Measure the ability to reach everyday destinations, not simply how much road a city has or how many people drive.
Measurement baseline v0.1, retrieved : 380 Digital Atlas observations for the 20 largest US cities by Census Vintage 2025 population. This is a simple analysis published for general information only. It is not advice, not a prediction and not for use in any decision. Read the disclaimer.
Baseline captured; full indices not yet scored.
All 20 cities have 19 returned measures. Of 380 observations, the atlas labels 280 “available” and 100 “proxy”. These are supporting indicators, not validated access scores. Comparable parking, safety, travel reliability and destination-access inputs remain missing.
Choose any two of the 20 cities; the paper sets New York against Fort Worth.
Transit, walking and cycling together represent 54.8% of workers’ commutes in New York and 1.9% in Fort Worth. Car commuting is 25.9% and 82.3%, respectively. These describe behaviour; they do not measure which destinations are accessible.
The share of non-home commuters travelling at least an hour is 25.3% in New York and 7.8% in Fort Worth. Shorter commutes alone do not establish better access, and these all-mode times cannot rank driving congestion.
How each city commutes, any one of the 19 captured measures, or two measures side by side. Numerical order is not a Driveability or Movability ranking.
| Population order | City | Car | Public transit | Walk and bike | Work from home | Other means | No-car households | An hour or more |
|---|---|---|---|---|---|---|---|---|
| 1 | New York | 25.9% | 44.0% | 10.8% | 16.9% | 2.4% | 55.3% | 25.3% |
| 2 | Los Angeles | 68.5% | 6.1% | 3.8% | 19.5% | 2.1% | 11.9% | 12.0% |
| 3 | Chicago | 52.7% | 17.5% | 7.1% | 20.7% | 2.0% | 25.9% | 13.5% |
| 4 | Houston | 78.6% | 3.0% | 2.4% | 13.1% | 3.0% | 9.4% | 7.8% |
| 5 | Phoenix | 74.9% | 1.9% | 1.9% | 19.1% | 2.1% | 6.6% | 5.8% |
| 6 | Philadelphia | 53.7% | 16.7% | 8.9% | 18.4% | 2.3% | 27.9% | 12.0% |
| 7 | San Antonio | 80.5% | 2.0% | 2.1% | 13.7% | 1.6% | 7.6% | 5.5% |
| 8 | San Diego | 70.3% | 2.7% | 4.1% | 21.2% | 1.7% | 6.6% | 4.5% |
| 9 | Dallas | 78.3% | 2.1% | 2.2% | 16.0% | 1.4% | 8.5% | 6.8% |
| 10 | Fort Worth | 82.3% | 0.5% | 1.4% | 13.8% | 2.0% | 5.6% | 7.8% |
| 11 | Jacksonville | 79.8% | 1.0% | 1.6% | 15.6% | 2.1% | 7.3% | 3.8% |
| 12 | Austin | 62.1% | 2.1% | 3.3% | 30.7% | 1.7% | 7.2% | 4.6% |
| 13 | San Jose | 73.6% | 2.4% | 2.3% | 20.3% | 1.4% | 6.2% | 7.3% |
| 14 | Charlotte | 65.9% | 2.0% | 2.0% | 28.7% | 1.4% | 6.5% | 4.4% |
| 15 | Columbus | 75.2% | 2.0% | 3.2% | 18.3% | 1.3% | 8.7% | 3.2% |
| 16 | Indianapolis | 81.3% | 1.4% | 2.1% | 14.1% | 1.1% | 8.4% | 4.6% |
| 17 | San Francisco | 34.6% | 18.7% | 12.7% | 31.0% | 3.1% | 30.2% | 11.3% |
| 18 | Seattle | 41.4% | 11.9% | 10.6% | 34.8% | 1.4% | 19.0% | 5.6% |
| 19 | Denver | 62.5% | 3.8% | 5.1% | 27.3% | 1.3% | 10.3% | 5.1% |
| 20 | Nashville | 74.6% | 1.5% | 2.2% | 20.4% | 1.4% | 6.7% | 4.9% |
Car is drive alone plus carpool; walk and bike combines those two modes; other means is the rest (taxi, motorcycle and other). No-car is a share of households; an hour or more is a share of commuters who travel to work. Indianapolis and Nashville use consolidated-government balance geographies. Atlas aggregates tract data using coverage fractions; these are approximate city estimates, not direct ACS place tabulations, and city boundaries and land areas differ substantially.
