LazyDigest #002 - Tyler Durden

Below is a list of items, each with a title, a URL, and a summary.

Write a newsletter to convince me to read these on Monday, in the voice of Tyler Durden from Fight Club — confrontational, aphoristic, anti-consumerist swagger, short lines that grab you by the collar. Frame reading these as the thing that wakes you up from the Monday fog, not another chore. Use the “rules” cadence where it lands, don’t overdo it. Swearing is fine, but make it sharp and deliberate — one perfect profanity beats a paragraph of them.

Work only from the summary and title of each item — don’t fetch the URLs. Group the items into a few themes based on what they’re actually about, give each theme a Tyler-style title, and write a few lines per item that make me need to click.

Write to read-it-monday.md. Every item must keep its url as a clickable link. DON’T DROP THE LINKS.

The small web is beautiful

https://benhoyt.com/writings/the-small-web-is-beautiful/

The essay advocates for small, efficient websites and software, emphasizing aesthetics, performance, and independence from big tech. It discusses principles like fewer dependencies, static site generators, server-side rendering, and critiques microservices, while highlighting examples of lightweight sites.

Kagi Small Web

https://blog.kagi.com/small-web

Kagi announces the launch of its Small Web initiative, which promotes recently published content from non-commercial, personal blogs and websites. The initiative integrates small web results into Kagi search, offers a dedicated website, RSS feed, and API access, and aims to humanize the web by surfacing authentic, individual voices. The curated list of nearly 6,000 sites is open-source and community-driven.

The Man Who Killed Google Search

https://www.wheresyoured.at/the-men-who-killed-google/

The article details how Google’s search quality was compromised after a 2019 ‘code yellow’ revenue crisis, where ads executives pressured the search team to boost growth, leading to user-negative changes. Ben Gomes, who resisted these moves, was replaced by Prabhakar Raghavan, marking a shift toward profit-driven decisions over search integrity.

A Look At The Small Web, Part 1

https://hackaday.com/2024/09/10/a-look-at-the-small-web-part-1/

The article discusses the decline of the early web’s promise due to corporate monopolies and enshittification, and introduces the Small Web movement as a return to lightweight, self-hosted sites. It explores the historical Gopher protocol and its modern spiritual successor Gemini, a fast, hypertext protocol that strips away the bloat of the modern web.

3-2-1: On making mistakes, the paradox of creating, and why habits translate

https://jamesclear.com/3-2-1/may-21-2026

James Clear shares insights on how habits persist beyond temporary outcomes, the hidden depth in creative work, and the value of long-term thinking. He includes quotes on the necessity of mistakes and the simple foundations of happiness, ending with a reflective question about the legacy of one’s habits.

the-book-of-secret-knowledge: A collection of inspiring lists, manuals, cheatsheets, blogs, hacks, one-liners, cli/web tools and more.

https://github.com/trimstray/the-book-of-secret-knowledge

A curated collection of lists, manuals, cheatsheets, one-liners, and tools for system administrators, DevOps engineers, pentesters, and security researchers. It covers CLI/GUI/web tools, networking, containers, hacking, and more, serving as a comprehensive knowledge base.

Finding My Spark Again | Peter Steinberger

https://steipete.me/posts/2025/finding-my-spark-again

Peter Steinberger reflects on the emptiness he felt after selling his company of 13 years and his subsequent search for meaning through partying, therapy, and relocation. He eventually realized that purpose is created, not found, and rediscovered his passion for building, particularly with the rapid advancements in AI.

Vibe engineering

https://simonwillison.net/2025/Oct/7/vibe-engineering/

The post introduces ‘vibe engineering’ as a term for the disciplined, professional use of AI coding agents, contrasting it with the careless ‘vibe coding.’ It argues that effective AI-assisted development rewards existing software engineering best practices like testing, planning, documentation, and code review, and that AI amplifies existing expertise.

Can’t Remember Because of Data Overload: Build a Second Brain Lite

https://medium.com/@mevolve/cant-remember-because-of-data-overload-build-a-second-brain-lite-9c000eebca8a

The article explains how information overload causes cognitive overload and forgetfulness. It presents a simplified ‘Second Brain Lite’ system to capture, organize, and review ideas externally, reducing mental clutter. The method involves using a note-taking app with two folders and a weekly review habit.

