Corrections

Corrected 2026-10-28: the rough 175,000–260,000 stall illustration mixed city-limit surface acreage with a ZIP-based garage estimate. It is now explicitly labelled mixed-geography, with conversions and exclusions. The acreage finding and numerical stall range are unchanged. Corrected 2026-10-29: Metro's bus bases came in the late 1970s and early 1980s, not the late 1970s and 1980s (BelRed Subarea Plan). The 38,092 downtown spaces were the Regional Council's 2010 survey as reported in the city's 2013 study, not a count the study made.

Corrected 2026-11-03: the page quoted the 2026–2032 affordable housing strategy as proposing 'evaluating whether the current incentive programs could be modified to better support larger units suitable for families'; those words are the Planning Commission minutes' summary of a 22 April 2026 staff briefing, not the strategy, which says 'Prioritize funding and incentives for larger units that meet the needs of families.' Figures unchanged. On 2026-11-01 the page removed the inference that buildings are on borrowed time where land assessments exceed building assessments; those values do not establish redevelopment economics or timing. Parcel counts and assessments unchanged. Seven earlier corrections. On 2026-10-29 the unit-mix chart caption said 10 of the 11 new BelRed buildings contain no three-bedroom homes; two of them have no bedroom breakdown at all, so 8 record none and 2 are unknown. On 2026-10-21 the parcel figures were corrected: the city's parcel layer still listed as BelRed 45 parcels that Wilburton's 2025 rezoning had moved into Wilburton, so BelRed is 465 parcels across 1.06 square miles, not 510 across 1.14, with land of $3,171M against $1,084M of buildings (was $3,441M and $1,138M); the 93 percent share and the unit-mix figures did not change. On 2026-10-07 the account of why the bonus is missing was corrected: the page called it a gap nobody caught, a drafting gap rather than a policy decision; the record shows staff proposed it in February 2023, commissioners asked for plan language in 2024 that staff softened, and a July 2026 request was left out while three companion items were added. On 2026-10-06 the explanation of why the bonus is missing was corrected: the page said both overlays descend from Part 20.25R and BelRed dropped the bonus, but 20.25R was Wilburton's code and BelRed's new part replaces its own; the bonus quote and one subsection citation were also fixed. On 2026-09-25 the delivered share moved from 1.0% to 0.9% and the citywide base from 3.3% to 3.0%, after Bellevue apartments were re-selected by parcel jurisdiction instead of ZIP code. The land figures were first measured through three census tracts covering 3.75 sq mi against BelRed's 1.30 — a corridor is the wrong shape for compact tracts — and every figure derived from them was retracted and remeasured parcel-exact. Separately, an earlier version reported BelRed pricing family-sized housing at one point per square foot against sixteen for Eastrail paving; those rates are Part 20.25C, the Wilburton Overlay. The corrected finding is that BelRed has no such provision at all.

Corrected 2026-10-27: the map and geographic method used broad latitude/longitude bands while describing Downtown and Overlake urban centres. Recounted the same 31 complexes using the City of Redmond's official Downtown Urban Center and Overlake Metro Center polygons and parcel area centroids. The counts and rounded 87% share are unchanged: 3,097 Downtown, 2,897 Overlake, 916 elsewhere. Replaced the map and renamed Southeast as Rest of Redmond. Earlier: published twice with the wrong headline. First as 2.5x, which used ZIP-based city assignment over a partial decade. Then the body was corrected to 2.4x and the writeup headline was left reading "a third of Redmond’s rate", which is 3x, so the piece contradicted itself. Original note: first published as 2.5x. That used ZIP-based city assignment over a partial decade; the per-capita comparison is 2.4x and the absolute one is 1.2x.

Corrected 2026-09-25: 780 is 737. Apartments were selected by ZIP code, which counted three Newcastle complexes (43 three-bedrooms) and missed eleven Bellevue ones with no ZIP or a mistyped one; they are now selected by the city on the assessor parcel record. Earlier: first published as one three-bedroom across the four highest-SCORING zones. The arithmetic was right but the label was not: those four zones are the four lowest-POVERTY ones, which swap Woodridge for Bennett. Ranked by score the finding reverses — 60 three-bedrooms across 1,041 units, above the citywide rate.

Corrected 2026-10-28: removed an unsupported claim that office floor areas differed by up to 15%; the named web-versus-brochure suite discrepancy remains. Also corrected the implication that sale marketing had begun and dated the city event permits to their actual seasons.

Corrected 2026-10-29: the page counted David Lawrence among eight fashion houses and 14 shops. The store closed on 15 April 2026 when its owners retired, though the center's directory still lists it. The courts hold seven fashion houses and 13 shops.

