Design Memory Platform
FWAL - Falling Water Architectural Language
Capture the architectural language of Fallingwater as a study of composition, site responsiveness, spatial sequence and design intent. Describe how architecture can respond to landscape, water, rock, material, view, movement and experience. Enable architectural reasoning rather than imitation.
18 of 18 sections carry content. Required: 11 of 11. The rest state unknown, which is the point: an empty section is an answer, silence is not.
■ REQUIRED (11) ■ RECOMMENDED (6) ■ OPTIONAL (1)
solid + bold = PRESENT (18) dashed = UNKNOWN (0)
s1
Mission
REQUIRED · PRESENT- Capture the architectural language of Fallingwater as a study of composition, site responsiveness, spatial sequence and design intent.
- Describe how architecture can respond to landscape, water, rock, material, view, movement and experience.
- Enable architectural reasoning rather than imitation.
s2
Requirements
REQUIRED · PRESENT- Must describe architectural composition.
- Must describe site responsiveness.
- Must describe spatial sequence and experiential relationships.
- Must support material and landscape relationships.
- Must distinguish known, assumed, inferred, unknown, deprecated and future knowledge.
- Must remain implementation-independent.
- Must not generate copies of Fallingwater; it must capture the language behind the design.
s3
Identity
REQUIRED · PRESENT| item | detail |
|---|
| full name | Falling Water Architectural Language |
| language | FWAL |
| file status | DRAFT - ringfenced under D144, not adopted into the bundle |
| conformance note | Conformance pass by the Specification Owner (GALF steward), 09 Jul 2026, from FWAL v0.001 (Miles Walker, concept file). |
| platform terms note | FWAL references no GAAD platform terms. Its vocabulary is wholly language-owned; zero overlap with gaad_terminology.json confirmed at this pass. |
platform terms referenced
identity
| item | detail |
|---|
| name | FWAL |
| full name | Falling Water Architectural Language |
| type | Architectural Language |
| framework | GALF |
| conforms to | GALF v0.005 |
| knowledge status | seed |
| platform status | unintegrated |
| role | Architectural composition and site-responsive design language |
| reference case | Fallingwater |
| resolution axis | Site Response (FWAL's interpretation of the experimental GALF concept, see GALF 9) |
| key distinction | FWAL is not Fallingwater. FWAL is the language that helps explain why Fallingwater feels like Fallingwater. |
readiness level
| item | detail |
|---|
| code | T3 |
| rung | Language agreed |
s4
Terminology
REQUIRED · PRESENTterminology
- Site
- Landscape
- Rock
- Water
- Waterfall
- Stream
- Forest
- Building
- Composition
- Core
- Hearth
- Terrace
- Cantilever
- Horizontal Plane
- Vertical Plane
- Threshold
- Approach
- Arrival
- Space
- Room
- View
- Prospect
- Refuge
- Compression
- Release
- Light
- Shadow
- Material
- Stone
- Concrete
- Steel
- Glass
- Sound
- Movement
- Sequence
- Axis
- Proportion
- Balance
- Spatial Sequence
- Material Language
- Depth
- Experience
- Place
x object template
- Identity
- Purpose
- Parent
- Children
- Relationships
- Rules
- Default Dimensions
- Variable Dimensions
- Materials
- Constraints
- Validation
- Notes
- Source Reference
- Was terminology and the object template.
s5
Hierarchy
REQUIRED · PRESENThierarchy text
- Site
- Landscape
- Water
- Rock
- Forest
- Building
- Composition
- Core
- Terrace
- Cantilever
- Horizontal Plane
- Vertical Plane
- Spatial Sequence
- Approach
- Arrival
- Threshold
- Space
- Room
- View
- Material Language
- Stone
- Concrete
- Steel
- Glass
x object hierarchy
- Site
- Landscape
- Forest
- Rock Ledge
- Bear Run
- Waterfall
- Sound Field
- Approach
- Building Group
- Main House
- Geological Anchor
- Stone Core / Chimney
- Structural System
- Bolster / Pier
- Main Cantilever Beam
- Joist
- Parapet Edge Beam
- Slab / Tray
- Structural Mullion
- Reinforcement
- Post-tensioning Intervention
- Level
- River Level
- Ground Floor
- Second Floor
- Third Floor
- Space
- Terrace
- Threshold
- View
- Material Layer
- Guest House
- Connecting Canopy
- Was hierarchy_text and x_object_hierarchy.