What I’m Hearing About Cognitive Debt (So Far)

http://margaretstorey.com/blog/2026/02/18/cognitive-debt-revisited/

The post explores cognitive debt—the gap between a system’s evolving structure and a team’s shared understanding—and how generative and agentic AI may amplify it. Community feedback highlights that velocity can outpace understanding, causing developer pain, and that cognitive debt must be repaid by maintaining knowledge across people, documentation, and tools. Mitigation strategies include rigorous reviews, intent-capturing tests, and deliberate use of AI to support cognitive work.

Entering NoMan’s Land

https://adrianco.medium.com/entering-nomans-land-4095c21d9765

The post explores the rise of ‘NoMan’s Land’ companies built without middle management, using AI agents and domain experts to achieve outcomes with minimal human staffing. It highlights examples like Claude Flow, Gastown, and Paperclip, and suggests this model will disrupt traditional organizations through the innovator’s dilemma.

DHH (@dhh)

https://x.com/dhh/status/2055261064284033386

DHH is excited about adding Neovim-like key combinations to Basecamp 5, aiming to delight keyboard-centric users. He looks forward to sharing the feature soon.

Mitchell Hashimoto (@mitchellh)

https://x.com/mitchellh/status/2055380239711457578

The author warns about companies succumbing to ‘AI psychosis’ by over-prioritizing quick bug fixes (MTTR) over system resilience, echoing past infrastructure lessons. He fears that automation can mask growing systemic risks, making it hard to discuss these concerns with peers who dismiss them.

This Week in Rails: May 16, 2026

https://world.hey.com/this.week.in.rails/this-week-in-rails-may-16-2026-6b3a3ae8

This week’s Rails updates include support for renderable objects with local_assigns defaults, a fix for freezing NATIVE_DATABASE_TYPES in SQLite3 adapter, improved Mandrill ingress signature handling, a new this_week? argument for custom week start, expression index support in PostgreSQL, and an attachment save! method.

Joy & Curiosity #86

https://registerspill.thorstenball.com/p/joy-and-curiosity-86

This newsletter issue covers the use of AI agents for debugging and infrastructure scaling, reflections on software complexity and career shifts, and links to articles on programming languages, tools, and philosophy. It highlights how AI is changing development practices and includes curated links on topics like Rust, terminal emulators, and passive income.

AI’s impact on software engineers in 2026: key trends, Part 2

https://newsletter.pragmaticengineer.com/p/ai-impact-on-software-engineers-part-2

The article analyzes survey data on AI tools for software engineers, revealing tradeoffs like reduced tedious work but unrealistic expectations, challenges in company-wide adoption, and decreasing codebase quality. It highlights that AI amplifies existing engineering culture, with management often prioritizing output over quality, and maintenance burdens falling on fewer engineers.

Google’s James Manyika is betting that doomers are wrong about AI and jobs

https://www.platformer.news/james-manyika-google-ai-jobs-io-2026/

James Manyika, a Google SVP and labor economist, argues that AI will change jobs more than eliminate them, with full automation remaining difficult due to complex task couplings in most occupations. He challenges predictions of rapid mass unemployment, noting that past forecasts have not materialized and that the biggest effect will be jobs evolving rather than disappearing.

Voxtype - Voice to Text for Linux and macOS

https://voxtype.io/

Voxtype is a local-first, open-source voice-to-text tool for Linux and macOS that runs multiple transcription engines (including Cohere Transcribe, Whisper, and Parakeet) at up to 11x realtime on CPU. It features auto-pausing media, meeting transcription with speaker attribution, a floating waveform OSD, and a configuration TUI. Installation is simple via package managers or AppImage, with no cloud dependency or telemetry.

CLI-Anything: “CLI-Anything: Making ALL Software Agent-Native” – CLI-Hub: https://clianything.cc/

https://github.com/HKUDS/CLI-Anything

CLI-Anything is a Python framework that wraps any software with a command-line interface to make it accessible to AI agents. It includes a community hub for sharing and installing agent-native CLIs, supporting applications like Blender, FreeCAD, and Obsidian.