Corrected 2026-10-07: the page said Bellevue's apartment minimum was one citywide number, 1.0 falling to 0.5 in 2027. Downtown has had its own 0.5 minimum, and since 2025 state law has barred any minimum within half a mile of a light rail station. The unsourced $60,000 per stall was replaced with UCLA's 2026 figures. Earlier: reframed twice after outside review by a parking-reform researcher. The first pass fixed the body and left the trap in the headline: the title, the lead figure and the one-line claim all still led with ownership against the minimum, which is the calibration reading the review warned about, so a reader met the correction last and only if they finished. Retitled around the counterexample, and no-single-citywide-number merged in, since after the reframing the two made the same argument. Original note: reframed after outside review by a parking-reform researcher. The earlier version argued the minimum was too high because ownership sits below it; that concedes minimums should be calibrated to observed ownership, and the same logic raises Kirkland's to 1.44. The numbers were not at issue — the inference was. Now part of Bellevue's parking lots cover about 5 to 6% of the city.

Half of this is now wrong, and the wrong half was the headline. The corpus was a 38-of-765 street convenience sample and could not be counted; it is now a full sweep of every street in five cities, 827,057 permits, and Bellevue counts no longer need a records request. The labelling objection survived and grew: building on the complete corpus found four separate crosswalk defects, including a dwelling gate that existed, was tested, and was called by nothing, under which 39% of Bellevue's new single-family permits built no house. Now part of Redmond is half Bellevue's size and builds more apartments. Nearly all of them went up in two neighbourhoods..

Corrected 2026-09-09. The claim read "more parties than ever are asking Bellevue what they can build." The chart on this page refutes it: predevelopment filings peak around 2020 and 2025 sits below that peak. The original reached its conclusion by comparing 2010 to 2025 and stepping over the intervening high point, which is the failure this project's own conventions name first. The ratio of inquiries to filings, which is what the finding is actually about, is unaffected and is now the headline. The 2026 point is a partial year and is now labelled as one. Now part of What a Bellevue building permit costs.

Retracted in full on 2026-08-26, the same day it was written, before it was sent anywhere. It rested on 30 of BelRed's 301 documents. The full record of 301 documents refuted two of its three pillars: the school district had bought land in BelRed for an elementary school, and 'school district' appears ten times in the full record rather than zero. Retained rather than deleted because the failure is more instructive than the claim was. Now part of BelRed's draft code has no incentive for family-sized homes. It was asked for three times and left out each time..

Corrected 2026-09-25: first published as "Counted homelessness grew 67%. Crisis beds grew 20%.", which compared 2015 with 2024 across the 2022 change in counting method. Now compared within each method; the count outgrew the beds under both. Now part of About 1,900 people were homeless on the Eastside at the last count. Here is who they are, and how that number is made..

Our own fetcher built broken links: it added .pdf to any link that did not already end in it, so a Word file's link ended in .docx.pdf and returned an error. 307 of the 311 recorded failures had the doubled ending. The apparent decay with age tracked older matters carrying more Word files. All 303 documents were fetched again without a failure. The rule it cost: an error from your own fetcher is evidence about the fetcher until that has been ruled out. What survives is the reason for collecting at all: an archive taken now keeps what a portal may stop serving later.

Both halves failed a check against the archived tracker: it contains $16.50, $13.00, a $27-$42 range, ten percentages and three FAR figures, and this piece quoted one of those dollar figures four paragraphs later. The count of 74 is not derivable from the document text either. Caught before the claim went to an editor, in a pass that verified every figure against the corpus rather than against the writeup that produced them.

Three figures exist and two were paired across denominators: 654 three-bedrooms is 3.1% of the 20,869 units in complexes of 20 or more, while 780 is three-bedroom-and-larger across all 21,831 units. Both were correct; putting them side by side was not.

The two rankings agree on three zones of four and diverge on the fourth: Woodridge outscores Bennett 80 to 73 but has three times the low-income share, and holds 811 units and 59 three-bedrooms. Every number was right and the grouping label was not. Stated by score the finding reverses — 5.8% against a citywide 3.1%. The scarcity tracks affluence.

The first figure used ZIP-based city assignment and a partial decade. Per-capita is the comparison that holds.

The tracts covered 3.75 sq mi against BelRed's 1.30, and BelRed is a corridor while tracts are compact — wrong in shape before wrong in size. Every figure derived from them was retracted.

Detailed race and origin coding changed at the 2022 vintage. Three cities step in the same year. Margins of error did not catch this and could not.

Corrected 2026-10-15: that cause was wrong too. There was no methodology change. The Census Bureau renumbered table B02015 at the 2022 release, and our script read the 2022 column numbers for every year, so its 2019–2021 "Chinese" figures were Asian Indian and Filipino. Read with each year's own columns, Redmond's Chinese population did not fall, and no city steps at 2022. A margin of error cannot catch a misread column; checking each year's column labels does, and our tests now do.

Framing minimums as something to calibrate against observed ownership concedes the premise, and the same logic raises Kirkland's minimum to 1.44. Ownership also responds to the built environment and to parking supply, so it cannot forecast behaviour under a different rule. Corrected after outside review by a parking-reform researcher; the numbers were not at issue, the inference was.

The first number compared against different, pricier buildings.