**Note** from Bead: 96597c32 · [canonical source](https://redfish.acequia.io/guerin/.agents/96597c32-7f1c-4e95-acd5-635673461781/2026-05-16/notes/data-sovereignty-and-sousveillance.md) · session 2026-05-16 · discussion: Talk: 96597c32
The foundational diagnosis the DLA talk has been circling but had not yet named cleanly: > **All the imagery and location data that is the foundation for geospatial digital landscape architecture is in the corporate clouds. Unscriptable. Revoked APIs. No sovereign control of the data. And the government surveillance systems are gone. There is no architecture for sousveillance. But students are learning to design these architectures.** — Stephen Guerin This note states the diagnosis precisely, names sousveillance as the missing architectural category, and locates it inside the talk's other moves.
## The three-way pinch on the data substrate Digital landscape architecture, as a practice, depends on a primary substrate of geospatial data — imagery, location histories, hydrology layers, sensor streams, plant/animal observation records, infrastructure status, climate data. **The substrate has been quietly captured from three directions at once:** ### 1. Corporate enclosure — your data, locked in their cloud Every commercial platform has followed the same arc: open early to capture users and content; restrict the API once captured; eventually revoke programmatic access entirely. | Platform | What was open | What was revoked / when | |---|---|---| | **Google Photos Library API** | Programmatic access to a user's own photos | Deprecated 2024; restricted 2025; effectively unscriptable for third-party tools by 2026 | | **Google Maps API** | Reasonable free tier for civic and academic use | Massive price hike 2018; prohibitive for civic-scale use 2024+ | | **Twitter / X API** | Free tier for researchers, public-interest tools, civic apps | Free tier killed Feb 2023; thousands of civic tools and academic studies dark | | **Reddit API** | Free programmatic access | Free tier killed June 2023; killed third-party clients and research scrapes | | **Meta (FB / IG) research access** | CrowdTangle, IG Graph API | CrowdTangle shut down Aug 2024; IG research access progressively restricted | | **Apple Photos** | Never had a usable third-party API | Closed by construction | | **Ring (Amazon)** | None | Cloud-only by design since 2018 Amazon acquisition; recording requires subscription; footage subpoena-able by law enforcement without homeowner notice via *Ring Neighbors* partnerships | | **Nest (Google)** | Local NAS bridging deprecated 2019 | Cloud-only; Google account required; Nest WebAPI killed 2020 | | **Wyze / Arlo / Eufy** | Limited local RTSP on some models | Local-access paths frequently broken in firmware updates; cloud-default reinstated by patch | | **Apple HomeKit Secure Video** | iCloud-only encrypted storage | Apple-tied; no third-party scripting | **The pattern is uniform.** Open until network effects lock users in; restrict once locked; revoke once a model has been trained on the captured corpus. Even the photos *you* took on *your* phone, stored on *your* paid Google account, cannot now be programmatically accessed by *you* for scripting purposes. **The data is yours in the legal sense and theirs in every operational sense that matters.** #### Home security cameras — the most intimate case Home cameras deserve their own emphasis because they sit at the *most intimate* layer of the data substrate. Every Ring doorbell, Nest cam, Wyze unit, or Arlo camera installed at the front door, in the kitchen, in the child's room is by default an Amazon/Google/Apple ingestion endpoint. **You bought the hardware. You powered the device. You set the password. The footage goes to the platform.** Local-only home cameras exist as a niche; the firmware-update treadmill regularly breaks the local-RTSP path, deprecates manufacturer bridging tools, or silently removes the local-storage option. **The closed loop is the business model.** For landscape work this matters because the camera in the kitchen and the camera at the front door are *primary geospatial sensors* for the home as digital landscape — and the inhabitant cannot script them. This is what the [acequia manifesto](https://acequia.org/acequia-manifesto.md) calls *tenancy on land owned by platform lords*. For landscape architects, it means the primary substrate of your professional work is *not yours to compose with*. #### The atrophied edge — your devices as crippled publishers The diagnosis goes one layer deeper than API revocation, into the architecture of the network itself. **The hardware and software to *publish* from your own device has been present on every consumer machine since the 1990s.** IIS ships with every Windows install. Apache is available on every Mac and Linux. Nginx, lighttpd, and a half-dozen others are one apt-install away. **These are not exotic capabilities.** Every laptop, every desktop, every phone is a fully capable server by hardware and operating-system construction. What has happened across the same window is the **systematic atrophy of the edge as a publisher**: - **ISP terms of service** — virtually every residential broadband contract since the early 2000s prohibits "running a server." Enforced by traffic-shaping, **inbound port 80 and 443 blocks on most residential connections**, and contract termination for repeat offenders. - **CGNAT and IPv4 exhaustion** — most residential broadband and all mobile connections sit behind Carrier-Grade NAT. **You do not have a public IP.** Inbound connections to your device are architecturally impossible without a relay you don't control. - **Mobile carrier port blocking** — Verizon, AT&T, T-Mobile block well-known server ports on consumer plans. The phone in your pocket cannot receive an inbound TCP connection on port 80, 443, 22, 25. - **OS defaults** — Windows since Vista hides IIS by default in obscure "Programs and Features" sub-menus; macOS removed the "Web Sharing" toggle from System Preferences in 2012; iOS has never offered server mode without jailbreak; Android's local-server apps require configuration the average user will not perform. - **Browser security model** — WebRTC and ServiceWorker offer the *technical* capability for the browser to act as a server (see [service-worker-as-edge-agent.md](service-worker-as-edge-agent.md)), but the integration into native *publishing* workflows is intentionally absent. Every "publish your photos" workflow defaults to a platform upload, not a local serve.
