The Time Machine — Same Codebase, Three Eras, One Right Answer
Show me this app written in 2005, 2015, and 2035. Same functionality, different era's idioms. Pick the version that's actually best — and explain why we don't write that way today.
Why this exists
Every era believes its conventions are the correct ones. The 2005 engineer was sure XML and stored procedures were the adult choice. The 2015 engineer was sure microservices and single-page apps were the adult choice. The 2026 engineer is sure of something too — and is equally wrong about half of it.
Reading the same business logic through three lenses surfaces which choices are durable and which are fashion. The function that solves the problem doesn't change. The scaffolding around it does, repeatedly, and not always in the direction of progress.
What you get back
- Three full reimplementations of the same app — 2005 idioms, 2015 idioms, 2035 idioms — each internally consistent with the era's deployment model, language features, and conventional wisdom.
- A side-by-side feature parity check confirming all three versions actually do the same thing.
- A "fashion vs. fundamental" annotation on every major architectural choice in your current codebase.
- The model's pick of the best version, with a justification grounded in what the code is actually trying to accomplish — not in which era it came from.
When to reach for this pattern
Before a refactor — to find out which parts of the existing system are load-bearing and which are inherited habit. While teaching architecture history — three eras of the same code teaches more than any survey course. When evaluating a new framework — to decide whether it's real progress or a rebrand of something you already had in 2009.
Show me what this codebase would look like if it had been written in
2005. Then 2015. Then in 2035. Same functionality. Different era's
idioms, frameworks, deployment model, and conventional wisdom. Pick
the version that's actually best — and tell me why we don't write
that way today. Be specific about which 2026 conventions are
load-bearing and which are fashion.Paste this into Claude, Cursor, or Copilot. Change one thing that matters to you.
What I learned shipping it
- Most 'best practices' are dated. They were answers to constraints that no longer exist.
- Reading your code through three eras exposes which choices are fundamental and which are inherited reflexes.
- The future isn't a straight line forward. It's frequently a return to an older idea wearing new clothes.