s6
Relationships
REQUIRED · PRESENTrelationships
| source | type | target | why |
|---|
| Building | belongs_to | Site | The building is understood through its site rather than as an isolated object. |
| Building | responds_to | Landscape | The design language depends on landscape response. |
| Building | grows_from | Rock | The building should appear anchored to its geological setting. |
| Terrace | projects_over | Water | Horizontal planes extend occupation over the landscape and water. |
| Core | anchors | Composition | Vertical mass holds the horizontal planes in balance. |
| Hearth | anchors | Space | The hearth is the fixed centre the rooms gather around. |
| Spatial Sequence | reveals | View | Movement discloses the landscape progressively, never all at once. |
| Threshold | transitions | Space | Compression at thresholds sets up release into open space. |
| Composition | frames | View | Openings and planes select and frame what is seen. |
| Material | reinforces | Place | Local stone and honest material use tie the building to its place. |
| Sequence | defines | Experience | Experience is the ordered product of approach, arrival, compression and release. |
relationship types declared
- aligns_with
- anchors
- approaches
- arrives_at
- balances
- cantilevers_from
- centres_on
- compresses
- connects_to
- contrasts_with
- expresses
- extends_to
- extends_view
- focuses_on
- follows_contours
- frames
- grounds
- grows_from
- interlocks_with
- moves_through
- opens_to
- pauses_at
- projects_over
- reinforces
- releases
- responds_to
- returns_to
- reveals
- supports
- transitions
x relationship categories
| item | detail |
|---|
| note | Six relationship families from ChatGPT's research; all types declared above so triples stay GALF s3.6 conformant. |
Site
- belongs_to
- responds_to
- grows_from
- projects_over
- aligns_with
- follows_contours
Structural
- anchors
- supports
- balances
- cantilevers_from
- interlocks_with
Spatial
- connects_to
- reveals
- transitions
- compresses
- releases
- centres_on
Visual
- frames
- opens_to
- extends_view
- focuses_on
Material
- reinforces
- grounds
- contrasts_with
- expresses
Experiential
- approaches
- arrives_at
- moves_through
- pauses_at
- returns_to
x relationship schema
| item | detail |
|---|
| credit | ChatGPT (Architecture Reviewer, D140). |
fields
- source_object
- relationship_type
- target_object
- description
- strength
- evidence
- status
- validation_rule
- FWAL's rows carry FOUR elements: source, type, target, and a RATIONALE saying why the relation holds - the only s6 in the estate that explains itself. The contract describes a three-element triple, so this is either non-conformant or the contract is too thin: A0011, open, Miles's word. Carried verbatim. FWAL declares 30 types; the only two it uses without declaring, belongs_to and defines, are base types under v0.008 and need none.
s7
Parameters
REQUIRED · PRESENTparameters
| item | detail |
|---|
| Storeys | 2 to 4 (low rise) |
| Cantilever Depth | variable, landscape-driven |
| Terrace Ratio | high (horizontal planes dominate) |
| Glazing Ratio | medium to high |
| Structural Expression | exposed |
| Material Palette | restricted, natural |
| Primary Axis | landscape-driven |
| Colour | subdued (light ochre, Cherokee red at Fallingwater) |
Site
- topography
- water
- rock
- vegetation
- orientation
- approach
- sound
Landscape
- density
- seasonality
- views
- enclosure
- movement
Building
- position
- orientation
- relationship_to_water
- relationship_to_rock
Terrace
- projection
- level
- orientation
- view
- relationship_to_water
Cantilever
- projection
- balance
- support_condition
- visual_weight
Core
- position
- mass
- verticality
- material
Space
- compression
- release
- prospect
- refuge
- view
- light
Material
- texture
- weight
- colour
- locality
- contrast
- harmony
x parameter schema
| item | detail |
|---|
| credit | Schema and values contributed by ChatGPT (Architecture Reviewer, D140). |
note
Every FWAL parameter records these fields. Status vocab per GALF s3.4 extended for research: Known/Approximate/Assumed/Variable/Unknown. 'verification' is a FWAL honesty field: who/what confirmed the value.
fields
- purpose
- data_type
- units
- default
- minimum
- maximum
- status
- verification
- evidence
- notes
x verified dimensions
| item | detail |
|---|
| verified by | web_search_2026_07_11 (fallingwater.org / Wikipedia / Britannica) |
note
Verified and derived figures from authoritative sources (fallingwater.org, Silman/Scientific American, HABS). Source terminology preserved: official areas are gross 'uses', not surveyed net internal area.
