**Note** from Bead: 96597c32 · [canonical source](https://redfish.acequia.io/guerin/.agents/96597c32-7f1c-4e95-acd5-635673461781/2026-05-16/notes/anysurface-as-digital-landscape.md) · session 2026-05-16 · discussion: Talk: 96597c32
**AnySurface** is Stephen Guerin / Simtable / Redfish's ongoing work, since 2009, of turning rooms into interactive computational surfaces. A camera and a projector form a perception-action loop on a physical surface — a table, a wall, a patient, a fireshed map. The camera senses what is on the surface and how it changes; the projector lights the surface with computed content; together they make the surface *responsive*. The room becomes the substrate. This note positions AnySurface as the load-bearing concrete instance of **landscape as computation** — definition 4 of the four-definitions arc — in domestic, civic, and pedagogical space. **Source artifacts:** - Speaker transcript (Stephen Guerin himself, ~2015): [anysurface.transcript.md](../artifacts/anysurface.transcript.md) (mirror in this bead's `artifacts/`) and the canonical at [simtable.com/videos/anysurface.transcript.md](https://simtable.com/videos/anysurface.transcript.md). - Source video: [youtube.com/watch?v=wuepu_PPmNI](https://www.youtube.com/watch?v=wuepu_PPmNI). - Related current pitch: AnySurface 3D Scan — projector-camera calibration + structured light + RGBD fusion + Acequia GeoCamera integration ([guerin.acequia.org/redfish/pitches/anysurface-3d-scan.md](https://guerin.acequia.org/redfish/pitches/anysurface-3d-scan.md)). - Companion site: [anysurface.com](https://anysurface.com) (acequia.io add-on domain).
## Why this is *digital landscape*, not "interactive projection" AnySurface is not a screen. It is not a tool *applied to* a room. It is **the room becoming computational substrate.** The distinction matters and locates AnySurface precisely in the four-definitions arc: | Definition | Where AnySurface lives (or doesn't) | |---|---| | 1. Numerical solutions to equations | Not here. AnySurface uses simulations, but the contribution is not the equations. | | 2. Computational tools for the landscape architect | Not quite. It is a *platform*, but its claim isn't "better tools." | | 3. Simulation of landscapes | Partially here. The wildfire spread, the agent-based model — these are simulations. But the simulation isn't running *on a screen*; it is running *on the territory it models*. | | 4. **Landscape as computation** | **Here.** The surface itself is the computational substrate. Sensors and actuators close the loop with the inhabitant. Intelligence is in the surface, not applied to it from a separate cockpit. | Stephen's own phrasing in the 2015 video, almost verbatim: *"the light bulb needs to fundamentally be interactive… you are the author of the room."* That is definition-4 stated in everyday vocabulary — the *room* is the computation; the inhabitant is its co-author.
## The perception-action coupling AnySurface implements perception-action coupling literally, in physical light, on real surfaces: ``` [INHABITANT] │ moves, gestures, places objects │ ▼ ┌────────────────────────────────────────┐ │ CAMERA — perception (any phone, any │ │ webcam, any networked sensor) │ └─────────────┬──────────────────────────┘ │ depth, motion, object tracks ▼ ┌────────────────────────────────────────┐ │ BROWSER — compute (HTML5, edge-local, │ │ no install, runs in any tab) │ └─────────────┬──────────────────────────┘ │ updated state, projected frame ▼ ┌────────────────────────────────────────┐ │ PROJECTOR — action (light onto the │ │ surface; the surface itself is the │ │ output device) │ └─────────────┬──────────────────────────┘ │ ▼ [SURFACE] changed by the projected light → seen by camera again → loop closes ``` This is **active inference** in the technical sense (the system minimizes prediction error by acting on the environment, then sensing the change). It is **embodied cognition** in the Hutchins / Clark sense (the cognition lives in the loop, not in the box). It is **stigmergic** in the bead-game sense (each cycle deposits a pheromone in the form of projected light + camera response that conditions the next cycle). And it is **landscape architecture** in the literal sense: a coordinated design of surface, light, sensing, and inhabitant behavior on a shared territory, with the substrate as the medium of the design.
## Matrix, patches, corridors — Forman applied to the room Forman's landscape-ecology vocabulary maps cleanly onto an AnySurface-instrumented space: - **Matrix** — the room itself: walls, floor, ceiling, fixtures, ambient light, network. The dominant background that patches and corridors are embedded in. - **Patches** — designated surfaces: the table that runs the wildfire simulation, the wall that displays the data overlay, the patient on the surgery bed, the section of floor that becomes a control panel. Each patch is a region where compute-into-light happens. - **Corridors** — the *light paths*, the *gaze paths*, the *gestures and movement* that link patches. A projector beam from ceiling to table is a corridor. The inhabitant's gaze sweeping from data overlay to physical map is a corridor. The 5 GHz network linking phone-camera to projector is a corridor. Designing AnySurface deployments is *matrix design*. Which surfaces are patches? Which light paths are corridors? What is the surrounding matrix doing — supporting, interfering, framing? Landscape architects' instincts about edge effects, succession, and stewardship apply directly.
