Dynamic Project States — Brick Cylinder (Toy Model)

A course-by-course build where each course can be laid cw or ccw. The branching choices are rendered as “possible trajectories”.
First try · Make two choices, then read the history
  1. Leave the fresh-page defaults: 10 courses, 24 bricks/course and Manual policy. If you have already started, press Reset.
  2. Press Lay next course ↻ cw, then Step course.
  3. Press Lay next course ↺ ccw, then Step course.

Read the result: two rows are complete, and the trajectory string is ↻↺. Choosing a direction places the first brick; stepping finishes that course. The tree highlights this history among all unrestricted direction histories—it is not a count of futures remaining after your choices.

Then try a policy: reset, select Alternate cw/ccw, and step courses without choosing each direction yourself. The rule chooses; the resulting sequence records what happened.

Model and limits below · Next: add curing and a shared hoist in #13. The pause slider controls presentation time; this primer does not model physical curing or resource conflicts.

Visualisation

Unwrapped cylinder “blueprint”: rows are courses (bottom → top), columns are bricks around the circle.
booting… cw: 0 ccw: 0
—
Tips: Space play/pause · → step brick · Shift+→ step course.

Controls


70ms
0ms
A deterministic policy selects a direction at each course start. Random mode instead samples a direction using the displayed probability.

Tangent choices at the current state

At the start of a course, the action fibre is {cw, ccw}. In manual mode you must choose.
—

Trajectory

Top: blueprint progress bar. Bottom: all unrestricted course-direction histories (depth-limited), not a count conditioned on the selected policy or current prefix.
—
8

Analytics

What this model lets you inspect

The enquiry asks what a project state must retain to generate possible futures. This primer isolates direction choice, policy and history; the other essays add different dynamics.

1
Work and its history
The unwrapped cylinder fills brick by brick, then course by course. The completed rows show work done; the direction string records how it was done.
2
Choices and policy
At a course start, the action set is {cw, ccw}. Manual mode asks you to choose; a deterministic policy supplies that choice. This discrete fibre is a set, not a tangent vector space.
3
Possible trajectories
The depth-limited tree displays unrestricted direction histories and highlights the realised prefix. It does not recompute a count of futures still available under the selected policy.
!
The boundary of this primer
One subsystem places courses. The pause controls presentation time, not physical curing. For represented curing rules see essay 8; for shared hoists, timer state and coupled brick-ring dynamics see essay 13. These are separate models, not features enabled by this primer.

Restatement of the toy model

State and dynamics (macro)

S_min = ℕ × {cw,ccw}
I = {cw,ccw}
O = ℕ
read(n,dir) = n
update((n,dir), choice) = (n+1, choice)

This is the “course-by-course” system. The animation tracks additional micro-state (brick index, pause timer) for the replay. The macro model commits a whole course per step; its pause is presentation time, not physical curing. The trajectory prefix is also stored to show history; (n,dir) alone does not reconstruct it.

Tangent-bundle intuition (discrete)

At each course-start state, the “tangent fibre” is the set of allowable next directions. In this base toy it’s always {cw, ccw}. A deterministic policy such as always-cw or alternate is a section selecting a direction. Manual mode supplies choices as you go. Random mode is a stochastic rule; its sampled run is a trajectory, not a deterministic section defined on all states.

Lens / open-system packaging (informal)

The system exposes an output (current course count) and accepts an input (direction choice). We can wire it to other systems by composing on the input/output interface.