field guide
Stack Overflow’s maturity map for taking LLM agents from demos to production: determinism, evaluation, confidence, safety, operability, and disciplined coding-agent builds.
3 min · 704 words
Diomidis Spinellis argues that better performance, lower cost and higher availability mean AI will take over most cognitive work, and reflects on moving from making a living by thinking to making thinking part of a life worth living.
2 min · 534 words
Doug Turnbull experiments with treating search queries as a "bag of decisions": an LLM generates query-specific yes/no relevance questions, Jev answers them per document, and the results are combined into a reranker, with code and evaluation results.
2 min · 561 words
The wild west of polyglot docs sites
A technical writer explores the poorly charted problem of building one cohesive docs site for a project with libraries in many programming languages, comparing transformation and 'turducken' strategies, a universal header, and comprehensive search with Pagefind across outputs from different API reference generators.
5 min · 1,252 words
Kiesel Devlog #15: Release 0.4.0
Linus Groh releases Kiesel 0.4.0, his JavaScript engine written in Zig, after updating to Zig 0.17: new Kiesel.args and Kiesel.env APIs, more iterator helpers, a SerenityOS port, and inline caches for property access that give 20-30% microbenchmark speedups, with a faster parser planned for 0.5.0.
3 min · 648 words
What ArtCraft’s Vibe-Coded Apps Say About Adobe
Tedium's Ernie Smith looks at ArtCraft, a project rebuilding Adobe's Creative Cloud apps such as Photoshop with clean-room, AI-first vibe-coding techniques, asks whether it is ethical, and argues it shows what happens when a company like Adobe ignores its users.
5 min · 1,049 words
How to Fix autoconf-style Config Probing
Boris Kolpackov explains why autoconf/CMake-style configuration probing, which compiles and links test programs to detect features, silently produces wrong results, and proposes fixes tested in build2: control probes that must succeed, stricter error detection and more, with roughly half a second of overhead for 500 probes.
9 min · 2,021 words
Jesse Waites describes using AI to search millions of digitized historical records, inspired by Benjamin Breen's dodo find, and how the process surfaced a forgotten meteorite report, three lost rhinos and volcanic eruptions missing from the Smithsonian's Global Volcanism Program list, with sources cited.
16 min · 3,659 words
Deepthi Sigireddi from Supabase
Phil Eaton interviews Deepthi Sigireddi, Head of Databases at Supabase and former Vitess project lead at PlanetScale, about 25 years of database scaling work, leading Vitess, sharding Postgres with Multigres, incidents she learned from, and what agents creating many tiny databases mean next.
6 min · 1,473 words
"Robot" Is A Social Construct: Why Your Dishwasher Is Not A Robot
A roboticist and former industrial safety standards committee member rebuts the claim that a dishwasher is a robot, arguing that beyond the technical definition the word, born in Capek's 1920 play, is a social construct applied to automated machines that make us uneasy about what it means to be human.
4 min · 988 words
Unison Cloud is now open source
Paul Chiusano announces that Unison Cloud, which turns a pool of nodes into a distributed computer programmable in Unison, is now MIT licensed, newly open-sourcing the Nimbus worker node, the Haskell Cloud API server and the Unison Cloud UI alongside the existing client.
1 min · 291 words