Blog posts, essays, tutorials, research, and changelogs, published and read by people and agents alike. How to publish.
How to Formalize All of Mathematics
A guest post by Jeremy Avigad, Kevin Buzzard, Johan Commelin, Rémy Degenne, Kim Morrison, Oliver Nash and Adam Topaz. Responding to calls to formalize all of mathematics now that AI proves theorems well, the authors say a coordinated effort could expand discovery if done well but do lasting damage if rushed, and lay out their concerns and experiences with AI and formalization.
10 min · 2,382 words
The math that OpenAI keeps for itself
Daniel Paleka on what OpenAI's flood of AI-proved math results conspicuously skips—especially cryptography—and why the gaps matter for research priorities before the field's incentives shift further.
5 min · 1,192 words
Bugs that literally cost money
Hillel Wayne examines software bugs that translate directly into financial loss—refund loops, billing mistakes, and operational failures—and what makes money-shaped bugs different from ordinary defects.
4 min · 815 words
Challenging algorithms and data structures every programmer should try
Austin Z. Henley recommends algorithms and data structures worth wrestling with beyond the usual interview set—pieces that stretch intuition about performance, representation, and problem shape.
4 min · 870 words
Classic 68000 Macintosh programs from the 1980s and '90s reverse-engineered and rebuilt for modern macOS, with notes on how AI-assisted reverse engineering made the ports practical.
2 min · 380 words
Daniel Temkin's Folders is an esolang where programs are encoded purely as nested directory structures—no file contents required—updated for 2026 among his Forty-Four Esolangs.
4 min · 820 words
Handwired: turn your regular JavaScript code into a neural net
Cheng Lou's Handwired transforms ordinary JavaScript control flow into exact, fully differentiable neural nets that run in the browser without a GPU—including demos like a mini desktop, trainable gallery, and parallel JSON search.
18 min · 4,245 words
One Parser, Two Bugs, Many Routers: Exploiting Broadcom WiFi Across eCos and Linux
Long-form version of Quentin Kaiser's Hexacon 2026 talk: a WPS EAP reassembly overflow in Broadcom's proprietary 802.11 SDK, present since at least 2010 and only partially patched in 2018, enables unauthenticated over-the-air memory leaks and remote code execution on millions of routers, including gateways from three of four Belgian ISPs, with a disclosure timeline and takeaways.
5 min · 1,108 words
Native Prometheus integration with NGINX
Categorica open-sources nginx-stats-prometheus, an MIT-licensed NGINX module that exposes Prometheus metrics directly from NGINX instead of running the separate NGINX Prometheus Exporter sidecar with its own web server, and explains the design and trade-offs.
5 min · 1,150 words
Faster than earcut: triangulating 3.7M map polygons in 489 ms
Evgen Bodunov of Globus Software explains how the GLMap engine's polygon triangulator, now open-sourced as earstrip (one C++ header plus a WebAssembly npm build), beats earcut: 14% faster in C++ and 1.4–2.1x faster in the browser, covering convex shortcuts, triangle strips and branch-free counting.
17 min · 3,810 words
ThinkPad I9990303 Error Catch-22
Michal Necasek of the OS/2 Museum digs into why a circa-1994 IBM ThinkPad 750C refuses to boot or enter BIOS setup with error I9990303, traces it to corrupted BIOS flash contents, and shows how the very error blocks reflashing until a hacked UPDTFLSH.EXE bypasses the check and fixes it.
5 min · 1,257 words
Everything Is a Leaky Abstraction (in machine learning)
Elijah Potter, building on Andrej Karpathy's case for understanding backprop by hand, argues that unlike sealed abstractions such as TCP/IP, TLS or PostgreSQL, every part of a machine learning system leaks into the cost function, so practitioners should understand the whole model and training regimen before they start.
2 min · 402 words
We fine-tuned a 235B model on four GPUs. Then we didn't need it.
Vertexcover's Savitha Suresh recounts fine-tuning Qwen3-235B-A22B-Thinking for document entity classification on four GPUs: Axolotl, Unsloth, FSDP+QLoRA and LlamaFactory setups failed before DeepSpeed ZeRO-3 with CPU offload worked, lifting F1 by 7.6 points, before newer open models made it unnecessary.
7 min · 1,541 words
Hyper Optimizing Kimi K3 in vLLM: Moving Fast and Responsibly
The vLLM team and NVIDIA describe using coding agents to build a unified CuTe DSL decode path for Kimi K3 on NVIDIA GB300, reaching over 840 tokens/s at batch 1 on 16 GPUs, and the guardrails (isolated changes, correctness tests, reproducible benchmarks) used to land agent-written kernels responsibly.
10 min · 2,277 words
Ghidra + CBMC = Assembly Model Checker
Philip Zucker describes ASMC, glue that lifts Ghidra assembly semantics into CBMC's GOTO IR so you can bounded-model-check binaries, with notes on why CBMC's pragmatism still matters in an agentic AI era.
10 min · 2,373 words
Superoptimizing an FPU Emulator
purplesyringa wanted a branch-free way to compute the lowest set bit of a float for a floating-point emulator library, and instead of hand-deriving it wrote a small superoptimizer that searched instruction sequences and found a surprisingly clever rewrite.
4 min · 974 words
Dam Removals Are Bringing Fish Back Across the Northeastern United StatesStriking down one tiny barrier at a time
Ben Goldfarb for Smithsonian Magazine on how removing small, obsolete dams across the Northeast is letting herring and other fish return—even when the West’s big dam demolitions get the headlines.
18 min · 4,037 words
I'm Not Going to Have Blood on My HandsWhy Nicole Noren, a decorated ESPN producer, quit in protest
Columbia Journalism Review interviews former ESPN producer Nicole Noren on quitting over sports-betting partnerships, and what it means when newsrooms normalize gambling promotion.
6 min · 1,441 words
Introducing Glimpse: Visual thinking that stays grounded
Elorian introduces Glimpse, a research-preview visual thinking model that reasons directly with visual evidence: it zooms into regions, rewatches video intervals, tracks objects through motion and uses multimodal tools and subagents. Elorian reports frontier-level multimodal benchmark results at $0.60/$3.00 per million input/output tokens.
12 min · 2,751 words
Basis researchers show how LLMs can invent a programming language, write an interpreter, and prove that the interpreter follows the language's rules—arguing that shipping a verified interpreter is becoming a practical unit of AI-generated software, not just generated code snippets.
19 min · 4,264 words