## Why AnySurface aligns with the acequia manifesto The [acequia manifesto](https://acequia.org/acequia-manifesto.md) sets up a fork in the road: cloud-extractive vs edge-sovereign architecture for ambient AI. AnySurface, by its 2015 design choices, already sits firmly on the edge-sovereign side of that fork: | Manifesto principle | How AnySurface realizes it | |---|---| | *"The architecture determines the outcome"* | Choosing browser-native HTML5 + local camera + local projector means data stays in the room. No cloud round-trip required. | | *"Your devices serve; they don't just consume"* | The phone is the *camera server*; the laptop is the *compute server*; the projector is the *actuator*. Each is owned by the inhabitant. | | *"Your laptop is your land. Let your browser be a server"* | Verbatim what AnySurface does. The browser is the runtime; no install; no IT permission; the user controls the deployment. | | *Common pool resources federated across parcelas* | Phone provides camera + portable compute. Laptop provides bigger compute + projector driving. Network provides corridor between parcelas. Each contributes what it has. | | *Commons governed by owners* | The room's compute is governed by whoever is in the room — not by a platform that hosts the session. | AnySurface is **eleven years of working evidence** that the architecture the manifesto calls for is buildable. It existed before the manifesto articulated *why*; the manifesto articulates the political/economic stake of the architectural choice AnySurface already made.
## Connections to the rest of this bead - **[hubler-network.md](hubler-network.md)** — AnySurface implements the **GET → COMPUTE → PUT** loop in light. Camera GETs surface state; browser computes; projector PUTs light back onto the surface. The polarization is dipolar in Stephen's sense: each surface region simultaneously *senses* and *is acted on*. Mead's coherence/resonance/entrainment is literal here — the camera and projector must be *coherent* in geometry (calibration), *resonant* in update rate (frame-locked), *entrained* with the inhabitant's perception (low-latency, perceived as continuous). - **[polymorphic-get-content-negotiation.md](polymorphic-get-content-negotiation.md)** — A single AnySurface surface renders differently per caller: as fire-spread animation to the firefighter trainer, as patient-anatomy overlay to the surgeon, as physics simulation to the sixth-grader, as point-cloud preview to the 3D-scan calibrator. Same URI/surface; different `Accept` per context. - **[spatial-cache-matrix-citizen-fires.md](spatial-cache-matrix-citizen-fires.md)** — AnySurface IS a spatial cache matrix, but the spatial layer is *the actual room* rather than a virtual map. The fire-spread simulation projected on the table is the same shape of data flow as citizen-photo participation in the network — local capture, local serve, opportunistic peer sync, multi-stakeholder shared view. - **[service-worker-as-edge-agent.md](service-worker-as-edge-agent.md)** — AnySurface anticipated this pattern by ~7 years. Each browser tab on an AnySurface deployment is an edge agent serving the room's URIs. - **[three-eras-of-web-topology.md](three-eras-of-web-topology.md)** — AnySurface is Era-3 architecture (stigmergic, participatory, polarized) implemented in domestic/civic light. - **[runtime-binding-vs-build-time-linking.md](runtime-binding-vs-build-time-linking.md)** — "no installation of software" + "browser is a server" = pure runtime binding. AnySurface deployments don't install dependencies; they bind them at runtime. - **[bead-game-as-cultural-interface.md](bead-game-as-cultural-interface.md)** — AnySurface is a culturally-evolved interface for navigating an otherwise-intractable plenum (complex-systems simulation made tangible). Sixth-graders manipulating a fire-spread model on a tabletop is the very claim of "cognitively graspable handles on conceptual plenum" — *operating in physical light*. - **[field-and-bead-reconciliation.md](field-and-bead-reconciliation.md)** — AnySurface makes the field/bead split tangible: the projected light is the *field* (continuous, modulated, ambient); the inhabitants' actions and the camera-tracked objects are *beads* (discrete, named, manipulable). The score is the running simulation; the performance is the projected light; the inhabitants improvise on the surface.
## What AnySurface gives the DLA-2026 framing For the 3-minute video on **"the architecture of the digital landscape / computational landscape architecture as design intervention"**, AnySurface is the *concrete demo that makes the abstract claim real*. The audience can be told: > "*The sensors are already in your home. The question is who governs the agents that process what they see and hear.* For sixteen years our team has been building one answer — AnySurface. Camera-and-projector pairs that turn any surface into a substrate for computation. Browser-native. No installation. Edge-deployed. The room becomes its own computational landscape, with the inhabitants as its authors. We've shipped sixty of these to firefighters since 2009. We're now building 3D structured-light scan into the same architecture. The question DLA-2026 should be sitting with: *what landscape are we choosing to design, and who gets to be its author?*" That is the talk's argumentative center, anchored by a project that has been working in the field for a decade and a half.
## Open thread The 2015 video transcript was cut at the end ("…you are the author of the…") — the YouTube auto-caption track ran out before the sentence ended. If the original mp4 (at `simtable.com/videos/anysurface.mp4` or elsewhere) is available, a manual transcription of the final 5-10 seconds would close the citation. Worth doing before quoting in a paper.
## A sentence to keep > AnySurface is *landscape as computation* realized in domestic and civic light: the room as the substrate, the camera-projector pair as the perception-action loop, the inhabitant as the author, and the browser as the parcela's sovereign runtime — the architecture the acequia manifesto demands, built and shipping since 2009.