Corrections & Updates

An earlier version said Buncombe County cannot separate the occupancy tax short-term rentals pay from what hotels pay, and put the rental share at $232 million a year and about $14 million in tax. We filed a public records request to test the first claim, and the county's answer went the other way. A hotel-versus-vacation-rental breakout does exist. The county compiles it and publishes it every month in the tourism authority's board packets, and it has since at least 2017. That made the second claim checkable too, and it was high: the ~$14 million was a Step Up AVL derivation from a private scraping estimate, and it was roughly the fiscal-2022 level. The county's own figures put vacation-rental lodging sales at $129.4 million in fiscal 2025, about $7.8 million in tax, and $173.8 million through the first eleven months of fiscal 2026, about $10.4 million. The post-Helene downturn hit this segment hardest and the piece had not carried that through to its headline number. Both figures now come from the county's series rather than a derivation, with the vintage stated. The article's argument is unchanged and better supported: the same response confirmed, from the county rather than from 2023 secondhand reporting, that it has not audited a single occupancy-tax remitter in five fiscal years, that platforms report only total gross receipts with no list of the properties behind them, and that the new breakout does not separate platform money from direct remitters. The piece now also reports the county's count of 374 active occupancy-tax accounts as of June 30, 2026.

An earlier version said that across the Carolinas about 45.5 percent of people facing hunger likely earn too much to qualify for SNAP, and that this was higher than the national share. Both halves were wrong, and not because the data changed. SNAP income limits are set by each state, and Feeding America measures each state against its own: North Carolina's limit is 200 percent of the federal poverty line, South Carolina's is 130 percent. The 45.5 figure was the plain average of the two states' percentages, which combines two different measurements into one that describes neither. It also pointed the wrong way for the state this site covers. North Carolina's share has been below the national one in both years we have cited, 38 percent against 44 nationally for 2023 and 40 percent against 48 for 2024. The combined figure has been removed, and the piece now gives the national share and the North Carolina share separately, with the reason they cannot be added. The paragraph's point, that food banks catch the households SNAP's income limits miss, is unchanged. A second copy of the same combined figure in The People Just Above the Line has been removed as well.

The sources note said the federal count of public-school students without stable housing, 1,548,191 in 2023-2024, was a 12.6 percent rise in two years. It is a rise of 12.6 percent in one year. Two years is 28.4 percent. The wording came straight from the source, which describes the same change as an increase "in the past two reporting years," counting data points rather than elapsed years; the same document elsewhere gives the annual increase as 12.63 percent. Quoting it accurately is what produced the error, so the note now gives both spans and explains the source's wording. The same sentence also called the figure the highest since national reporting began in 2004. That may well be true, but it is not stated in the document we hold, so it has been removed rather than left unsourced. The student count itself, and the finding that about three in four of those students were doubled up, are unchanged.

An earlier version said Buncombe SNAP benefits "are spent at local grocers and tailgate markets" at about $60 million a year. The figure was derived from a count the county has since moved past: about 29,000 residents on SNAP in spring 2025. As of June 2026 the number is about 25,000, so the money flowing to local stores now runs closer to $50 million a year. The piece now gives both the current count and the earlier one, and the corrected figure matches the companion piece, "Who Loses SNAP Now," which had already been updated. The point is unchanged: as the rolls shrink, that spending leaves local tills too. The error was a dated figure left in a present-tense sentence.

An earlier version described the housing authority's new work requirement as 15 hours a week of work, with exceptions for the elderly, the disabled, and primary caregivers. Both halves were off, in opposite directions. School, job training, volunteering, self-employment and gig work all count toward the 15 hours, and there is no minimum income. The caregiver exemption is narrower than it sounded: it covers a head of household caring for a child under six. The piece now follows HACA's own resident FAQ of June 30, 2026, and cites it. The phase-in dates were right and are unchanged.

An earlier version said the ROAD to Housing Act lets a city waive the cap on how much emergency-shelter grant money can go to beds and street outreach. A city can only ask. HUD approves or denies within 60 days, must deny a request from a city that clears camps without offering people housing, and the authority runs only for fiscal years 2027 through 2030 (H.R. 6644, sec. 503, adding 42 U.S.C. 11373(f)). The piece was written from the Senate committee section-by-section, as its own footer warned it might need checking; it now cites the enrolled text.

An earlier version said the drug-free zone around a domestic-violence shelter was the interior of the building only. The ratified bill also covers any outdoor area that can be reached only by going through the building, so an enclosed courtyard or back patio is inside the zone (House Bill 437, G.S. 90-95(e)(8a)a.2.). The body and the sources note now say so. There is still no hundred-foot ring for these shelters, and the rest of the piece is unchanged. Caught by a guard that checks each stated rule against the ratified text rather than a summary of it.

An earlier version said a local government could open a sanctioned campsite only if a state agency signed off within 45 days, which made the deadline a hurdle the city could fail. The ratified bill runs the other way. The state Department of Labor has 45 days to act on a complete request, and the site is deemed certified on the forty-fifth day if the department does nothing (House Bill 437, new G.S. 160D-917(e)). Certification is still required, and the other conditions are unchanged. The piece now says so. The argument is unchanged: the cost and the standards are the barrier, not the state's signature. Caught by a guard that checks each stated rule against the ratified text rather than a summary of it.

The estimate that HUD's replacement funding rules could put at least 97,000 people at risk of losing their housing was credited to the National Low Income Housing Coalition. The estimate is the National Alliance to End Homelessness's, from its analysis of the FY2026 funding notice; NLIHC carried it. The figure was right, the source was not, and the piece now credits the Alliance. The argument is unchanged. Caught by a full re-verification of the article's claims.