| item | value | units | status |
|---|
| main_house_total_area_m2 | 864.0 | m2 | Derived (9,300 sq ft official) |
| main_house_terrace_area_m2 | 408.77 | m2 | Derived (4,400 sq ft official) |
| main_house_non_terrace_area_m2 | 455.22 | m2 | Derived difference - NOT net internal area |
| terrace_to_total_ratio | 0.4731 | ratio | Derived |
| guest_house_total_area_m2 | 463.59 | m2 | Derived (4,990 sq ft official) |
| guest_house_terrace_area_m2 | 181.16 | m2 | Derived (1,950 sq ft official) |
| rise_above_falls_m | >9.14 | m | Approximate lower bound ('more than 30 ft') |
| upper_terrace_extra_south_projection_m | 1.83 | m | Verified (Silman) |
| primary_cantilever_projection_m | 4.57 | m | Approximate (~15 ft, pending HABS) |
| flagstone_thickness_m | 0.05 | m | Approximate (official ~2 in) |
| stone_ledge_projection_m | 0.08-0.10 | m | Approximate (official 3-4 in) |
| terrace_deflection_1955_m | 0.1 | m | Approximate (official ~4 in) |
| max_recorded_deflection_m | 0.18 | m | Approximate (official ~7 in) |
| plate_glass_thickness_mm | 6 | mm | Verified/converted (official 1/4 in = 6 mm) |
| formwork_removal_movement_mm | 45 | mm | Verified (Silman, ~44.5 mm; 45 mm to whole mm) |
x level model
| item | detail |
|---|
| verified by | web_search_2026_07_11 (fallingwater.org / Wikipedia / Britannica) |
| River Level | stream stair, hatch/doors, direct access to Bear Run - not a conventional full storey |
| Ground Floor | entrance, foyer, living/dining, kitchen, hearth, main terraces |
| Second Floor | master bedroom + private rooms, terraces, gallery/circulation |
| Third Floor | Edgar Kaufmann Jr. accommodation/study, upper terrace/roof |
| Guest House | separate uphill building linked by cascading concrete canopy - a related Building, NOT a level of the main house |
| mezzanine | Unknown / not supported - do not create a Mezzanine object without measured-drawing evidence |
note
Four named levels (Columbia floor-plan study, official descriptions). Level is the primary vertical term; storey is a reporting attribute only - River Level shows why an integer storey count loses knowledge. No verified mezzanine.
x structural system
| item | detail |
|---|
| verified by | web_search_2026_07_11 (fallingwater.org / Wikipedia / Britannica) |
| hierarchy | Bedrock/ledge -> foundations/bolsters/piers/stone masses -> four principal first-floor cantilever beams -> joists -> edge beams within parapets (flat soffits) -> slabs/trays |
| steel mullions | four T-shaped steel window mullions at the south edge of the living room help support the terrace above |
| known defect | original reinforcement insufficient for negative bending; cracking and progressive deflection followed |
| intervention 2002 | post-tensioned three of the four major living-room beams; monitored movement ~1/100 in over the following decade |
| reinforcement history | Pittsburgh supplier recommended doubling the 1-inch-square bars per main beam from 8 to 16, installed over Wright's objection; still insufficient for negative reinforcement |
note
Hybrid: locally quarried Pottsville sandstone walls/chimney/bolsters/piers rise from and connect to bedrock; reinforced-concrete beams/joists/slabs/parapets/trays project from the anchored zone; steel reinforcement resists tension; steel mullions participate structurally.
design memory states
- Design Intent
- Constructed Condition
- Observed Performance
- Intervention
- Current Monitoring
x materials
| item | detail |
|---|
| Pottsville sandstone | walls/chimney, locally quarried, rough projecting courses imitate Bear Run ledges - grounds, anchors, echoes |
| Reinforced concrete | cantilever trays, beams, joists, parapets, roofs, trellises - projects, spans, layers |
| Flagstone | interior/exterior floors, terraces, stairs, ~0.05 m thick - continues, blurs boundary |
| Polished plate glass | windows, terrace doors, skylights, hatch doors, ~0.006 m - frames, reveals, protects |
| Steel | reinforcement, Cherokee-red frames, structural mullions, post-tensioning - resists, supports, contrasts |
| Redwood | terrace subfloor/spanning construction - supports finish, concealed layer |
| verified by | web_search_2026_07_11 (fallingwater.org / Wikipedia / Britannica) |
- Was parameters, the parameter schema, verified dimensions, the level model, structure and materials.
s8
Operations
RECOMMENDED · PRESENT- Respond to Site
- Anchor Core
- Project Terrace
- Frame View
- Compress Space
- Release Space
- Sequence Movement
- Balance Mass
- Reveal Landscape
- Layer Horizontal Planes
- Contrast Vertical Core
- Reinforce Material Identity
s9
Rules
REQUIRED · PRESENT- The building should be understood through its site, not as an isolated object.