Workers 16+; ACS B08301. Commute mode, not service reliability. Source: ACS 2020–2024. Atlas status for all 20 cities: available; aggregation rule share.
| City | Public transit (%) | |
|---|---|---|
| New York | 44.0% | 2,237 |
| Los Angeles | 6.1% | 1,110 |
| Chicago | 17.5% | 794 |
| Houston | 3.0% | 925 |
| Phoenix | 1.9% | 378 |
| Philadelphia | 16.7% | 389 |
| San Antonio | 2.0% | 332 |
| San Diego | 2.7% | 337 |
| Dallas | 2.1% | 381 |
| Fort Worth | 0.5% | 216 |
| Jacksonville | 1.0% | 208 |
| Austin | 2.1% | 253 |
| San Jose | 2.4% | 231 |
| Charlotte | 2.0% | 254 |
| Columbus | 2.0% | 261 |
| Indianapolis | 1.4% | 239 |
| San Francisco | 18.7% | 240 |
| Seattle | 11.9% | 178 |
| Denver | 3.8% | 176 |
| Nashville | 1.5% | 168 |
| City | Drive alone | Commutes of an hour or more |
|---|---|---|
| New York | 21.6% | 25.3% |
| Los Angeles | 59.8% | 12.0% |
| Chicago | 45.2% | 13.5% |
| Houston | 67.9% | 7.8% |
| Phoenix | 64.1% | 5.8% |
| Philadelphia | 46.0% | 12.0% |
| San Antonio | 68.2% | 5.5% |
| San Diego | 62.6% | 4.5% |
| Dallas | 66.8% | 6.8% |
| Fort Worth | 71.8% | 7.8% |
| Jacksonville | 70.8% | 3.8% |
| Austin | 55.8% | 4.6% |
| San Jose | 63.0% | 7.3% |
| Charlotte | 57.3% | 4.4% |
| Columbus | 67.4% | 3.2% |
| Indianapolis | 71.8% | 4.6% |
| San Francisco | 28.6% | 11.3% |
| Seattle | 36.7% | 5.6% |
| Denver | 56.1% | 5.1% |
| Nashville | 66.5% | 4.9% |
The proposed weights below total 100 for each index. They are a research design, not existing Digital Atlas scores. Each component needs consistent data across all 20 cities before a composite can be published.
Not scorable yet: 90 of its 100 weight points have no comparable inputs and 10 rest on a proxy.
Population-weighted share of grocery, primary-care and school categories reachable in 20 min; jobs within 30/45 min reported separately.
Missing: citywide OD access.
For fixed origin–destination pairs, peak P95/P50 travel time; publish peak/free-flow ratio separately.
Missing: observed travel-time distributions.
Three-year fatalities + serious injuries involving motor vehicles per 100m vehicle miles; publish fatalities separately.
Missing: consistent exposure and injury data.
Destination-weighted median legal parking search + final walk time; price per hour separately.
Missing: comparable parking observations.
Population-weighted share retaining access to the destination basket after a critical link closure.
Proxy only: gateway redundancy and graph-bridge share.
Not scorable yet: 75 of its 100 weight points have no comparable inputs and 25 have context only.
Population-weighted destination-category coverage using the fastest feasible walk, bike, transit or car journey; jobs separately.
Missing: comparable access by mode.
Same destination coverage using walk/bike/transit only; share of residents with two feasible modes reported separately.
Context only: commute modes; no-vehicle households.
Share of planned all-day and evening transit journeys arriving within 5 min of schedule; frequency and service span separately.
Missing: consistent service and reliability sample.
Fatalities + serious injuries per million trips by mode, with absolute counts and vulnerable-user rates.
Missing: mode-specific exposure.
Cost of feasible monthly essential travel basket / household disposable income, by income quintile.
Missing: consistent travel-cost basket.
