San Francisco

Original stage labels are preserved. Unverified patterns are research audit entries, not deployments.

[]

# San Francisco — Smart City Dispatch v0.2.3 research report

**Status:** PROVISIONAL RESEARCH · prepared 2026-10-11 · **2B — provisional**
**Boundary:** Consolidated City and County of San Francisco; not entire Bay Area, not SFO airport or adjacent Silicon Valley.

**Scores:** Type 20.83 (exact 20.83333333333333333333333333); Autonomy 12.50 (exact 12.50000000000000000000000000).
**Conclusion:** San Francisco receives provisional 2B: Waymo provides material autonomous mobility, and citywide AI-assisted government analysis meets a second Type 2 domain, but no second distinct domain demonstrates independently executed operating choices. Water metering adds connected capability, not Level C.

## Source audit
Discovered: 11; examined: 9; accepted: 9; rejected: 0; inaccessible: 2; distinct publisher families: 6. These are this research edition's unique original URLs, not historical Sites search totals.

## Eight-domain decision table
| Bucket | Type winning technology | T,c,g | Type pts | Autonomy winning technology | A,c,g | Autonomy pts |
|---|---|---|---:|---|---|---:|
| MOB | TECH-001 | 2,1,1 | 8.33 | TECH-001 | 2,1,1 | 8.33 |
| ENG | — | —,—,0 | 0.00 | — | —,—,0 | 0.00 |
| WAT | TECH-004 | 1,1,1 | 4.17 | TECH-004 | 0,1,1 | 0.00 |
| BLD | — | —,—,0 | 0.00 | — | —,—,0 | 0.00 |
| SAF | — | —,—,0 | 0.00 | — | —,—,0 | 0.00 |
| ENV | — | —,—,0 | 0.00 | — | —,—,0 | 0.00 |
| GOV | TECH-002 | 2,1,1 | 8.33 | TECH-002 | 1,1,1 | 4.17 |
| HLT | — | —,—,0 | 0.00 | — | —,—,0 | 0.00 |

**Sum, without per-domain rounding:** 20.83 Type, 12.50 Autonomy. The scoring formula is `100 × Σ_d 0.125 × max_i(g_i c_i T_i/3)` and its analogous A formula.
**Mandatory gates:** Type≥2 domains at T≥2 = ['MOB', 'GOV']; Autonomy C domains at A≥2 = ['MOB']. Result: Type2 True; autonomy C False. Type3 and D gates not demonstrated.

## Technology inventory, evidence and exclusions
### TECH-001 — Waymo fully driverless robotaxi mobility network
- **Bucket:** MOB; **status:** operating production; **T=2, A=2, c=1, g=1**; scope W.
- Coverage evidence: Commercial citywide rider service, tens of thousands of trips/week already in 2024; still active in 2026.
- Human authority / actuation: Driverless road operation without routine onboard intervention; remote assistance and geographic constraints. Vehicle steering, braking, routing and safe stopping.
- Basis and limitations: SFMTA traffic and temporary event disruptions do not by themselves change coverage W. Sources: SRC-001;SRC-009.
### TECH-002 — Citywide municipal staff AI decision assistance (Copilot)
- **Bucket:** GOV; **status:** operating production; **T=2, A=1, c=1, g=1**; scope W.
- Coverage evidence: Deployed access to ~30k staff across >40 departments, >1m messages July 2025–April 2026.
- Human authority / actuation: AI summaries, research and analysis; human experts must fact-check and approve outputs. No automatically authorized city-service execution.
- Basis and limitations: Usage uneven; T2 based on staff analytical assistance rather than independently validated outcome optimization. Sources: SRC-002.
### TECH-003 — 311 Enhanced Knowledge Base / AI knowledge search
- **Bucket:** GOV; **status:** partially verified rollout; citywide status uncertain; **T=2, A=1, c=.35, g=0**; scope D.
- Coverage evidence: 2025 soft-launch with test group, Aug 2026 city newsletter search excerpt claims expanded use.
- Human authority / actuation: Knowledge search assistance to service agents; agents take action. 
- Basis and limitations: g=0 pending accessible primary source confirming substantial use; GOV score instead uses citywide Copilot. Sources: SRC-003;SRC-010.
### TECH-004 — SFPUC near-universal wireless water metering
- **Bucket:** WAT; **status:** operating production; **T=1, A=0, c=1, g=1**; scope W.
- Coverage evidence: Nearly all ~178k water customers; separate leakage notification for >170k devices.
- Human authority / actuation: Automated readings and alerts; physical service decisions remain with customers/utility. Meter telemetry and notifications.
- Basis and limitations: Hourly data and simple thresholds do not establish contextual actuation. Sources: SRC-004;SRC-005.
### TECH-005 — SFPUC rule-based leak alert detection
- **Bucket:** WAT; **status:** operating production; **T=1, A=0, c=1, g=1**; scope W.
- Coverage evidence: 170k+ customers alerted where triggers occur.
- Human authority / actuation: Preset 48-hour continuous-usage thresholds and night deviations; human repair. Automated warnings (not leak shutoff).
- Basis and limitations: Same WAT bucket as meters, not double counted. Sources: SRC-005.
### TECH-006 — Citywide demand-responsive parking pricing
- **Bucket:** MOB; **status:** operating production; **T=1, A=0, c=1, g=1**; scope W.
- Coverage evidence: Quarterly rate changes across most SFMTA-managed metered blocks.
- Human authority / actuation: Fixed 60/80 percent occupancy band rule with periodic staff-controlled rate schedules. Meter prices periodically updated under fixed rule.
- Basis and limitations: Not A2; no independent strategy optimization above fixed rule. Sources: SRC-006;SRC-007.
### TECH-007 — Automated red-light and bus-lane enforcement
- **Bucket:** MOB; **status:** operating production; **T=1, A=0, c=.35, g=0**; scope D.
- Coverage evidence: Limited camera network and corridors; no substantial N/W demonstration.
- Human authority / actuation: Human due process on detected infractions. Automatic capture only.
- Basis and limitations: No additional score beyond MOB winner; fixed capture not Level C. Sources: SRC-008.