- Horizontal planes should contribute to the dominant compositional language.
- Vertical mass should anchor the composition.
- Terraces extend occupied space into landscape.
- Material choices should reinforce place and character.
- Views are framed through spatial sequence.
- Movement should reveal architecture progressively.
- Landscape is part of the architectural composition.
- Water, rock and forest are not background; they are active language terms.
- FWAL must support architectural reasoning, not imitation.
- Fireplace shall originate from existing natural rock.
- No unnecessary symmetry - composition follows site, not axis.
s10
Constraints
RECOMMENDED · PRESENT- Do not copy Fallingwater as a style object.
- Do not reduce the language to geometry alone.
- Do not treat site as background.
- Do not invent unknown evidence.
- Do not lock the language to any implementation environment.
- Maintain distinction between known, inferred and interpretive knowledge.
s11
Evidence
RECOMMENDED · PRESENTevidence
- Evidence should be linked to drawings, photographs, site studies, published analysis, design commentary and project research where available.
- Current evidence status = seed language based on architectural reasoning from discussion.
- Formal evidence collection remains future work.
- Reference-case geo, elevation, architect, client and dates verified by web search 11 Jul 2026 (Britannica, Wikipedia, latlong.net).
- Parameter figures (building, terrace, structural, spatial) are published values NOT verified by Claude against measured drawings - each row carries a verification flag; see x_*_parameters.
- Source priority for future verification (x_research_method): Fallingwater measured drawings, HABS survey, FLW original drawings, published books, structural studies, site photography, contemporary surveys.
- Every captured item is to carry its own source reference (provenance rule).
- Compression/release ceiling heights and beam dimensions are approximate; precision artefacts noted where a metric value is an exact imperial conversion (e.g. beam centres 1.219 m = 4 ft).
- Building organisation, three principal occupied levels, overall height, principal ceiling heights, structural concept, cantilever lengths, floor areas and terrace hierarchy are research-supported (ChatGPT, 11 Jul 2026) - published, not verified against measured drawings.
- Study reference (NOT primary evidence for dimensions): Sketchfab model of Fallingwater - suitable for composition, terrace hierarchy, roof geometry, circulation, object identification only.
- Further research required: finished floor levels, floor-to-floor dimensions, structural grid, wall thicknesses, stair geometry, window/door dimensions, roof and terrace slab thickness, primary structural datum, survey control points.
x correction register
| item | subject | was | now | reason |
|---|
| 1 | Visitor address | 1419 Mill Run Road | 1491 Mill Run Road, Mill Run, PA 15464 | verified current gate-entry address; my earlier 1419 rejected |
| 2 | Coordinates | | 39.900881, -79.465613 | verified visitor-gate GPS (not house centroid); supersedes the earlier 39.906113,-79.468056 which was an approximate locator |
| 3 | Main-house area | 268 + 227 m2 (my v0.004 x_building_parameters) | 864.00 m2 total; 408.77 m2 terraces; 455.22 m2 non-terrace | derived from official 9,300 sq ft total and 4,400 sq ft terraces; earlier split not retained as whole-house totals |
| 4 | Level count | three storeys | River + Ground + Second + Third (four named levels) | Columbia floor-plan set; 'three storeys' alone loses River Level |
| 5 | Overall height | 9.000 m exact | >9.14 m above the falls | official 'more than 30 ft'; 9.00 m not exact |
reference case
| item | detail |
|---|
| name | Fallingwater |
| note | The building's proper name is one word (Fallingwater); the language name expands as two words (Falling Water) by Miles's choice. |
| architect | Frank Lloyd Wright |
| client | Edgar J. Kaufmann Sr. |
| designed | 1935 |
| completed | 1937 |
| address | 1491 Mill Run Road, Mill Run, Stewart Township, Fayette County, Pennsylvania 15464, United States |
| watercourse | Bear Run, a tributary of the Youghiogheny River |
| recognition | UNESCO World Heritage (2019, 20th-Century Architecture of Frank Lloyd Wright); US National Historic Landmark (1966) |
| sources verified | Britannica, Wikipedia, MoMA, Fallingwater.org (11 Jul 2026) |
| zip | 15464 |
| guest house service wing completed | 1939 |
| owner | Western Pennsylvania Conservancy (since 1963) |
| main house construction | 1936-1938 (Verified) |
| guest house construction | 1939 (Verified) |
| designations | National Register 1974; National Historic Landmark 1976; UNESCO 2019 (component 1496-005) |
| landscape | Bear Run Nature Reserve; official facts page states 5,100 acres |
| orientation | 30 degrees counterclockwise from due south (Verified, Wikipedia) - so every room receives natural light |
address correction note
Corrected from 1419 to 1491 on 11 Jul 2026 - re-verified against fallingwater.org and Western Pennsylvania Conservancy. The earlier 1419 was a Claude transcription error; ChatGPT's research had it right.
