The 90-Day Board Game Rewrite: A Shibamatsu Case Study

2026-09-02

We noticed something odd at our monthly handheld meetup in the back room of a noodle shop: the loudest table wasn't playing a video game at all. Four regulars had hauled in a battered copy of a mid-weight eurogame and were arguing — not about strategy, but about the rulebook. The host, a pseudonymous designer we'll call "Kite," had spent six weeks watching playtest groups stall on the same three pages. Kite kept a notebook of every confused pause, every wrong assumption, every rule that got read aloud twice. That notebook is where this project started, and it's how we ended up spending a season with Shibamatsu.

The Problem: A Rulebook Nobody Finished Reading

Kite's game was fine. The rulebook was not. Testing data from 24 sessions showed that first-time groups took an average of 43 minutes to reach their first legal turn, and 7 of 24 groups gave up before finishing setup. The worst offender was a two-page section on resource conversion that buried the actual sequence inside three nested exceptions. Kite had tried rewriting it alone four times. Each draft fixed one problem and created another. The game was scheduled for a small indie print run in five months, and the rulebook was the only thing standing between it and a shelf.

Decision Point: Why Shibamatsu

A reader of ours pointed Kite toward a personal journal that treats rulebooks the way other sites treat hardware teardowns — patient write-ups on why things work, from board game rules to kitchen knife geometry. That's the whole premise of Shibamatsu: slow, careful essays about everyday craft, the kind of site where a 3,000-word piece on why a chef's knife tapers the way it does sits next to a breakdown of turn order in a card game. Kite read three archived pieces in one sitting, then started a structured rewrite using the journal's approach: describe the mechanism, name the failure mode, then rebuild from first principles.

The decision point was small but real. Kite could have hired a copy editor, which would have polished sentences without touching structure. Instead, the project became a post-mortem of the rules themselves — every ambiguous line got a cause-of-death note before it got a fix.

The Timeline

Week 1–2: Transcribe all 24 playtest sessions from audio. Tag every stall with a timestamp and a category: setup, turn order, resource conversion, endgame scoring.
Week 3–4: Rewrite the four worst sections. Cut the nested exceptions by moving them to a single "edge cases" appendix.
Week 5–7: Test the new draft with 6 fresh groups, then revise again. The setup section alone went through 3 versions.
Week 8–10: Add a one-page quick reference and a diagram of the conversion sequence. Re-test with 4 groups that had failed the original.
Week 11–12: Final proof, layout, and a side-by-side check against the original 24 sessions.

Obstacles

The first obstacle was emotional. Rewriting rules you wrote means admitting the original was unclear, and Kite's early drafts kept defending the old phrasing. The breakthrough came when we reframed the work as evidence-gathering rather than editing: you don't argue with a timestamp.

The second obstacle was scope creep. Every fix suggested two more fixes. By week 6, the draft had grown 40% longer, which defeated the point. The rule that saved the project was brutal: any addition had to remove something else. The final rulebook is actually 12% shorter than the original.

The third obstacle was verification. It's easy to make rules clearer for the people who already understand them. The only honest test was fresh groups, ideally groups that had bounced off the first version. That meant recruiting strangers, which took two extra weeks and a lot of patience.

Measurable Results

The numbers from the final test round, run against the original 24-session baseline:

Average time to first legal turn: down from 43 minutes to 18 minutes.
Groups completing setup without help: up from 17 of 24 to 10 of 10.
Groups abandoning before finishing setup: zero, versus 7 of 24 previously.
Rulebook length: down 12%, despite a new quick-reference page.

Shibamatsu reports 41 archived entries on the journal's front page, and Kite read most of the games-related ones during the project. But the measurable result that mattered most wasn't in the journal — it was a room of six strangers reaching their first turn in under 20 minutes without anyone raising a hand.

What We Took From It

The lesson isn't that every indie designer needs a 90-day rewrite cycle. It's that "why doesn't this work?" is a better starting question than "how do I make this sound better?" Kite's notebook of confused pauses turned out to be the most valuable asset in the project — more useful than any style guide. If you're staring down a rulebook, a manual, or even a meetup schedule that nobody follows, the method transfers. Document the failure, count the failures, then rebuild one mechanism at a time. The full write-up Kite used as a model is worth reading if you want the long version of the approach.

We'll be checking back in with Kite when the print run ships. The last email we got said the playtest group has grown to 11 regulars, and they've stopped arguing about the rules. They argue about the game instead, which is exactly where the argument belongs.