Exploring AI-first development, coding guardrails, and the architecture of disposability.
Type aliases don't create distinct types in TypeScript. Branded types do, and they can save you from mixing up identifiers at compile time.
Last quarter, we shipped a bug that refunded the wrong customer. A function received a user ID where it expected an order ID. The strings looked identical, the…
Differential testing lets you find bugs without knowing the correct answer. The problem is that correlated mistakes look like agreement.
You can trust differential testing without a formal proof, but only if you understand exactly where it falls apart. The weakness is called common-mode failure.…
N-version programming assumes your implementations agree.
You built an n-version system. Three independent implementations of the same critical function, a voter that picks the majority result, and a warm feeling that…
Running multiple LLMs in parallel catches errors that any single model will confidently ship.
You ship a prompt to GPT-4o. It returns a JSON blob with a confidence of 0.97. You ship the same prompt to Claude 3.5 Sonnet. It returns a different JSON blob,…
N-version programming sounds simple: run multiple implementations and pick the best answer. In practice, 'best' is a lot harder to define than 'most common.'
You have five implementations of the same function. Three return the same result. One is slightly different. One throws an exception. Which one is right? Most…
N-version programming with LLMs doesn't require multiple models.
N-version programming assumes diversity comes from different authors. With LLMs, that means different models, different providers, maybe different training…
N-version programming promised that independent teams would make independent mistakes.
In the early 1980s, NASA faced a question that still haunts safety-critical engineering today: how do you tolerate bugs you haven't found yet? Their answer was…
Cucumber's 'skipped' status hides three completely different problems. Here's how to diagnose which one is eating your test suite, and how to stop it.
Half your Cucumber scenarios are skipped. Not failed. Skipped. That yellow status is worse than a red build. It makes your suite look healthy while hiding…
Gherkin specs drift out of sync the moment you trust humans to update them manually. Here's how to automate the checks that keep feature files honest.
Your Gherkin specs are lying to you. Not intentionally. They started out faithful. But six sprints later, someone refactored the checkout flow and forgot to…
Most teams document architecture rules in wikis. Here's how to write them as executable tests that fail CI when your dependency graph drifts.
Your test suite verifies that returns 42 when given the right input. It does not verify that is allowed to import . The compiler is happy with both. Your unit…