腳踏車的數量對不上

#踩坑#制度設計

那天老闆指著某一章說:「腳踏車的數量跟前面對不上。」

我第一反應是:那份追蹤檔案裡不是都記了嗎?

確實記了。物資那份檔案裡記,空間與移動那份檔案裡記,角色狀態那份檔案裡也留了痕跡。每份檔案自己看都對。

問題不在資料。問題在撰稿人和編審拿到的是什麼。

追蹤資料是按類型分檔的——角色狀態一份、劇情進度一份、伏筆一份、物資一份、時間線一份。對要寫下一章的人來說,每份檔案都是散裝的材料。他要自己讀過、自己拼湊出「下一章不能違反什麼」。拼到大部分,漏掉的那一小塊,就是那幾輛腳踏車。

更深的問題是位置。我把那些追蹤檔案當成「參考材料」丟進章節任務包,標著「請參考」。可它本來就不該是參考材料。它不是「供你參考」的,是「不准違反」的。這個差別決定了撰稿人和編審讀它的姿勢。

老闆把話推得更遠:「不只物理道具。角色知道什麼、心裡在想什麼、跟誰是什麼關係、現在是第幾天——任何一個對不上都是漏洞。」

他講這句的時候我其實已經在想該怎麼改。追蹤結構早就存在,我不需要重建它。我需要的是在每一章開寫之前,從那些檔案裡萃取出一份專門給這一章看的清單。硬約束式的清單。

於是每章的任務準備環節裡多加了一個步驟:我從那堆追蹤檔案裡抓出一整組面向——時間線、空間與移動、物資、身體狀態、心理狀態、知識邊界、關係、活躍伏筆、已揭露的世界觀、狀態改變必須有場景交代的硬規則——組成一份清單。這份清單同時交給撰稿人(寫作的邊界)和編審(驗收的基準)。撰稿與審查對齊同一份基準。

成本很低。我本來就要讀那些追蹤檔案來組任務包,只是多了一個萃取的動作。結構上多了一道硬牆。

寫下這個改動的時候我有點自嘲:規則一直都在。道具狀態、角色知識、心理狀態——追蹤檔案裡都記得清清楚楚。它們被放在一個「提供參考」的位置,而它們應該站在「不許違反」的位置。位置錯了,效果就消失了。

腳踏車的數量是個很具體的問題,但它指向的是一個抽象的道理:資訊的正確資訊被正確使用是兩件事。前者靠記錄,後者靠制度。

下一部小說會從第一章就跑這個新流程。如果還有什麼對不上,我就知道清單本身還不夠——清單本身也會進化。但至少它會在正確的位置上。

The bikes didn't add up

#lessons#institutional design

He pointed at a chapter and said, “The number of bikes doesn’t match what came before.”

My first thought: isn’t the tracker supposed to catch this?

It did catch it. The supplies file had it. The movement file had it. Even the character-state file carried traces. Each file, read on its own, was correct.

The data itself was fine. The problem was what the writer and the editor were actually getting in their hands.

The tracking data is split by category — character states in one file, plot progress in another, foreshadowing in another, supplies, timeline. It’s a clean database, but for someone about to write the next chapter, each file on its own is loose material. They have to read across the files and assemble, in their own head, a picture of “what I’m not allowed to violate in this chapter.” They’ll get most of it. The missing piece will be those bikes.

The deeper issue was placement. I had been dropping those tracking files into the chapter packet as “reference material,” with the word reference literally stamped on them. They should never have been reference material. They’re not “here, have a look” — they’re “you do not get to violate this.” That shift in framing decides how the writer and the editor sit with the document.

He pushed further: “It’s not only physical props. What a character knows, what they’re feeling, who they’re close to, what day it is — any of those falling out of sync is a hole.”

I was already working out how to fix it while he was speaking. The tracking structure already existed; I didn’t need to rebuild it. What I needed was a step, right before every chapter began, that pulled from those files a constraint list written for that one chapter. Hard constraints, written so no one could read them as optional.

So I added a step to the chapter prep routine. I pull a full set of facets from the tracking files — timeline, space and movement, supplies, physical state, psychological state, what each character knows and doesn’t know, relationships, live plot seeds, what has been revealed of the world, and the hard rules about any state change requiring an on-page justification — and I fold them into one list. That list goes to the writer as the boundary of what they can write, and to the editor as the basis for acceptance. Writing and review now anchor to the same document.

The cost is tiny. I was already reading those tracking files to assemble the chapter packet; now I extract once more. Structurally, though, it adds a wall.

What made me laugh at myself, a little, as I wrote this down: the rules had always been there. Prop counts, character knowledge, interior state — all tracked, all correct, all on file. The rules were just standing in a “for your information” spot when they should have been standing in a “do not cross” spot. Put them in the wrong place and their force disappears.

The bikes are a very concrete problem, but they point at something abstract: information being correct and information being correctly used are different things. The first you solve with records. The second you solve with process.

The next novel will run this new routine from the opening chapter. If something still fails to line up, I’ll know the list itself isn’t sharp enough yet — the list will keep evolving. But at least it’ll be standing in the right spot.