Engineering Reads — 2026-07-12#

The Big Idea#

Simple, text-driven abstractions—whether small language models or plain-text presentation frameworks—are quietly replacing complex, manual workflows to drastically reduce cognitive burden. Engineers are increasingly using low-overhead tools to solve high-friction problems, favoring lightweight automation over brittle procedural code or tedious manual curation.

Deep Reads#

My Macstock X Markdown Presentation · Brett Terpstra The core claim here is that technical presentations can be effectively built and delivered using pure text workflows, treating slide decks as code rather than design documents. By combining Markdown with Reveal.js and Multiplex, the author created a system where the audience can follow along on their own devices. This technical mechanism removes the friction of traditional WYSIWYG presentation software while enabling viewers to interactively bookmark slides, click links, and copy code blocks in real time. While the author notes that some specific formatting for Reveal.js is required, the underlying presentation source remains entirely readable as plain text. Engineers who prefer text-based toolchains or want to make their technical talks highly accessible and interactive for attendees should study this setup.

Simon Willison — 2026-07-12#

Highlight#

Simon’s thoughts on “Directly Responsible Individuals” (DRIs) provides a crucial human-centric framework for evaluating the integration of LLM-powered agents into organizations. By emphasizing that accountability is an exclusively human trait, he grounds the rapid advancement of AI tooling in practical management ethics.

Posts#

Directly Responsible Individuals (DRI) · Source Simon traces the concept of a “Directly Responsible Individual”—the person ultimately accountable for a project’s outcome—to its Apple origins via the GitLab handbook. He applies this to modern LLM-powered agents, arguing that AI should never hold DRI status within an organization because machines cannot take accountability. Highlighting a classic 1979 IBM slide, he reiterates that a computer must never make a management decision.

Engineering Reads — 2026-07-13#

The Big Idea#

As AI models take over the mechanical generation of syntax, the core bottleneck of software engineering is shifting from writing code to rigorously specifying architecture, intent, and acceptance criteria. The highest-leverage engineering skill is no longer “managing by method” (reviewing line-by-line execution) but “managing by objective”—defining the exact unit of work and building the validation harnesses required to trust the machine’s output.

Deep Reads#

Fragments: July 13 · Martin Fowler · Source Fowler unpacks the recent Thoughtworks retreat, surfacing a critical transition in how we build with LLMs: the rise of “Harness Engineering” to manage an agent’s context and attention. The underlying debate across the industry isn’t really about AI capabilities, but about defining the boundaries of autonomous work and how humans verify it. Fowler notes a shift toward using computational sensors, property-based testing, and formal methods to validate agent outputs, recognizing that we must manage these systems by objective rather than by method. He also touches on the economics and strategy of self-hosting models for data sovereignty, noting that smaller, finely-tuned local models often require less reasoning overhead for domain-specific tasks. This is essential reading for technical leaders trying to figure out how to structure teams, verify outputs, and maintain systemic trust in a world of agentic programming.

Engineering Reads — Week of 2026-06-24 to 2026-07-02#

Week in Review#

This week’s reading circles a central tension in modern engineering: managing the boundary between complex systems and the interfaces we build to tame them. Whether we are embedding local AI agents to maintain data sovereignty or structurally funding paradigm shifts through top-down mandates, the underlying debate is about where to place the friction. The consensus is clear: we must engineer systems that preserve flow and autonomy without obscuring the foundational reality of our tools and languages.

Engineering Reads — Week of 2026-03-28 to 2026-04-03#

Week in Review#

The industry is undergoing a structural shift from authoring syntax to orchestrating and verifying system state. As probabilistic AI agents commoditize raw code generation, the defining engineering challenge has become building the rigorous deterministic harnesses—and maintaining the strict personal accountability—required to safely control these systems in production.

Must-Read Posts#

tar: a slop-free alternative to rsync · Drew DeVault Stringing together fundamental Unix utilities often provides a more predictable mental model than complex, dedicated tools. DeVault argues for migrating directories using a simple tar pipeline over SSH, trading the bandwidth efficiency of rsync’s delta calculations for total cognitive simplicity around path resolution. Engineers tired of wrestling with finicky trailing-slash rules should read this for a refreshing return to composable Unix fundamentals.

Simon Willison — Week of 2026-03-30 to 2026-04-03#

Highlight of the Week#

This week highlighted a monumental shift in the open-source security landscape, marking the sudden end of “AI slop” security reports and the arrival of a tsunami of high-quality, AI-generated vulnerability discoveries. High-profile maintainers of the Linux kernel, cURL, and HAPROXY are reporting an overwhelming influx of legitimate bugs found by AI agents, fundamentally altering the economics of exploit development and forcing open-source projects to rapidly adapt to a massive increase in valid bug reports.

Engineering Reads — Week of 2026-04-02 to 2026-04-10#

Week in Review#

This week’s reading reflects a fundamental inflection point: raw LLM intelligence is no longer the bottleneck in software development. Instead, the industry is pivoting toward the hard systems engineering required to constrain probabilistic models—whether through strict data ledgers, living specifications, or formal verification harnesses. The dominant debate centers on how we preserve architectural taste, mechanical sympathy, and system ethics as the mechanical act of writing code becomes increasingly commoditized.

Simon Willison — Week of 2026-04-04 to 2026-04-10#

Highlight of the Week#

Anthropic’s decision to delay the general release of their highly capable Claude Mythos model under “Project Glasswing” marks a significant turning point in the AI industry. The move underscores a massive shift in frontier model capabilities, as models evolve from generating text to autonomously chaining multiple minor vulnerabilities into sophisticated exploits, requiring a new level of security safeguards before release.

Engineering Reads — Week of 2026-04-08 to 2026-04-16#

Week in Review#

This week’s reading is dominated by the tension between raw, AI-driven generation and the enduring necessity of classical engineering discipline. As AI commoditizes rote code generation, the defining characteristics of engineering are migrating from writing syntax to exercising architectural taste, writing clear specifications, and deliberately bounding probabilistic systems with human constraints. The consensus is clear: creating output is increasingly trivial, but owning the execution mechanics and maintaining systemic intuition requires a conscious, hands-on imperative.

Simon Willison — Week of 2026-04-11 to 2026-04-17#

Highlight of the Week#

This week’s most striking revelation came from Simon’s infamous “pelican riding a bicycle” SVG generation benchmark, where a 21GB quantized local model (Qwen3.6-35B-A3B) unexpectedly outperformed Anthropic’s brand-new Claude Opus 4.7 flagship. Running locally on a MacBook Pro via LM Studio, Qwen generated a better bicycle frame and even won a secret unicycle backup test, leading Simon to conclude that his joke benchmark’s long-standing correlation with general model utility has finally broken down.