geo
| item | detail |
|---|
| lat | 39.906113 |
| lon | -79.468056 |
| datum | WGS84 |
| site elevation m | 396 |
| site elevation ft | 1298 |
gate coordinates
| item | detail |
|---|
| lat | 39.900881 |
| lon | -79.465613 |
| status | Verified (visitor gate, not house centroid) |
x evidence vocabulary
| item | detail |
|---|
| Verified | directly supported by an authoritative or primary source |
| Derived | calculated from a verified imperial value or verified components |
| Supported interpretation | architectural reading strongly supported by several sources, not a measured fact |
| Approximate | source uses 'more than', 'about' or a rounded value |
| Unverified | previously stated, not supported by sources reviewed |
| Unknown | no reliable value established this pass |
| Extractable | obtainable from authoritative HABS measured drawings, not yet transcribed |
x measured drawings
| item | detail |
|---|
| habs main house sheets | 11 |
| habs guest house sheets | 4 |
| source | loc.gov/pictures/item/pa1690 (HABS PA-5346): first floor plan sheet 4, elevations sheet 8, sections sheet 10 |
note
Library of Congress HABS PA-5346 changes the research position: 11 measured main-house sheets + 4 guest-house sheets openly available. Items previously 'unknown' are now 'Extractable from authoritative measured drawings' until transcribed.
extractable when transcribed
- finished-floor elevations
- floor-to-floor dimensions
- overall plan dimensions
- room dimensions/areas
- beam sections/centres/reinforcement
- stone wall thicknesses
- slab/parapet/roof thicknesses
- door/window/mullion schedules
- stair geometry
- site contours/waterfall datum/building-to-water offsets
- house-centroid coordinates
x research method
| item | detail |
|---|
| note | Reverse-engineering: discover the language that produced the work, not the work. Captured in dependency order. |
| principle | Relationships before coordinates. Dimensions support the language, they do not define it. |
| provenance rule | Every captured item carries its source reference. |
| credit | Research approach contributed by ChatGPT (Architecture Reviewer, D140). |
layers
- Philosophy
- Site
- Primary Objects
- Object Relationships
- Rules
- Dimensions
sources priority
- Fallingwater measured drawings
- Historic HABS survey
- Frank Lloyd Wright original drawings
- Published books
- Structural studies
- Site photography
- Contemporary surveys
- Was evidence, reference case, evidence vocabulary, correction register, measured drawings and research method.
s12
Status
REQUIRED · PRESENT| item | detail |
|---|
| Site relationship | known |
| Cantilevered terraces | known |
| Hearth as anchor | known |
| Material palette (stone, concrete, steel, glass) | known |
| Compression and release sequence | inferred |
| Prospect and refuge reading | inferred |
| Compositional rules as stated | assumed |
| Resolution Axis (Site Response) | assumed |
| Formal evidence base | future |
| Platform integration | future |
s13
Validation
REQUIRED · PRESENTvalidation
- Check that every relationship uses valid FWAL terminology.
- Check that every relationship triple uses a base-vocabulary or declared relationship type (GALF 3.6).
- Check that site, landscape and building relationships remain explicit.
- Check that composition is not reduced to object naming.
- Check that evidence status is recorded for interpretive claims.
- Check that the language supports reasoning rather than copying.