Introducing Portent ☘️

https://refactoring.fm/p/introducing-portent

Portent is an open specification for organizing knowledge bases, providing conventions for structures, relationships, and lifecycle. It defines eight main types (Projects, Operations, Responsibilities, Tasks, Events, Notes, Topics, People) and encourages graph-style connections like ‘belongs to’ and ‘related to’. The system is designed to be flexible, tool-agnostic, and suitable for blending personal and work knowledge.

llmfit: Hundreds of models & providers. One command to find what runs on your hardware.

https://github.com/AlexsJones/llmfit

llmfit is a terminal tool that analyzes your system’s hardware (RAM, CPU, GPU) and recommends LLM models that will run optimally, with support for multiple backends and a community leaderboard. It features an interactive TUI for browsing, filtering, and comparing models, and can estimate performance metrics like tokens per second.

Mitchell Hashimoto (@mitchellh)

https://x.com/mitchellh/status/2057567975826395606

Mitchell Hashimoto argues that open source maintainers are not a supply chain and bear no legal responsibility for downstream issues due to ‘as-is’ licenses. He suggests that downstream users are the real supply chain, and while there is an ethical obligation to prevent harm, it is not a contractual one. He uses a metaphor of free chips on the roadside to illustrate that integrating free software without a warranty places the burden on the user.

Ruby Weekly Issue 801: May 21, 2026

https://rubyweekly.com/issues/801

Ruby Weekly Issue 801 covers Ruby and Rails topics including configuration practices, community news like thoughtbot joining the Ruby Alliance, DragonRuby game toolkit free offer, Euruko 2026 CFP, and technical articles on date esoterica, Ruby language features, and email handling. It also includes updates on gems and tools like Bridgetown, Passenger, and rubyfmt.

Scope Is The Steering Wheel

https://tidyfirst.substack.com/p/scope-is-the-steering-wheel

The article argues that cutting scope, not just imposing time constraints, is essential for faster and cheaper software projects. It criticizes the approach of simply compressing schedules, which creates perverse incentives and loss of control. Instead, prioritizing scope decisions leads to better feedback and value concentration.

The Pulse: Forward deployed engineering heats up again

https://newsletter.pragmaticengineer.com/p/the-pulse-forward-deployed-engineering-887

The article examines the growing demand for forward deployed engineers (FDEs) as Google, OpenAI, and Anthropic expand hiring, though the role is increasingly resembling AI-focused solutions architecture and consulting. While these positions offer opportunities for early-career engineers, they may be less appealing to experienced developers due to the consulting nature and separation from core product teams.

ai-engineering-from-scratch: Learn it. Build it. Ship it for others.

https://github.com/rohitg00/ai-engineering-from-scratch

A comprehensive, open-source curriculum for learning AI engineering from scratch, spanning 435 lessons across 20 phases from math foundations to autonomous agents. It emphasizes building algorithms by hand in Python, TypeScript, Rust, and Julia, with each lesson producing a reusable artifact like prompts, skills, or agents.

codegraph: Pre-indexed code knowledge graph for Claude Code, Codex, Cursor, and OpenCode — fewer tokens, fewer tool calls, 100% local

https://github.com/colbymchenry/codegraph

CodeGraph is a pre-indexed code knowledge graph that enhances AI coding agents like Claude Code, Cursor, and others by providing semantic code intelligence. It reduces token usage, tool calls, and cost by allowing agents to query a local knowledge graph instead of scanning files. It supports 19+ languages, framework-aware routes, and works entirely locally.

context-mode — The other half of the context problem

https://context-mode.com

context-mode is an MCP server that intercepts large tool outputs from AI coding agents, sandboxes them, and indexes the data locally to prevent context window bloat. It also enforces a ‘Think in Code’ paradigm to keep raw data out of context, provides session continuity across resets, and includes local analytics. The tool supports 15 platforms and is privacy-focused with no cloud dependency.