The result is **architectural asymmetry**: your devices are full-duplex publishers *in capability* and **functionally consumers only** *in deployment*. The intelligence-collection apparatus — corporate clouds, the surveillant state — runs full-duplex on you. They collect from your devices, train on your data, serve you back the inference. **You have been put on the receiving end of an asymmetric communication architecture that your hardware does not require and your operating system inherits without naming.** This is the deeper architectural fact beneath the API-revocation surface. The cripple is **policy + network design + UX defaults**, not capability. IIS, Apache, Nginx, WebRTC, ServiceWorker — all *literally already installed.* The means of production are already in your hands, exactly as the [acequia manifesto](https://acequia.org/acequia-manifesto.md) says. **The architectural intervention is to use them.** ### 2. Government surveillance systems — being dismantled in real time The other half of the diagnosis is just as load-bearing. Public-good earth observation, climate monitoring, and civic data infrastructure — the public-sector substrate that landscape architecture has relied on for decades — has been substantially **gutted across 2025–2026**: - USGS Landsat / earth-observation programs under funding threat - NOAA satellite, weather, and climate programs cut - USDA monitoring scaled back - EPA environmental monitoring reduced - USFS Burned Area Emergency Response (BAER) scope narrowed - USAID closure (Jan 2025) — many overseas earth-observation and development monitoring programs gone - US Wildland Fire Service consolidation reducing rather than expanding capacity - HUD housing-data programs reduced - Census Bureau capacity contraction - NSF / NIH / USGS budget cuts The pattern is uniform here too: **the public-good observation infrastructure that has, for the past half-century, made landscape architecture's analytical work possible is being defunded, restructured, or eliminated.** Some of these programs may return; many will not return in recognizable form. ### 3. The sousveillance gap — no architecture for citizens watching back The third leg of the pinch is the one with no incumbent and no name in the discipline: **citizens have no architecture for observing the landscapes they live in, at the scale and integration required to fill the gap left by (1) and (2).** The vocabulary for this exists. Steve Mann coined **sousveillance** in 1995 — *watching from below*, the inverse of surveillance. Where surveillance is institutions observing citizens (top-down, asymmetric, extractive), sousveillance is citizens observing institutions and conditions on the ground (bottom-up, distributed, accountability-shaped). The classic case is the citizen smartphone video of police misconduct. But the concept generalizes: - Surveillance: Google Photos / Apple Photos / Ring / Nest watching your home - Sousveillance: citizen phone imagery documenting post-fire watershed conditions, neighborhood flood risk, infrastructure status, planting recovery - Surveillance: USFS / NOAA satellite imagery (when funded), licensed and metered to citizens through institutional intermediaries - Sousveillance: federated citizen-camera networks ground-truthing the same territory — at higher temporal resolution, at finer spatial resolution, with stronger local context **Sousveillance is the design category for the third leg.** And **the architecture does not yet exist at the scale or with the governance the moment requires.**
## Why this matters for DLA specifically Of all the design disciplines, **landscape architecture has the deepest stake in this gap.** The discipline's primary substrate IS geospatial primary data. CAD-era and GIS-era DLA could afford to treat that substrate as a free public good (Landsat, NAIP, Census, FEMA flood maps, USGS topo, county parcel data — all once free, all once stable). **That settlement is dissolving in real time.** A landscape architect in 2026 who depends on: - Google Maps tiles for context imagery → priced out for any non-trivial workflow - Google Photos API for client-site imagery management → unscriptable - USFS BAER reports for burn-scar data → reduced or absent - NOAA climate data for site-suitability analysis → uncertain future - USGS topo for terrain → still mostly there but resource-constrained - ESRI / commercial cloud GIS → tenancy on the platform lord's land …has had the substrate quietly walked out from under them. The discipline has not yet articulated this as a crisis because the dissolution is slow and the alternatives are not yet visible. **Naming it on stage at DLA-2026 is the move.