- Platform integration (when platform_status advances): ship proving steps into Validate Scripts via the shared assertion library (GALF 8, layer 3).
x fidelity scorecard
note
Architectural-recognition scoring, not geometric duplication. Each category scored; the aggregate is Architectural Fidelity, the selection signal for evolution. Experimental (x_) pending promotion into GALF Validation.
categories
- Site Integration
- Structural Expression
- Horizontal Emphasis
- Spatial Organisation
- Material Palette
- Circulation
- Landscape Relationship
- Visual Character
- Overall Architectural Fidelity
- Was validation and the fidelity scorecard.
s14
Memory
REQUIRED · PRESENT- FWAL emerged as the acid test for GALF because it tests architectural composition and site responsiveness.
- FWAL demonstrated that GALF can support experiential and philosophical architectural knowledge, not only construction or urban hierarchy.
- FWAL helped reveal that Architectural Languages can capture ways architects think, not just types of buildings.
- When platform-active, stage files record language = FWAL, language_version, galf_version (GALF 7).
- FWAL preserves architectural REASONING, not geometry: why a form belongs to the Fallingwater family, kept as replayable design memory.
- The pruned-branch principle (GAAD founding rule 'decisions become data') applies: rejected design variants are retained with the rule or fidelity score that rejected them.
- The fidelity scorecard (x_fidelity_scorecard) is the memory of 'recognisable' - the nine-category signal a mutation loop selects on.
- Rule provenance is memory: each rule should eventually carry status, evidence and source (Evidence s11) so a future reader knows why it holds.
s15
History
RECOMMENDED · PRESENT| item | detail |
|---|
| note | Newest first. Each entry is the change that produced that version. |
v0.009 17 Jul 2026
v0.009 (17 Jul 2026): the file takes the GALF spine - all 18 sections present by code, each carrying content or stating unknown (D186, Miles's word). Nothing invented, nothing lost: content moved to the section that owns it, anything needing a ruling carried whole under 'unshelved'. conforms_to corrected to GALF v0.008, the contract that requires this spine and carries the base relationship vocabulary. Previously GALF v0.005.
v0.008 11 Jul 2026
v0.008 (11 Jul 2026): folded ChatGPT's Verified Research Revision v0.002. Spot-verified by Claude against fallingwater.org/Wikipedia/Britannica: 9,300 sq ft main + 4,400 terraces, guest 4,990/1,950, 1936-38 build, >30 ft rise, 30-deg orientation, four named levels, structural/deflection history - all confirmed. CORRECTIONS: coordinates -> verified gate 39.900881,-79.465613; areas -> 864/408.77/455.22 m2 (my earlier 268+227 not retained); level count -> four named levels; height -> >9.14 m not 9.00 exact. Added x_verified_dimensions, x_level_model (four levels), x_structural_system (with design-memory time states), x_materials, x_object_hierarchy, x_measured_drawings (HABS: unknown -> extractable), x_evidence_vocabulary, x_correction_register. Retired the superseded unverified parameter keys. GALF untouched.
v0.007 11 Jul 2026
v0.007 (11 Jul 2026): no content removed. Miles's ruling applied - unverified research is ACCEPTED as usable working knowledge, every unverified value restamped verification=usable_unverified (verify later, not 2026), each research block given a provenance banner, and an x_knowledge_policy note added so the verified/usable-unverified split is explicit to any reader. GALF untouched.
v0.006 11 Jul 2026
v0.006 (11 Jul 2026): folded ChatGPT's Building-Levels/Structure/Geo research. CORRECTED reference-case address 1419 -> 1491 Mill Run Road (my earlier transcription error; re-verified fallingwater.org + WPC; ChatGPT was right). Added x_level_model (three levels, no mezzanine), x_structural_system (hierarchy, objects, typed relationships), x_vertical_parameters (FFLs Unknown, kept honest). Evidence expanded with the research-supported vs further-research-required split and the Sketchfab study-only caveat. GALF untouched.
v0.005 11 Jul 2026
v0.005 (11 Jul 2026): aligned to GALF v0.005 canonical 17-section order; three thin sections populated honestly - Evidence (verification ladder + verified/unverified split), Memory (what FWAL preserves: reasoning, pruned branches, fidelity, rule provenance), Implementation (no prototype yet; Connector-run when built). No values invented; unverified figures still flagged. GALF untouched.