An earlier version said the region's food bank served about 158,000 pantry visits a month before Hurricane Helene, and more than 190,000 by June 2025. Neither figure could be traced to MANNA FoodBank or to any published report, and the 158,000 appears to have come from a count of people served, not visits. The piece now uses only figures we can cite: more than 170,000 monthly visits in September 2025, "the highest level in the organization's 42-year history" (Community Foundation of WNC); roughly 195,000 by that November (828 News Now); and more than 200,000 now (MANNA). The claim that visits are "more than a quarter higher than before the storm" was removed with the unsourced baseline. The point is unchanged: pantry traffic set a record and has not come back down. Caught by a three-model verification pass.

The home page's headline figure said 200,000+ Western North Carolina neighbors face food insecurity. The number is real but it counts something else: MANNA FoodBank reports that its pantries "are visited more than 200,000 times every month," and a household can visit many times. MANNA's own count of people is over 120,000 facing food insecurity across its region. The figure stays and the label is now correct: monthly visits to the region's food pantries. Every article on this site already used it that way; the home page was the one place it read as a headcount.

An earlier version said North Carolina "already puts more than $800 million a year into mental health." The figure is right; the framing was not. The more than $800 million the piece's own sources describe (NC Health News; the Governor's office) was invested in the state mental-health system in the prior budget cycle, a one-time commitment, not an annual one. The text now says so. The point is unchanged: House Bill 1104 reworks involuntary commitment while adding no new money. The error was caught by a new automated check that compares shared figures across every page of this site.

Earlier versions said roughly $250 million in hospitality investment and 1,200 new hotel rooms were entering the Asheville market across 2025 and 2026. That pair traced to no findable source. The pieces now use the figure city records support: nearly 1,700 new hotel rooms built, under construction, or in the development pipeline since the city lifted its hotel moratorium in February 2021 (City of Asheville records, via Asheville Watchdog). The first piece's title changed to match. The argument is unchanged: the boom is real, and the workforce to staff it cannot afford to live near it.

Five pieces said Buncombe County pantry visits were up about 20 percent since Hurricane Helene. We could not trace a Buncombe-specific figure to any source. The pieces now state what MANNA FoodBank's own reporting supports: monthly pantry visits across its WNC region rose from about 158,000 before the storm to more than 200,000, an increase of roughly a quarter, with no return to pre-storm levels. Superseded July 9, 2026: the roughly 158,000 pre-storm baseline used here could not itself be traced to MANNA or any published report and was retired (it appears to have been a count of people served, relabeled as visits). The "increase of roughly a quarter" framing was dropped with it, and the affected pieces now use only citable visit counts. See the July 9, 2026 correction above.

An earlier version anchored the 1980s rise in homelessness at "roughly 125,000 (1980)," a number that traces to no primary source. The text now uses the first federal estimate on record: 250,000 to 350,000 (HUD, 1984), rising to an estimated 500,000 to 600,000 by the late 1980s. A header note claiming early word processors did not recognize "homelessness" as a word was also unverifiable and was replaced.

An earlier version gave exact SRO counts for New York (129,000 rooms in 1960 to 25,000 by 1978) and Los Angeles that we could not trace to primary sources. The text now uses the documented record: New York City engineered the loss of more than 100,000 SRO units by 1985, and owners destroyed two-thirds of the remaining stock between 1976 and 1981 alone.

Earlier versions presented the Denver Social Impact Bond trial's cost comparison ($25,554 for the control group vs. $18,678 for people placed in supportive housing) as savings "with the housing itself included." The Urban Institute's evaluation says otherwise: those figures are use of other public services (jail, ER, detox, shelter), and the roughly $6,876-a-year offset covered about half of the program's own cost, not more than all of it. All four pieces now state the offset and the program cost honestly. The argument is unchanged: housing people returns a large share of its cost in avoided public spending and keeps 77 percent stably housed at three years, but it is not free.

Earlier versions framed Battery Park Apartments and the city's public housing as serving "the same kind of neighbor," presenting the contrast in their condition as a near-controlled experiment. Battery Park is in fact restricted to low-income seniors (age 62 and over), while public housing serves families and individuals of every age. Both pieces now disclose that difference. The argument is unchanged: the gap in condition traces to the financing platform each building can reach, not to the residents.

Two figures from the Harvard Joint Center for Housing Studies were mislabeled "as of 2024." The cost-burden shares (58 percent of older renters, 43 percent of older homeowners with a mortgage) are 2023 data; the count of about 11.2 million cost-burdened older adults, up from 9.7 million in 2016, is from 2021. The older-homeowner share was also corrected from 44 to 43 percent. Separately, the Urban Institute's 37 percent rise in older-adult homelessness between 2019 and 2022 refers specifically to sheltered homelessness, and the text now says so.

The musical-chairs framework was misattributed. It belongs to the sociologist Kay Young McChesney, "Family Homelessness: A Systemic Problem," Journal of Social Issues 46, no. 4 (1990). An earlier version misnamed her and conflated her paper with a separate 1990 editorial of a similar title by Elliott D. Sclar in the American Journal of Public Health. Both are now cited correctly.

An earlier version said food insecurity "roughly doubles" the odds of depression. The meta-analysis cited (Pourmotabbed et al., Public Health Nutrition, 2020) supports about 40 percent higher odds overall, and about 75 percent higher for adults 65 and over. The text and footnote now match the study.

The national supply figure was corrected from 37 to 35 affordable and available homes per 100 extremely low-income renter households, per NLIHC's The Gap (2026). The 37 matched no edition of the report.

Earlier versions described the 1980s federal housing cuts with three differing sets of figures drawn from different budget series. All three pieces now use one sourced framing: budget authority for subsidized housing peaked at about $82 billion in 1978 (in constant dollars) and was cut by more than half in the early 1980s (NLIHC / Cushing Dolbeare, Changing Priorities, 2002).