D = Driveability; M = Movability. Use fixed, published policy anchors to scale indicators to 0–100; reverse lower-is-better indicators. Do not rescale to the best and worst city in each new year. Do not fill missing values with zero or silently redistribute their weights.
Follow each major road or transit intervention from before opening through 36 months. Compare changes with matched untreated areas and account for population, jobs, fuel prices and land-use changes.
The current cross-sectional baseline cannot demonstrate induced demand or establish causality.
20 largest incorporated cities / consolidated balances by Census Vintage 2025 population; city boundaries, not metros. Indianapolis and Nashville use balance geographies.
Atlas census_place_tracts contexts with coverage fractions; source boundaries TIGER 2024. Approximate city aggregates; dropped cells and zero-weight tracts retained in source records.
Retrieved 2026-11-02. ACS 2020–2024, NTM 2026-03-09, roads full_v2 with no source date returned. This is a mixed-vintage baseline, not a single-year observation.
Both indices not scored for every city. Weights above are proposed analytical design, not official or validated Digital Atlas scores. Missing inputs are not zero and weights are not redistributed.
Before launch, preregister fixed policy anchors L/U per indicator. Higher-better score = 100 × clamp((x−L)/(U−L)); reverse for lower-better. Weighted sum only after all five components pass coverage checks.
Proposed: at least 90% of resident population represented for each component and same sampling period across all cities. Publish coverage and uncertainty; no rank for overlapping confidence intervals.
Use population-weighted residential block/grid origins; allow destinations beyond city limits. Grocery, primary care and schools as separate categories, one required per category; jobs use counts, not category share.
Use common weekday peak/off-peak and weekend windows over four representative weeks. Include walking, waiting, transfers and parking; accessible routes separately. Do not use a city centroid as the city.
Report median and 10th-percentile access plus low-income, disability and no-car-household gaps. City averages alone conceal exclusion.
Capture VMT/person, trip distance, lane-km, mode share and access before works and 6/12/24/36 months after; compare matched untreated corridors, adjusting for jobs, population, fuel prices and land use.
Lasting improvement means access persists while travel time/cost and safety improve. Rising VMT with fading access gains is an induced-demand warning; this cross-section cannot establish causality.
Freeze this cohort for trend analysis; refresh membership separately each Census release. Requery Atlas releases, retain metric IDs and hashes, append vintages rather than overwrite; QA before republishing. No refresh is scheduled.
Atlas returned no aggregate MOE for these 380 values. Available is a service status, not proof of unbiased estimates. Small differences should not be ranked.
Cohort. Census Vintage 2025 city estimates, released 14 May 2026. Top 15 checked against the release; ranks 16–20 follow the published reproduction of the Census ranked table. These are cities, not metropolitan areas.
Atlas. 20 explicit Census place contexts; metrics.fetch with MOE and support requested. Releases: people comprehensive_v3 (ACS 2020–2024), roads full_v2, mobility national_tracts_v2 (NTM 2026-03-09), boundaries layer00_geography_v2_polygons. The download below gives every value with its metric ID, Atlas status, aggregation rule and the tracts used and dropped.
Uncertainty. No aggregate 90% margins of error were returned. Five measures per city are tract-median proxies. “Available” does not mean exact, complete or contemporaneous.
Roads. No source timestamp was returned for roads full_v2. Legal cycle access is not protected infrastructure or comfort. Modelled freeway-entry time is uncongested. Graph bridges are network cut edges, not physical bridges.
Transit inventory. NTM coverage is incomplete and overlapping feeds can duplicate records. Boarding-stop counts and all-stop density have different universes. Inventory measures are retained for audit and are excluded from transit-quality conclusions.
Excluded measure. Atlas summed tract-level mean 15-minute walking-reach lengths at city scale. That result was rejected as a city-access indicator because those means are not additive.
Download the 380 observations (CSV)
Prepared using Propheus Digital Atlas and documented Census cohort sources. This is a reproducible baseline and proposed measurement specification; no official index, causal conclusion or automated refresh is claimed.
Case studies are simple analyses that show what Digital Atlas data can do. Read them on these terms:
Use of this website, including case studies, is subject to our Terms of Service. Read the Terms of Service.