graysonarts / graysonarts/aboutface

Put λ in the operator config and tune it by eye

Open
#15 0 comments 0 reactions 0 assignees View on GitHub
ready-for-human
Dominant language
Rust
Stars
0
Forks
0
PR merge metrics
No merged PRs in 30d

Description

## What to build

λ weights the movement penalty in the Assignment cost (ADR-0003). It is a tuning dial for how sticky the layout feels, and the ADR expects it to be set by eye during installation rather than derived.

Two parts. The plumbing: λ moves out of the code and into `booth.toml`, reported at startup with everything else the self-check reports. The tuning: sit in front of the wall, press the Shutter, and find the value where the crowd reorganizes without anyone teleporting.

Needs a human. "Reads as the crowd reorganizing" is not a number, and the honest way to find λ is to watch the wall.

Worth knowing whether λ can be changed without a restart. If tuning means rebuilding between every value, the tuning will not actually happen.

## Acceptance criteria

- [ ] λ is read from `booth.toml` and reported at startup
- [ ] An absent or nonsensical λ fails at startup with the reason, the way missing models already do
- [ ] The tuned value is committed as the default, with a sentence on what it looks like too low and too high
- [ ] Total movement per Re-solve is reported at the tuned value, so a later change can be compared against it

## Blocked by

- #14

Contributor guide

No contributing guide indexed for this repository

Research direction

Read ADR-0003 and trace the existing startup self-check that reports model configuration; then locate the booth.toml loading path and the Re-solve movement calculation. Done means λ is validated and reported at startup, its tuned default and low/high visual behavior are documented, and the tuned total movement is recorded.

Written by the indexing model from the issue text.

Assessment

Tech stack
rust
Domain
computer-vision
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Quiet
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.