Writing short: cutting 317 scenes to 202
We cut the first season of Hayat Dosyası, infancy and the first two years, from 317 scene records to 202. The average scene went from 41 words to 21. Here is what we cut and what we could not.
Yasper Games
On 9 October we wrote ourselves a sentence: the game must not bore anyone, and everyone should be able to play it. The measurement we had that day was this: playing the life of a child born in 1980 to a working-class family on the real engine, the first season (ages 0 to 2) came to 105 screens, about 27 minutes of reading. Up to the age of forty-seven it was 727 screens, about three and a half hours. About ten scenes per chapter; the average scene text was 41 words.
A well-written life does not mean a life that makes you read a lot.
The numbers
We take the first season (Birth) as the example. The same thirty-one chapters, but:
| measure | before | after |
|---|---|---|
| scene records | 317 | 202 |
| scenes per chapter (avg.) | 10.2 | 6.5 |
| scene text (avg. words) | 41.2 | 20.8 |
| longest scene text | 79 words | 25 words |
"Scene records" here means every scene row in the season file, cards and interludes included. The number of screens a player sees in one life will differ, depending on household and decade variants.
The other seasons moved the same way. Across all eight seasons, scene records went from 2,309 to 1,490. The average scene text in each season went from 33-44 words to 19-22.
The rules
We did not cut by "make it a bit shorter". We wrote a rule table and rewrote every chapter against it:
- 5-6 scenes per chapter (7 at most).
- Scene text of 25 words at most: two or three short sentences.
- A dialogue scene of 4 lines at most, a montage of 5 items at most.
- Option text of 8 words at most; 3 options per decision (4 at most).
- Outcome text of 15 words at most: one sentence, ending on an image.
We set rules for the language too: short sentences, everyday Turkish, no literary ornament; one idea per sentence; be concrete, one object, one sound, one movement. And one ban: draft sentences used to cover every possibility with "either ... or ...". Now each household type and each decade gets its own exact sentence.
In the same round we made another decision: the director's language in scene text ("Camera… Cut.", "the scene goes dark") goes away. The same moment is told in plain sentences.
An example: the first night
The first chapter is "The First Night". Before it had 11 scenes, averaging 48 words, the longest 100. Now it has 6 scenes, averaging 18.5 words, the longest 20. Outcome lines went from 24 words on average to 11. The "either ... or" draft sentence went from 6 to 0.
The mistakes we caught while cutting turned out to be the real work:
- In the first scene a fluorescent light was flickering. In a home birth there is no fluorescent light; the scene now says a lamp and a stove.
- The grandmother was waiting at the door of the hospital delivery room on the first morning. She cannot wait there; the scene moved to noon.
- A "Window" scene described things a newborn cannot see. In a scene written from the baby's eyes that was wrong; it was cut.
Writing short is not only counting words. It means testing every sentence again: is this right in this context?
The trace of a cut scene
This is the hardest part. When you cut a scene you may cut not only the scene but the trace it left (a flag, a seed, a memory marker). If another scene reads that trace, that scene now works wrongly.
Our rule: if the trace of a cut scene is read somewhere else, it does not disappear; it moves to one of the remaining options or is written openly in the proposal. We check from a map which trace is read where. A decision scene none of whose options is read later is the first candidate to cut; or one of its options gets tied to a meaningful trace.
Consistency comes from the same context. A scene must be right in every context that can lead to it: household type, decade of birth, gender, and earlier choices inside the chapter. A person who does not exist in this life, has died or has left does not speak.
The cost
Shortening has a cost. When chapters got shorter, their contributions to the counters changed too, which can shift how season endings and final thresholds are distributed. So re-running the bot measurement after each round of cutting is one of our open items; we wrote about how it works in Balancing a life simulation with bots.
There is also a loss: atmosphere scenes sometimes go for the sake of brevity. We decided that when a chapter has a single emotional "breathing" scene, it stays, shortened; but the rule also means we lose some of the atmosphere scenes. We accepted that knowingly.
