My dearest reader,
Happy Fortnight Week to you and yours! I hope this week has been kind to you. Either way, tomorrow is Friday, which is enough reason to be happy.
”If generation gets easier, judgment must get stricter. Otherwise, you’re not engineering anymore.”
Chad Fowler
Relocating Rigour: The traditional architectural rigour, which relied on upfront planning, is being replaced by a more fluid, iterative Phoenix model, where the ability and skill to rapidly regenerate, refactor, and discard code is more valuable than trying to make it perfect from the start. LLMS have lowered the engineering cost of generating code, and the rigour of the developer now lies in their ability to orchestrate these cycles and maintain integrity across a system that is constantly rebuilt. Chad emphasises a shift from treating software as a monument to treating it as a living organism that maintains its form through continuous renewal.
“The engineers who thrive in this environment will be the ones who relocate discipline rather than abandon it. They'll treat generation as a capability that demands more precision in specification, not less. They'll build evaluation systems that are harder to fool than the ones they replaced. They'll refuse the temptation to mistake velocity for progress.”
Chad Fowler
The malleable computer: AI is fulfilling the open-source promise of software freedom by making complex codebases accessible and customizable by everyone. AI is making it possible for users to customise their forks to suit their personal needs.
“As always, the future is already here; it's just not evenly distributed.”
David Heinemeier Hansson
AWS Distinguished Eng: Learnings From 3000 Incidents And How Engineering Is Changing: This is an interview with Marc Brooker, a Distinguished Engineer at AWS, who shares that being on call and analysing over 3,000 postmortems provided insights into distributed systems. Real-world failures offer a unique opportunity to learn and identify improvements. He also shared nuances such as caching in distributed systems and the importance of writing as a tool for mental clarity. He also mentions that engineers at all levels should embrace AI as a lever for innovation, and the ability to build faster and experiment with the new tools will redefine software engineering as a more creative and highly leveraged profession.
Claude Design: Anthropic’s new canvas for designers, PMs, and developers: Claude just released a new Canvas for design targeted to non-technical stakeholders to iterate and prototype faster through natural language.
The Hidden Cost of AI: We tend to default to the most powerful AI models without considering the financial and performance overhead of each API call. Fulop advocates a more deliberate approach to selecting models based on the specific complexity of a task. The hidden cost is not just the literal price tag, but also the missed opportunity for smarter, more efficient architectural decisions that balance capability with sustainability.
What Happens After AI Writes The Code: There is an assumption gap when AI agents generate code from repository files without access to the actual runtime environment, leading to assumption bugs that are caught later in CI or staging. This article suggests using tools like Mirrord to enable AI agents to run code locally while connected to clusters, so they can observe failures and adjust their assumptions.
Your Friends Are Hiding Their Best Ideas From You: People used to get validation for their ideas from their friends rather than doing the hard work of turning them into reality. AI has taken the role by offering uncritical praise and low-fidelity prototypes that leave the creator satisfied without ever launching a business. This article reminds us that an idea is only a spark, the true value is in execution, which requires moving past addictive validation into the messy work of building.
“There are limits to feats of skill, beyond which lie the realms of nonsense. Everything is quite difficult enough as it is, and what is simple actually comes hardest.”
Hans Werner Henze
The AI stack I Developed To Help Me Context Switch: The article outlines a personal productivity system designed to manage the high cognitive overhead of juggling multiple projects. The stack relies on three interconnected tools:
Gemini Meeting Notes for hands-free transcription and summaries, which allows him to stay fully present during calls
Obsidian as a local markdown knowledge base that provides a consistent organisational structure for all project contexts
Claude Projects (linked via an MCP plugin) to maintain intelligent, area-specific memory without starting every chat from scratch.
By automating information capture and centralising context in a standardised way, you can switch between diverse tasks more fluidly while reducing the mental exhaustion of constant context switching.
“What’s In It For Me” Architecture: Technically brilliant architectural design is worthless if the architect cannot convince stakeholders to actually implement it. The article shifts the focus from technical to diplomatic, noting that success depends on identifying the actual decision-makers and tailoring arguments to their specific motivations. The role of an architect is as much about social and political navigation as it is about software design. It requires an understanding of the diverse human contexts that exist within a single project.
The word of the day is futz. Futz means to waste time, often by fussing over minor things instead of tackling something important.
Example in a sentence:
Instead of starting his chores, he decided to futz with his collection of old cards.
I am making my grand departure into the unknown.
Take care of yourself!
Until the next fortnight, my treasured reader, go forth, and may the odds be ever in your favour! 👏 🤖 ✊ ☠️ 🏹 🪖
Do you enjoy the latest issues of my newsletter? Buy me a piping cup of hot chocolate today!
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.