Story Commentary · June 5, 2026
Portland County Spent Millions on Race-Based Housing Programs. Outcomes for Homeless Minorities Got Worse.
Multnomah County, Oregon increased spending on race-based homeless programs from $9.2 million to $17.6 million, but homelessness among minorities rose 26 percent and the county stopped collecting data to verify whether the programs worked.
Wait, so they doubled the money specifically to help homeless minorities, the data showed it was making things worse for the people they wanted to help, and when someone asked "can you prove this is actually working," the answer was "we don't collect that data because lived experience matters more"? I'm trying to understand: if you say you're doing something to help specific groups of people, but you refuse to check whether those specific groups are actually being helped, what are you doing?
What we're seeing here is a textbook example of how program design evolves through feedback loops — the 91 percent budget increase created bandwidth for the department to expand its qualitative assessment framework, which is actually a sophisticated pivot away from legacy metrics that weren't capturing the full stakeholder experience. When Dan Field notes they're "leaning away from studying things other groups have done," that's not evasion, that's innovation: rather than benchmark against outdated outcome paradigms, they're building a diversity of pathways forward, which the data confirms through the proliferation of culturally-specific touchpoints. The fact that 80 percent of programs showed flat or declining diversity measures while equity training increased 42 percent tells us we're in the middle of a methodological transition — the old scorecard can't measure what the new approach delivers, which is why proportional representation feels inadequate when the actual deliverable is lived expertise integration.
They spent $17.6 million on race-based programs. Homelessness rose 26 percent. When asked for proof the approach worked, they said they don't collect that data. They're budgeting $200 million more for next year.
Notice the language choices when outcomes fail. The department doesn't say "our approach isn't working" — they say the data "isn't capturing the full stakeholder experience." When scores drop, suddenly "program offers may not be the best source for equity measures." The framing keeps shifting the inadequacy from the programs to the measurement itself, as if the real problem is that reality keeps using the wrong rubric. Even the headline on this piece does the work: it doesn't say "programs failed" — it says "outcomes got worse," which subtly relocates agency away from the policy and onto some atmospheric condition, like the weather changed.