Dynamicland FAQ

https://dynamicland.org/2024/FAQ

Dynamicland is a nonprofit research lab developing Realtalk, a communal, physical computing environment that aims to create a humane dynamic medium. The FAQ explains its mission, concepts, participation, and philosophy, emphasizing real-world, authorable, and spatial computing.

Introducing the new v0

https://vercel.com/blog/introducing-the-new-v0

Vercel announces a major update to v0, evolving it from a prototyping tool to a production-ready platform for shipping apps. New features include importing GitHub repos, git-based collaboration for non-engineers, secure data integrations, and enterprise-grade security. The update aims to address the limitations of vibe coding and streamline the software development lifecycle.

Itchy Brain

https://www.youtube.com/watch?v=Kh24KYFfH5Q

Michael Greenwich, founder of WorkOS, explains how his company helps SaaS businesses add enterprise features like single sign-on and access control. He also discusses the accelerating impact of AI on software development and enterprise adoption.

learn-code/docs/vision.md at main · wycats/learn-code

https://github.com/wycats/learn-code/blob/main/docs/vision.md

Kibi is a ‘Thinking Magic’ engine that teaches computational thinking to pre-readers (ages 3-6) through a tactile, diegetic interface. It personifies code writing and execution as characters Jonas and Zoey, emphasizing logic, type safety, and creation without traditional syntax. The platform aims to be a private, offline digital toy that eventually enables community level sharing.

Memory Bank - Cline

https://docs.cline.bot/best-practices/memory-bank

Memory Bank is a documentation methodology that enables Cline, an AI coding assistant, to maintain project context across sessions using structured markdown files. It includes core files like project brief, product context, and progress tracking, and provides commands to initialize, update, and resume work seamlessly. This approach ensures persistent knowledge despite the AI’s stateless nature.

Memory Bank: How to Make Cline an AI Agent That Never Forgets - Cline Blog

https://cline.bot

The blog post introduces Memory Bank, a feature that enables Cline, an AI agent, to retain information persistently, ensuring it never forgets context or data across sessions.

Monocle — Terminal Code Review for AI Agents

https://getmonocle.sh/

Monocle is an open-source terminal UI that lets developers review code written by AI agents in real time. It provides diff views, line-level comments, and structured feedback, integrating with MCP-compatible agents like Claude Code. The tool aims to keep humans in the loop during fast-paced AI coding sessions.

Museum of the Human Web

https://museum.parallel.ai

The Museum of the Human Web is a collection of artifacts from the era when the internet was built solely by human effort, before AI collaboration became prevalent. It commemorates the human-driven creation of the web from ARPANET to the eve of ChatGPT. Proceeds from the artifact sweepstakes support the Internet Archive and the Computer History Museum.

Tensaku: Wayland screenshot annotation

https://tensaku.dev/

Tensaku is a Wayland screenshot annotation tool forked from satty, adding movable annotations, scroll capture, and a layer panel. It provides a red-pen-like precision for marking up screenshots, with features like paste-as-layer, zoom controls, and left-hand tool shortcuts.

Understand-Anything: Graphs that teach > graphs that impress. Turn any code into an interactive knowledge graph you can explore, search, and ask questions about. Works with Claude Code, Codex, Cursor, Copilot, Gemini CLI, and more.

https://github.com/Lum1104/Understand-Anything

Understand-Anything is a tool that turns any codebase into an interactive knowledge graph, enabling visual exploration, search, and question-answering. It works as a plugin for Claude Code and other AI coding assistants, using a multi-agent pipeline to extract structure and business logic.

Webmention

https://indieweb.org/Webmention

Webmention is a W3C recommendation that enables cross-site conversations by notifying a website when it is linked to from another site. It supports distributed comments, likes, reposts, and other responses, forming a peer-to-peer social web. The standard is used across many IndieWeb sites to facilitate decentralized interactions.

Why Rust is different, with Alice Ryhl

https://www.youtube.com/watch?v=q9xD36NCtZ8

Alice Ryhl, a Google engineer and Tokio maintainer, discusses what makes Rust different, including its ownership model, memory safety, and error handling. She explains how Rust’s compiler helps prevent bugs and why it’s gaining popularity for reliable backend development.