FAQ — anticipated reviewer questions

This page answers the questions a planner, economist, or hiring reviewer is most likely to press on. It is deliberately candid about what the analysis does and does not establish. Method detail and data lineage are on the Methods & Sources page.

Capacity

Is “soft-site capacity” the same as net-new homes?

No — and the memo no longer calls it that. Soft-site capacity is the gross zoned capacity on parcels that are currently underbuilt (vacant, surface parking, or low-rise). It is not net of the existing units on those parcels. Most soft sites are non-residential (parking lots and low-rise commercial), so the subtraction would be small, but the honest framing is “zoned capacity on redevelopable land,” not “new units delivered.” A parcel-by-parcel subtraction of existing units is the right next refinement.

Why apply a uniform 350 du/ac to the whole Downtown core?

Downtown (DC) density is governed by the Envision 2040 General Plan Downtown designation (with the Diridon Station Area Plan on top of it), which permits up to 800 du/ac in the core. We apply a deliberately conservative 350 du/ac cap instead: 800 du/ac towers are rare in practice, and height limits, FAR, FAA/airport surfaces, historic districts, parking/loading, and tower-spacing rules typically bind well before the legal maximum. Even 350 is a ceiling, not a typical yield — which is why every number is labeled theoretical zoned capacity and why the soft-site figure, not the 120,000 gross, is the one the memo leads with. A height/FAR-aware envelope would refine the Downtown number and is the natural next step.

How are Planned-Development (PD) parcels handled?

Each parcel’s zoning is stripped of its (PD)/(CL) overlay to its base district (e.g., DC(PD) → DC). I checked whether the approved PD densities (PDDENSITY) could be used instead: within the station area they are null or zero for every target parcel (including all 101 DC(PD) parcels), so the base General-Plan/Title 20 maximums are the only available basis. Where a city later publishes site-specific PD entitlements, those should override the lookup.

Is the “City plans ~34,300 homes” comparison apples-to-apples with the envelope?

Yes — by construction. An earlier draft compared the DSAP’s ~12,900-home program (a ~262-acre plan area) against the full ring’s envelope, which overstated the gap. The current figure sums the City’s own Growth Areas 2040 programs across every plan area intersecting the ring — Downtown, the DSAP, and six urban villages — each counted only for the share of its land inside the 1-mile circle, so plan and envelope share the same geography. Two adjustments are footnoted in the memo: the DSAP polygon overlaps the Downtown growth area (its footprint is subtracted from Downtown’s before apportioning), and the amended DSAP (2021) program of ~12,900 homes replaces the pre-amendment figure (2,710) still carried in the City’s layer. Per-area detail: output/tables/growth_areas_in_ring.csv.

Why include Downtown but exclude R-M and other residential zones?

The question the memo asks is where the station-area upzoning lever is — and that is the Downtown core plus the mixed-use/urban-village districts. Lower- density residential zones like R-M are largely built out and are not where a TOD capacity debate plays out, so including them would add existing-neighborhood capacity that dilutes the station-area story. It would be a reasonable sensitivity to add; it does not change the finding that Downtown dominates.

Is UVC really zero residential capacity?

Yes — under Title 20 Table 20-136, the Urban Village Commercial district does not permit stand-alone residential, so it contributes 0 to residential capacity. It is retained in the parcel set for completeness and shows as 0.

Soft sites

Why footprint coverage instead of assessor improvement-to-land ratio?

The improvement-to-land ratio is the textbook soft-site measure, but Santa Clara County does not publish parcel-level assessed values — they are a paid bulk-data product. Building-footprint coverage from OpenStreetMap is the open, fully reproducible substitute. The Methods & Sources page shows the soft-site total is stable across coverage thresholds (~34.2k–53.9k from 5%–25%).

Could non-developable land (rail, the station, civic, parks) inflate the number?