**
## The post-fire flood recovery case shows all three pathologies at once The wildfire example ([three-front-pedagogy.md](three-front-pedagogy.md)) is sharper *because* all three pathologies hit at once: - **Corporate capture (#1):** the citizens' own phone images of the burn scar are siloed in Google/Apple Photos, unscriptable for the watershed-scale coordination the recovery requires. - **Government absence (#2):** USFS BAER closes; FEMA exits; NOAA flood forecasting is degraded or absent; state monitoring inadequate for the 3–5 year flood-risk window. - **Sousveillance gap (#3):** residents have no integrated architecture for collectively observing debris-flow precursors, vegetation regeneration, infrastructure status, downstream impacts, and rebuild decisions across years. **All three of the substrate's legs are missing in the precise window where the architecture is most needed.** This is why the design problem has both a *design* dimension (governance, observation architecture, computational substrate) and a *political* dimension (data sovereignty, public-good infrastructure, citizen agency). The discipline has been trained to read the design dimension; the political dimension is what needs naming.
## What the architecture would have to do Sousveillance for landscape requires several composable pieces, most of which exist in fragmentary form and none of which compose yet: 1. **Edge-native capture** — phones, low-cost cameras, dashcams, drones, doorbells, sensor packs operating as parcelas (manifesto sense), storing locally, serving optionally. AnySurface's projector-camera pair is a small example ([anysurface-as-digital-landscape.md](anysurface-as-digital-landscape.md)). The geo.camera project is another. 2. **Acequia-governed federation** — peers federating into commons-governed observation networks. Citizen contributes locally, shares per agreement, retains sovereignty over their own corpus. 3. **Cryptographic provenance** — when an image is shared into the commons, its origin, capture context, and chain of custody are auditable. (This solves the deepfake/manipulation worry without surrendering the imagery itself.) 4. **Polymorphic delivery** — same observation served as image to the citizen, GeoJSON to the analyst, JSON-LD to the agent, point cloud to the modeler. See [polymorphic-get-content-negotiation.md](polymorphic-get-content-negotiation.md). 5. **Cross-scale composition** — neighborhood sub-acequias nest into watershed acequias nest into regional federations, without surrendering local agency at each layer (Ostrom's nested-enterprise principle, see [three-front-pedagogy.md](three-front-pedagogy.md)). 6. **Sym-sovereign identity** — observers are themselves identified in the commons by transaction-attested identity, not platform-issued account. The manifesto's third-path identity move. 7. **Mayordomo protocol** — coordination roles for the network that are *protocol* not *position*. Replaceable. Auditable. Reformable. **Each of these pieces is buildable today** with current technology. The integrating architecture is what does not exist yet. The thesis: landscape architecture is the discipline that should be building it, and students are learning how.
## What the students are doing (already) From [three-front-pedagogy.md](three-front-pedagogy.md): - **Front 1 (field)** = citizen phone imagery, federated capture, peer-synced storage. This is sousveillance in its primitive form. - **Front 2 (lab)** = the EmRiver streamtable, where ground-truth physics is verifiable and demonstrable. This is the *interpretability layer* for sousveillance data — what you can know with confidence. - **Front 3 (synthetic)** = AnySurface projector-camera, edge-deployed, browser-native, no install. This is the *room-as-substrate* version of sousveillance: the room observing itself with the inhabitants as co-authors. - **Tooling autonomy** = students using AI to author their own CV pipelines, ML classifiers, geospatial ABM. This is the *capability sovereignty* leg. Tools they own, deployable to the acequia substrate. Plus the 30+ domain federation already running sym-sovereign identity on the acequia substrate — a working layer of the sousveillance architecture, in production. **The architecture is being built incrementally, by students, on a working prototype, deployed to a working substrate, governed by a working four-hundred-year-old institution.** The work is not aspirational; the gap it fills is real; the dissolution of the alternative is current.
