This is the personal blog of Ben Houston (bio here), a 3D computer graphics software developer. For a quick overview of the work I am proud of, see my portfolio. I've worked on projects like Web3DSurvey, Three.js, Deadline, Krakatoa, and Exocortex VFX plugins. I'm also an entrepreneur, having founded Land of Assets, Threekit, Clara.io, Exocortex, and Frantic Films Software.
Traditional specifications failed because they required exhaustive detail. With LLMs, we can now embrace "Sufficient Specification" - providing just enough intent to guide generation without drowning in details. Learn how iterative specification specificity and generator patterns make intent-based programming practical for the first time.
Recursive generation allows intent-based programming toolkits to ingest their own outputs and build entire applications from minimal specifications. This essay explores how recursive intent turns code generation into a compiler-like pipeline that scales small prompts into structured systems.
Charles Simonyi's Intentional Programming was a prescient attempt to shift software from languages to abstractions. In the age of AI and intent-based toolkits, many of his core insights are newly relevant. This essay explores the parallels and differences between Simonyi's vision and today's declarative, AI-driven development models.
In traditional development, refactoring is a costly, manual effort to improve code structure without changing behavior. But in intent-based programming, refactors are reframed: they emerge naturally from specification shifts and cost little more than regeneration. This essay explores how we can reconceptualize refactoring in a world where code is ephemeral and intent is the source of truth.
Code coverage tells you what code has been touched. Spec coverage tells you what intent has been verified. In the era of intent-based programming, it's time to stop measuring tests by lines of code and start measuring them by meaning.
How TanStack Router runs beforeLoad and loader functions, what data each one can see, and where slow route work belongs.
Optimize your README.md files to empower both human developers and AI coding assistants alike.
My home network now runs on 3Gbps fiber, 10G switching, and NVMe storage. The useful lessons came from the mistakes: too few Ethernet drops, the wrong 10G transceivers, and underestimating WiFi coverage.
AI coding assistants are excellent at generating tests with high coverage. But there is often a critical problem: they are testing that code behaves as written, not as intended. This "Test Theater" creates a dangerous illusion of quality while providing almost no real protection against bugs.
Learn how to develop 3D web applications that work reliably across all devices using real-world browser capability data from Web3DSurvey.com with a practical two-tier development strategy.
Exploring the simplest possible implementation of an agentic coder using just 200 lines of TypeScript, demonstrating foundational concepts clearly and concisely.
How to build an autonomous agentic coder that leverages advanced LLMs to transform software development. A talk given at the Ottawa Forward JS meet-up.