Partly, and this is the figure’s main weakness. Parks, public/quasi-public, and agricultural zones (OS, PQP, A) are already excluded — they are not housing-permitting zones, so they never enter the capacity set. The residual risk is large DC-zoned parcels that are functionally rail right-of-way, the station, or other civic uses but carry Downtown zoning: 219 of 351 soft sites have zero detected building coverage (~21,900 units — about half the soft-site total), and the largest few are multi-acre Downtown parcels. Excluding every zero-footprint parcel gives a conservative floor of ~20,600 units, so the defensible range is ~20,600–42,500. A parcel-level ownership/use screen (civic/rail, active entitlements) would tighten this — the City’s Major Private Development Footprints layer can flag soft sites with active projects (several likely overlap Google’s Downtown West) but records no unit counts, so netting them out needs hand-collected data. Until then the memo reports the range and treats the Downtown figure as a ceiling. The Methods & Sources page lists the top soft sites with their coverage and a verify/underbuilt flag so the parcels driving the total can be checked directly.

Is “38% underbuilt, much of it surface parking” a verified land-use statistic?

Partly. The 38% is measured: the share of Downtown’s zoned land with less than 15% building coverage. The “much of it surface parking” is an inference from that low coverage, not a classified land-use tabulation — which is why the memo says “underbuilt” for the number and names parking/vacant lots as the likely composition rather than asserting a verified split.

How complete is OpenStreetMap downtown, and what if footprints are missing?

OSM building coverage in downtown San José is strong, but it is not perfect. Missing footprints would over-count soft sites (a built parcel read as empty) — which pushes in the same direction as the non-developable-land issue above. Both are reasons the soft-site capacity is best read as an upper bound rather than a point estimate.

Equity

Are the demographics for residents inside the 1-mile ring, or whole tracts?

Whole census tracts that intersect the ring, weighted by occupied housing units — so the profile describes the surrounding neighborhoods, not only the population strictly inside the circle. Tracts are coarser than the parcel buffer; apportioning ACS to the buffer is possible but adds false precision to margin-of-error-laden estimates.

Could a high-income, renter-heavy Downtown tract get flagged “vulnerable”?

Yes — the renter and no-vehicle flags can fire in an affluent downtown tract, and an earlier version of this score had exactly that problem: high-income downtown tracts out-scored genuinely low-income ones. The score now includes an explicit low-income flag (median household income below the citywide median) alongside the City’s Equity Index, so income directly enters the measure. Under the corrected five-flag score, ~22% of soft-site capacity sits in higher-vulnerability (3+) tracts — including the station area’s low-income tracts — rather than the misleadingly reassuring ~4% the four-flag version produced. The score measures exposure, not a prediction of displacement.

Which ACS vintage is this, and is it current?

The figures are from the ACS 2019–2023 5-year estimates (year=2023), the most recent 5-year release. The Census data API requires a key, read at runtime from the CENSUS_API_KEY environment variable or the macOS Keychain and never written to disk. The structural facts the memo relies on — a renter-majority, transit-dependent downtown — move slowly across ACS vintages.

Scope & feasibility

Where is the pro forma / financial feasibility?

Out of scope, by design. This is a capacity and equity screen, not a feasibility study. It answers “where could homes go and who lives there,” not “what penciling rents, construction costs, parking, impact fees, and inclusionary requirements would make a given site build.” A residual-land-value or achievable-density sensitivity is the natural next phase and would pair well with the soft-site list.

Reproducibility

Can a reviewer regenerate these numbers?

Yes. environment.yml pins the spatial stack; the four scripts run in order (02→05) from code/. The pipeline now fails fast — it asserts the station selection returns the expected order of magnitude (~5,279 parcels in 1 mile, ~1,199 housing-permitting parcels, ~13 tracts) and writes outputs atomically (temp file → replace), so a bad run cannot silently overwrite good outputs. Generated tables/figures are committed so the published memo is always traceable to a specific pipeline run.