## The political-economic frame Step back: the dissolution of public earth observation + the corporate enclosure of citizen data + the absence of sousveillance architecture is **a single coordinated condition.** Not the result of any single conspiracy — the result of structural forces (capital concentration, austerity politics, autopoietic ontology — see [sympoiesis-over-autopoiesis.md](sympoiesis-over-autopoiesis.md)) that all push in the same direction. The same forces dismantling the federal civic-data infrastructure are the ones building the surveillance-capitalist matrix. The same vacuum that opens in (#2) is the one the platforms in (#1) move into. **Sousveillance is the only available structural counter** — and it has to be designed deliberately, because nobody is going to provision it from above. This is the manifesto's claim made operational for landscape: *the revolution is a URL, and where you host it*. For DLA, the URL is the geospatial observation endpoint, and the question is whether the citizens or the corporations host it.
## How this changes the talk Adds a third pathology to **Act 3 (Governance turn)**: 1. The matrix is brownfield (Forman applied to digital). 2. The government is absent (post-fire flood recovery; multi-year window with no incumbent). 3. **The data substrate is captured (corporate enclosure of imagery/location/observation; revoked APIs; no sovereign control). And there is no architecture for sousveillance.** The Ostrom-acequia third-way move now answers *three* failure modes, not one. The talk's diagnosis is more comprehensive; the prescription is the same; the working prototype demonstrates all three remediations simultaneously. The single-sentence version for stage: > *Your imagery is in someone else's cloud. Your home cameras feed Amazon and Google by default. Your government's eyes are being closed. Your own devices have full-duplex publishing capability that has been crippled by ISP contracts and OS defaults. There is no architecture for citizens watching back. This is the design problem.* (Or if 3-minute economy demands tighter compression — *"Your imagery is in someone else's cloud. Your government's eyes are being closed. There is no architecture for citizens watching back."* The longer version is for the paper / Q&A; the shorter is for the stage.)
## Connections - [three-front-pedagogy.md](three-front-pedagogy.md) — the prototype demonstrating sousveillance architecture under construction. - [sympoiesis-over-autopoiesis.md](sympoiesis-over-autopoiesis.md) — the metaphysical move that grounds why "self-sovereign data" is autopoietic-trapped and why sym-sovereign is the correction. - [cognitive-landscapes-and-three-intelligences.md](cognitive-landscapes-and-three-intelligences.md) — sousveillance is collective intelligence with explicit governance. - [anysurface-as-digital-landscape.md](anysurface-as-digital-landscape.md) — local edge-native capture and rendering; component of the architecture. - [spatial-cache-matrix-citizen-fires.md](spatial-cache-matrix-citizen-fires.md) — the spatial coordination layer. - [service-worker-as-edge-agent.md](service-worker-as-edge-agent.md) — browser-as-server; the deployable substrate. - [polymorphic-get-content-negotiation.md](polymorphic-get-content-negotiation.md) — same observation served to different consumers per Accept. - [Acequia manifesto](https://acequia.org/acequia-manifesto.md) — political-economic framing. - [Three-eras-of-web-topology.md](three-eras-of-web-topology.md) — Era 2's AdTech extraction = the same architecture as corporate data capture; Era 3 is the response.
## References (for the talk's confidence layer, not for the slide) - Steve Mann — coiner of *sousveillance* (1995, MIT). His [wearcam.org / wearable computing](https://wearcam.org/) work is the original reference. - Steve Mann, Jason Nolan, Barry Wellman — "Sousveillance: Inventing and Using Wearable Computing Devices for Data Collection in Surveillance Environments." *Surveillance & Society* 1(3), 2003. - Shoshana Zuboff — *The Age of Surveillance Capitalism* (2019). The corporate-enclosure half of the diagnosis. - Elinor Ostrom — *Governing the Commons* (1990). The third-way half. - The acequia precedent — 400 years, Northern NM.
## A sentence to keep > The geospatial substrate of digital landscape architecture has been captured from three directions at once: corporate enclosure of citizen imagery (down to the camera in the child's room), government dismantling of public observation, and the absence of any architecture for sousveillance — all on top of an architectural asymmetry that has progressively crippled every consumer device's full-duplex publishing capability. The discipline's intervention is not to lament the loss but to build the architecture the loss creates room for — and students, on three substrates, with edge-native tools already installed on every machine they own, are already doing it.