San Francisco's Superior Court acknowledged its own data limitations in writing — then published the numbers anyway to justify dismantling a 50-year-old pretrial services nonprofit, Sheriff Paul Miyamoto charged in a letter to city officials Thursday.

The methodological dispute is no longer academic: the city budget now before the Board of Supervisors would provide $12.7 million to the Probation Department to absorb the SF Pretrial Diversion Project's caseload — $4.7 million more per year than the nonprofit's current $8 million allocation — justified in part by court data the sheriff says fails to meet national standards and which court officials themselves flagged as incomplete before releasing it.

The fight between San Francisco Superior Court and the SF Pretrial Diversion Project, the nonprofit that has supervised defendants awaiting trial for five decades, has been simmering since early last year. Now it has produced dueling letters — and a starkly different set of numbers, depending on who's counting.

In a letter obtained by the SF Standard, Court CEO Brandon Riley cited court-compiled data showing that only 80 percent of pretrial diversion clients appeared for their hearings last year, and just 69.5 percent avoided rearrest or new charges while awaiting trial. Those figures were presented as evidence that the nonprofit's own self-reported performance — 95 percent court appearance, 97 percent rearrest-free — was misleading.

Sheriff Miyamoto, whose department holds the contract with the nonprofit, fired back Thursday with a letter of his own to city officials challenging the court's methodology.

"Transparency requires more than the publication of data," Miyamoto wrote, according to the Standard. "It requires a shared understanding of what is being measured, how it is being measured, and whether the metrics being compared are, in fact, measuring the same outcomes."

The sheriff said the court's analysis failed to use national standards and could "lead to different and sometimes inaccurate conclusions regarding program performance."

At the center of the methodological dispute is a straightforward but consequential difference: the court counts any arrest against SF Pretrial's record, even arrests that never resulted in charges. The nonprofit and the sheriff use the nationally recognized standard of counting only arrests that led to formal charges — because, advocates note, an arrest alone doesn't establish that a crime was committed.

There is a second, arguably larger flaw. The court's case management system, according to SF Pretrial's chief impact officer Matt Miller, "doesn't track when clients leave our supervision," meaning missed hearings and arrests continue to accumulate against SF Pretrial's account long after its supervisory relationship with a defendant has ended.

"The court acknowledges this limitation in its own caveats and publishes the numbers anyway," Miller said in a statement.

Riley, in his letter, dismissed the nonprofit's 97 percent figure as "misleading" and pointed to a state pilot program study that used the broader arrest-based definition of recidivism. That same state study found 61 to 73 percent of defendants were not rearrested — figures that align with the court's 69.5 percent calculation for SF Pretrial clients.

SF Pretrial CEO David Marouff said the court never gave the nonprofit a chance to address concerns before moving to replace it.

"We were … never given an opportunity to address the court's concerns, even after multiple requests," Marouff told the Standard. "We tried to honor 50 years of partnership. In contrast, the Superior Court is publicly impugning our agency to validate their decision that is being robustly challenged."

The feud broke out in earnest after SF Pretrial came under scrutiny for mismanaging its retirement funds. The city placed the nonprofit on a watch list of troubled nonprofits, and the court demanded access to its financial records. SF Pretrial declined, citing no legal obligation to comply. The court responded by announcing it would transfer the program's functions to the Probation Department.

The Board of Supervisors has not yet voted on the sheriff's contract with SF Pretrial. But the Probation Department's request for $12.7 million in city funds to build a new pretrial services unit is included in Mayor Daniel Lurie's budget proposal — representing a $4.7 million, or nearly 60 percent, increase over what the nonprofit currently costs the city.

In its own Thursday letter, the Sheriff's Office stopped short of fully endorsing the nonprofit but called for continued cooperation rather than a rupture, warning against ending a decades-long program. The sheriff's department recently awarded SF Pretrial a new contract.

For decades, San Francisco and Santa Clara counties were the only California counties providing pretrial services. In 2018, the state Legislature authorized county probation departments to run pretrial operations statewide — but created a specific carve-out for SF Pretrial because of its long-established role in the city's justice system. Whether that carve-out survives the current budget cycle may depend on whether the Board of Supervisors accepts the court's framing of the data dispute, or the sheriff's.