Site: a tab listing every recorded option on every screen #50

Open
opened 2026-11-19 21:41:24 +00:00 by kathleen · 1 comment
Owner

A new tab on the site: every step of the threshing, in game order, with every option a rider has been offered there. No probabilities, counts or advice: just the choices as the game words them.

Why. The route model and the site name choices in their own shorthand ("East", "West"). What a rider actually clicks can differ. For example, with squadmates the direction screen offers:

Company Options on screen
1 squadmate Go west with your squadmate · Convince them to head east instead · Leave your squadmate
2 squadmates Go east with your squad · Convince them to head west instead · Leave your squadmates

The squad's preferred direction has been the same in every bot run (West for one squadmate, East for two), so "convince vs go along" is the same choice as East vs West for a given company size. That isn't visible anywhere on the site today. A plain list of the screens makes this kind of thing obvious.

Where the options come from.

  • The bot's screens (data/bot/scenes.json): option text exactly as shown, but only 46 screens and 118 options so far. The bot plays a narrow set of routes, so most bonding encounters are seen only at their first answer.
  • The community route PDF (data/public/pdf_encounters.json): every answer inside each bonding encounter, as the guide words it.
  • The community form (core/spec/form.yaml): answer categories for the early steps (reaction to the first dragon, cadets, fight), sometimes paraphrased.

Proposed.

  • Build step: analysis/screens/build_screens.py writes data/public/screens.json. It merges the three sources into one list per step and keeps the bot's exact wording where the bot has seen the screen.
  • Screens that vary are listed per variant:
    • by company size (the direction screen);
    • by which dragon came first (the reaction screen);
    • by encounter.
  • Site: a "Screens" tab, a collapsible list by stage, with no numbers on it.
  • Rebuilt by python -m collection rebuild, with a --check in CI like the other builders.
  • Only option text: no scene narration or other game prose.

Open questions.

  • Mark where each option was recorded (bot / PDF / form)? It isn't probability data, but it says which options are confirmed on screen and which come from the guide.
  • One list per screen, or also show where each option leads (the next screen), as a tree?

Related: #26 (the route's labels for the squad direction; the bot's experiment list will add the two "convince" options, never played so far).

A new tab on the site: every step of the threshing, in game order, with every option a rider has been offered there. No probabilities, counts or advice: just the choices as the game words them. **Why.** The route model and the site name choices in their own shorthand ("East", "West"). What a rider actually clicks can differ. For example, with squadmates the direction screen offers: | Company | Options on screen | |---|---| | 1 squadmate | Go west with your squadmate · Convince them to head east instead · Leave your squadmate | | 2 squadmates | Go east with your squad · Convince them to head west instead · Leave your squadmates | The squad's preferred direction has been the same in every bot run (West for one squadmate, East for two), so "convince vs go along" is the same choice as East vs West for a given company size. That isn't visible anywhere on the site today. A plain list of the screens makes this kind of thing obvious. **Where the options come from.** - **The bot's screens** (`data/bot/scenes.json`): option text exactly as shown, but only 46 screens and 118 options so far. The bot plays a narrow set of routes, so most bonding encounters are seen only at their first answer. - **The community route PDF** (`data/public/pdf_encounters.json`): every answer inside each bonding encounter, as the guide words it. - **The community form** (`core/spec/form.yaml`): answer categories for the early steps (reaction to the first dragon, cadets, fight), sometimes paraphrased. **Proposed.** - Build step: `analysis/screens/build_screens.py` writes `data/public/screens.json`. It merges the three sources into one list per step and keeps the bot's exact wording where the bot has seen the screen. - Screens that vary are listed per variant: - by company size (the direction screen); - by which dragon came first (the reaction screen); - by encounter. - Site: a "Screens" tab, a collapsible list by stage, with no numbers on it. - Rebuilt by `python -m collection rebuild`, with a `--check` in CI like the other builders. - Only option text: no scene narration or other game prose. **Open questions.** - [ ] Mark where each option was recorded (bot / PDF / form)? It isn't probability data, but it says which options are confirmed on screen and which come from the guide. - [ ] One list per screen, or also show where each option leads (the next screen), as a tree? Related: #26 (the route's labels for the squad direction; the bot's experiment list will add the two "convince" options, never played so far).
kathleen added the site label 2026-11-19 21:41:24 +00:00
Author
Owner

Decided:

  • No source marks. Options are listed without saying whether the bot, the PDF or the form recorded them.
  • As a tree. Each option shows the screen it leads to, so the tab reads as the game's branching from the first screen to every ending.
Decided: - **No source marks.** Options are listed without saying whether the bot, the PDF or the form recorded them. - **As a tree.** Each option shows the screen it leads to, so the tab reads as the game's branching from the first screen to every ending.
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: DragonKind/threshing-atlas#50