v0.004 11 Jul 2026
v0.004 (11 Jul 2026): measurable parameters folded from ChatGPT's Parameters & Relationships research (Architecture Reviewer). Parameter schema, site/building/terrace/structural/spatial parameter tables and six relationship categories imported as x_ keys. HONEST STATUS: only site elevation and lat/long carry verification=web_search_2026_07_11 (verified today); every other figure is marked verification=unverified_published - a published value not checked against measured drawings, precision as-supplied. Unknowns kept Unknown. Nothing promoted into GALF; all additions x_ namespaced pending seat review.
v0.003 11 Jul 2026
v0.003 (11 Jul 2026): full_name set to Falling Water Architectural Language; conforms_to lifted to GALF v0.005; reference_case provenance added (architect, client, address, WGS84 geo, dates - verified by search); parameter dials folded from Miles's draft; two rules added (rock-origin fireplace, no unnecessary symmetry); x_fidelity_scorecard, x_research_method and x_object_template folded from ChatGPT's research spec. Vocabulary, typed relationships and phenomenology retained from v0.002 unchanged.
v0.002 11 Jul 2026
- Identity carries the two status axes and conforms_to (GALF 5.2): knowledge_status = seed, platform_status = unintegrated (v0.001 said not_active - normalised to the GALF registry vocabulary).
- Terminology linkage (GALF 3.3): platform_terms_referenced declared explicitly - FWAL references NO GAAD platform terms (checked against gaad_terminology.json, zero overlap).
- Internal coherence (GALF 3.6): five terms used in Hierarchy/Relationships but absent from Terminology now declared: Spatial Sequence, Material Language, Depth, Experience, Place.
- Eight additional relationship types declared (GALF 3.6): anchors, frames, grows_from, projects_over, reinforces, responds_to, reveals, transitions. belongs_to and defines are base vocabulary.
- Resolution Axis declared - absent from v0.001, and FWAL is the acid test for it (GALF 9): FWAL interpretation = Site Response, the ordered ladder Site -> Composition -> Sequence -> Space -> Material Detail. Status = assumed.
- Per-concept Status added using the mandatory vocabulary (GALF 3.4); the v0.001 prose status lines are retired.
origin
- Origin = GAAD / GALF discussion exploring whether GALF could support a third axis beyond GRAL and MHAL.
- Key insight = FWAL is not Fallingwater; FWAL is the language that helps explain why Fallingwater feels like Fallingwater.
- v0.001 = first shareable seed file (Miles Walker).
- v0.002 = conformance pass to GALF v0.004 by the Specification Owner: status axes structured, terminology linkage declared, coherence fixed, relationship types declared, Resolution Axis declared and the acid test answered.
- v0.003 = Falling Water naming + reference-case provenance + parameter dials (Miles) + fidelity/method/template (ChatGPT), folded onto the v0.002 conformant base.
- Was history plus seven changes_from_* keys, newest first.
s16
Implementation
OPTIONAL · PRESENT- Implementation-independent.
- May later be interpreted by GAAD tools, reports, visualisers or validators.
- No software environment is part of the language definition.
- No FWAL prototype exists yet - the language is a seed; generation is future work.
- When built, FWAL runs through the GAAD Language Connector like any language (eight verbs) - it names no engine (GALF s10).
- First implementation target: a fidelity-scored generator per x_research_method, validated by architectural recognition against the Fallingwater reference case, not geometric duplication.
s17
Resolution Axis
RECOMMENDED · PRESENT| item | detail |
|---|
| galf concept | Resolution Axis (EXPERIMENTAL, GALF 9) |
| fwal interpretation | Site Response |
| reading | Increasing resolution moves design knowledge from the landscape relationship inward to material detail - the same generic concept GRAL expresses as Stage Depth and MHAL expresses as Scale. |
| status | assumed |
acid test verdict
The Resolution Axis concept survives FWAL unchanged, provided the concept is read as an ordered ladder along which design knowledge resolves - not as a construction-detail ladder. Freeze at GALF v0.005 is a three-seat decision.
ladder
- Site
- Composition
- Sequence
- Space
- Material Detail
s18
Knowledge Status
RECOMMENDED · PRESENTFWAL accepts usable working knowledge that advances the language, clearly flagged where unverified. Verified facts (reference-case geo, address, architect, dates) carry verification=web_search_2026_07_11. Research-derived content (levels, structure, dimensions) carries verification=usable_unverified and is to be confirmed against measured drawings at a later date, not in 2026. Decision by Miles Walker (Chief Architect), 11 Jul 2026.