## Why not a higher classification
Type 3 requires three distinct material T3 service domains and a score≥30; not demonstrated. Level D requires three independent goal-directed adaptive domains and score≥30; not demonstrated.
this city has confirmed candidate autonomous mobility but no second distinct service domain satisfying the full test.

## Sensitivity, limitations, editorial check
The GOV Type 2 designation depends on accepting broadly used staff AI analytics as municipal operating decision support. If it is treated only as generic document drafting (T1), Type 2 breadth is unproven and the class would be downgraded; 311 EKB rollout cannot substitute without verified scale.
Reconfirm 311 knowledge-search use across city operations and separate direct operational decisions from generic AI drafting; validate Waymo route exceptions around temporary event geofencing.

## Source provenance
- SRC-001 [Waymo One open to everyone in San Francisco](https://waymo.com/blog/2024/06/waymo-one-is-now-open-to-everyone-in-san-francisco/) — accepted. Substantial tens of thousands weekly AV trips; service across city.
- SRC-002 [Municipal AI usage across >40 departments, July 2026](https://www.sfchronicle.com/sf/article/ai-city-worker-san-francisco-22326139.php) — accepted. City-shared data 30,000 staff license eligibility and over 1m interactions, with uneven active use.
- SRC-003 [2026 Budget and Performance presentation: AI 311 knowledge tool](https://media.api.sf.gov/documents/2.27.26_Budget_and_Performance_2026__Meeting_-_Presentation_FINAL.pdf) — accepted. 311 Knowledge Search initially soft-launched to test agents; not alone proof of full citywide rollout.
- SRC-004 [About your water meter](https://www.sfpuc.gov/accounts-services/about-your-bill/about-your-water-meter) — accepted. Nearly all of ~178,000 city water accounts have transmission devices.
- SRC-005 [Leak Alerts: Find and Fix Leaks](https://www.sfpuc.gov/conserve-water/leak-alerts-find-and-fix-leaks) — accepted. 170,000+ meters and 165,000+ customers; threshold-rule warnings, not independent operational actuation.
- SRC-006 [Citywide Meter Rate Adjustments July and October 2026](https://www.sfmta.com/notices/citywide-meter-rate-adjustments) — accepted. Parking pricing governed by fixed occupancy bands, staff quarterly price adjustments; no Level C.
- SRC-007 [Sec.402 citywide variable parking meter rates](https://codelibrary.amlegal.com/codes/san_francisco/latest/sf_transportation/0-0-0-52465) — accepted. Explicit predetermined 60/80-percent occupancy formula.
- SRC-008 [Automated red-light and transit-only lane enforcement](https://www.sfmta.com/getting-around/drive-park/red-light-camera-and-other-automated-enforcement) — accepted. Cameras operate with human/judicial enforcement processes.
- SRC-009 [Expanded San Francisco and Silicon Valley coverage](https://support.google.com/waymo/answer/16339356?hl=en) — accepted. Cross-boundary expansion should not be credited as in-city scale.
- SRC-010 [August 2026 City Administrator newsletter — 311 Enhanced Knowledge Base](https://www.sf.gov/office-of-the-city-administrator-august-2026-newsletter) — inaccessible. Search index snippet described EKB in use but direct original blocked by bot challenge; do not count as independently examined.
- SRC-011 [311 public service information](https://www.sf.gov/get-information-311) — inaccessible. Direct source blocked; public service available by alternate govt records.

## Scope and import rules
This is a research edition. Match existing canonical SCD city records, retain the source/technology/evidence IDs with edition-specific namespacing, preserve earlier versions, and recalculate deterministic scores before publishing. Do not treat search snippets or inaccessible PDFs as examined source material. Do not treat a score of 0 in an unsupported domain as proof of traditional/manual infrastructure.
21 capability tags describe technologies; they do not alter eight independent 0.125 scoring buckets. No inferred duplicate credit across systems.