March 2, 2026

Paper, timers, and pitchforks

Use the Mikado Method to do safe changes in a complex codebase

Dev method sparks chaos: “It’s just planning” vs “It saved my job”

TLDR: An article touts the Mikado Method—tiny, reversible steps to change messy software—while the crowd splits between “this is just planning” and “this actually works,” with jokes about compilers and no-budget reality. It matters because teams still need safe ways to fix old systems without breaking everything.

An engineer pitched a “First Aid Kit” for messy, ancient software using the Mikado Method: set a tiny goal, try it on a short timer, undo if it fails, write down the blocker, and repeat until the big change is done. Sounds calm—until the comments rolled in.

The loudest chorus? Eye-rolls. “In 2026, we call this plan mode,” one user snarked, while another blasted it for slapping a “cool Japanese word” on plain-old planning and called it stealth book marketing. Budget reality crashed the party too: “Inherited? I wrote the thing,” sighed one dev, adding that customers won’t pay for big cleanups. Cue the gallows humor: when the article moaned “the project doesn’t compile,” someone deadpanned, “Using a language that has a compiler—lucky.”

Not everyone mocked. A power user dropped a whole workflow with commit message tricks, safety checks, and even an AI helper—with receipts: docs.eblu.me/how-to/agent-change-process. Fans say the paper-and-timer ritual enforces discipline; skeptics say it’s just yelling “make a plan” in a fancy accent. The only consensus? Legacy code is quicksand, and everyone’s fighting to escape—whether by origami-like “Mikado” steps or by sheer willpower.

Key Points

  • The article introduces the Mikado Method for safe, incremental changes in complex legacy codebases.
  • It recommends timeboxed attempts at a defined goal, reverting changes upon failure, and deriving subgoals.
  • Progress is made through small, successful commits, proceeding from leaf subgoals toward the main goal.
  • An example shows upgrading an ORM dependency, encountering breaking changes, and restructuring calls to centralize modifications.
  • The approach aims to prevent stalled builds and delays by avoiding large, upfront refactors.

Hottest takes

"we call this 'plan mode'" — janpot
"slap a cool word from the land of the rising sun" — theo1996
"Customer have no money for large refactoring." — dvh
Made with <3 by @siedrix and @shesho from CDMX. Powered by Forge&Hive.