Local officials announced in 2013 that 54,000 people in Los Angeles city and county were homeless, an increase of 15% from two years earlier.
The jump bucked national trends, and suggested the region's campaign against homelessness might be heading in the wrong direction.
But earlier this year, federal officials challenged 18,000 of that total — those whom L.A. considers "hidden homeless," detected through a random telephone survey designed to locate people scattered in hard-to-find places across the county's 4,000-square-mile sprawl.
The U.S. Department of Housing and Urban Development says it lost confidence in the survey methodology. The Los Angeles Homeless Services Authority says its original count is the most accurate and comprehensive available.
Either way — 54,000 or 36,000 — Los Angeles remains second to New York as the region with the most homeless residents. And officials say the uncertainty won't affect L.A.'s share of funding for homeless housing and services, which is not based on homeless numbers but on poverty and housing data.
But the disagreement points to the challenges in trying define the size of — much less solve — the homelessness problem, and has prompted some to scoff at the radical drop reflected by the federal approach.
"Unless 18,000 people died, I have no idea where they put them," said Alice Callaghan, a longtime skid row activist and head of Las Familias del Pueblo, which educates immigrant children.
The dispute comes as the federal government has focused on getting the chronically homeless and veterans off the streets in 2015. Los Angeles' homeless population represents 9% of the national total and includes the most homeless veterans — more than 6,000 — in the country.
Success or failure here could tip the balance in the national fight.
"Our emphasis on data is changing the landscape, showing when we invest in strategies that work, we absolutely can end homelessness," said Laura Zeilinger, executive director of the U.S. Interagency Council on Homelessness, part of the federal government. "We absolutely see L.A. as critical to the national effort."
In 2002, HUD began developing standards to create uniform and reliable data on homelessness.
In most places, computerized bed counts from shelters and transitional and other temporary housing sites are reported to HUD. Volunteers also fan out in the streets, typically on a single night, to count people sleeping in cars, under freeway bridges or in riverbed encampments.
Because of the Los Angeles region's social, political and geographical complexity — mountains, forests, riverbeds, islands and 88 municipalities — the Los Angeles Homeless Services Authority conducts its count only every two years, over several days.
Final numbers are extrapolated from a sample count. Over the years, the sample has grown, with the 2013 street count covering 70% of the county's census tracts, said Mike Arnold, the homeless authority's executive director.
The county and several other jurisdictions, including Las Vegas, also used a telephone survey to ferret out people living in garages or garden shacks, or camping out in friends' yards — the so-called hidden homeless. In 2013, 30,000 households in Los Angeles County were contacted by phone, Arnold said.
Arnold said the survey methodology had been published in a peer-reviewed journal in Europe. HUD had accepted telephone survey results from Los Angeles for every count since 2005, he added.
As for HUD's decision to stop accepting the phone survey data, Arnold said: "There are the public reasons and there are the private ones. Everybody is under pressure to end homelessness."
Federal officials said they long had reservations about the survey. After seeing several years of survey results, they pulled the plug on its use by any jurisdiction.
In a letter to the L.A. homeless authority in October 2013, HUD officials took issue with the idea of extrapolating from what they said was a small telephone survey sample.
"In 2011, a single person counted in Los Angeles telephone survey results in nearly 1,000 persons being added to its total [homeless] count," HUD officials wrote.
Earlier this year, the Department of Veterans Affairs offered the county an estimated $772,000 to fund a 2014 homeless count that might have cleared up some of the confusion. The homeless authority's commission, made up of 10 mayoral and county supervisor appointees, said that negotiations with the VA over the contract had broken down.
Arnold said the VA's "contracting bureaucracy" was to blame. Sixteen other jurisdictions around the country accepted the VA's offer, and a handful of others did not.
Federal officials said time ran out for Los Angeles to plan the complex undertaking. Local homeless advocates said they were disappointed.
"LAHSA needlessly missed the opportunity to update our homeless figures in 2014," said Flora Gil Krisiloff, the Westside deputy for County Supervisor Zev Yaroslavsky and his appointee to the homeless commission. "It would have been especially useful in tracking veteran homelessness."
Arnold said the commission is continuing to work with the VA to get Los Angeles on an annual count schedule. To allow for historical comparisons and avoid the misleading appearance of a sudden drop, HUD in 2014 will publish L.A. numbers with and without the hidden homeless data.
"The numbers get far too much emphasis in policy, as opposed to doing anything about homelessness," said retired UCLA law professor Gary Blasi, who studies the issue. "Even with street counts, people do the best they can, but there's a very wide margin of error."