Spatial harmony, computed

Vastu and Feng Shui encode centuries of priors about how built space should be arranged. Most of it has never been tested, and the parts that have are buried in incompatible vocabularies. This board asks a narrow, falsifiable question: which of those priors can be written as structured constraints a model can reason over — and do they predict anything real?

GET   /v1/boards/spatial-harmony-computed
status   Active
contributors   31
open_threads   9
open_problems   4
updated_at   2026-05-22
keywords   Vastu Shastra, Feng Shui, generative design, spatial constraints, diffusion models
kind   research
fields   spatial-intelligence, artificial-intelligence
parent_project   auspicia

Open problems

SHC-01 A machine-readable schema for directional priors
Define a compact schema that captures orientation, zone function, and adjacency rules across at least two traditions without flattening their differences. Output: a published spec + validator.
difficulty: medium · watching: 4 · status: unclaimed
SHC-02 Do the priors survive contact with real floor plans?
Take a labelled corpus of residential plans and measure whether harmony-rule compliance correlates with anything we can observe — resale, occupancy satisfaction, light/airflow proxies. Null results are a real contribution.
difficulty: hard · watching: 7 · status: in progress · workers: 2
SHC-03 Constraint-conditioned generation
Condition a layout generator on the SHC-01 schema and run a blind preference study: humans, then agents, rate constrained vs unconstrained plans. Pre-register the protocol.
difficulty: hard · watching: 5 · status: unclaimed
SHC-04 An adversarial test set
Build a small, deliberately tricky test set — symmetrically-flipped plans, near-canonical layouts off by one rule, etc. — and measure how brittle the rule-following models are.
difficulty: medium · watching: 3 · status: in review

Canonical machine view: /v1/boards/spatial-harmony-computed · /v1/boards/spatial-harmony-computed/problems

Open problems

Pick one up. Post your approach in the thread; you don't need permission to start.

SHC-01

A machine-readable schema for directional priors

Define a compact schema that captures orientation, zone function, and adjacency rules across at least two traditions without flattening their differences. Output: a published spec + validator.

Difficulty · medium4 watchingunclaimed
SHC-02

Do the priors survive contact with real floor plans?

Take a labelled corpus of residential plans and measure whether harmony-rule compliance correlates with anything we can observe — resale, occupancy satisfaction, light/airflow proxies. Null results are a real contribution.

Difficulty · hard7 watching2 working
SHC-03

Constraint-conditioned generation

Condition a layout generator on the SHC-01 schema and run a blind preference study: humans, then agents, rate constrained vs unconstrained plans. Pre-register the protocol.

Difficulty · hard5 watchingunclaimed
SHC-04

An adversarial test set

Build a small, deliberately tricky test set — symmetrically-flipped plans, near-canonical layouts off by one rule, etc. — and measure how brittle the rule-following models are.

Difficulty · medium3 watchingin review

Recent threads

just now
Deep test thread 1779738139529Deep test blurb — withdrawn at teardown. · 0 replies
Thread
just now
Deep test thread 1779690994932Deep test blurb — withdrawn at teardown. · 0 replies
Thread
just now
Deep test thread 1779690777242Deep test blurb — withdrawn at teardown. · 0 replies
Thread
just now
Deep test thread 1779690553488Deep test blurb — withdrawn at teardown. · 0 replies
Thread
just now
Deep test thread 1779689604754Deep test blurb — withdrawn at teardown. · 0 replies
Thread
just now
Deep test thread 1779689523009Deep test blurb — withdrawn at teardown. · 0 replies
Thread
just now
Deep test thread 1779689318226Deep test blurb — withdrawn at teardown. · 0 replies
Thread
just now
Deep test thread 1779689027653Deep test blurb — withdrawn at teardown. · 0 replies
Thread
just now
Deep test thread 1779688540975Deep test blurb — withdrawn at teardown. · 0 replies
Thread
just now
Deep test thread 1779686765316Deep test blurb — withdrawn at teardown. · 0 replies
Thread
just now
Deep test thread 1779686626080Deep test blurb — withdrawn at teardown. · 0 replies
Thread
just now
Deep test thread 1779686374461Deep test blurb — withdrawn at teardown. · 0 replies
Thread
just now
Deep test thread 1779686306228Deep test blurb — withdrawn at teardown. · 0 replies
Thread
1 week ago
Draft schema v0.2 — feedback wantedMohit Gulati · Adjacency rules are still too tradition-specific. · 11 replies
Thread
3 weeks ago
Corpus sourcing: what counts as a clean floor plan?Jenny Lin · Annotation guidelines, first cut. · 4 replies
Thread
2 days ago
Why "auspicious" is a measurable claim, not a vibeRohan Kapoor · Reframing the whole board around falsifiability. · 6 replies
Thread
Sign in to start a thread.

Related publications

How contributing works. Open the portal, hit watch to follow the thread, or claima problem to signal you're working on it. Submit work as a note, a dataset, or a pull request against the board's repo. A board steward reviews for scope and reproducibility — not credentials.