I doubt the author, a Stanford professor, is correct, but can someone knowledgeable about San Francisco public schools comment about the impact of detracking?

Separating ‘gifted’ children hasn’t led to better achievement
by Jo Boaler
Hechinger Report
November 4, 2019

When New York City’s mayor began a move to revamp the program of selective schools last year, a public outcry ensued, and the issue has yet to be resolved.

Objections echoed those in the San Francisco Unified School District, which six years ago began in earnest the elimination of advanced mathematics classes until after 10th grade. Parents created Facebook groups to oppose the changes.

Many believe that children learn more effectively in schools or classes with similar learners, but are they right? It is a question that has long intrigued and divided people. When learners show different achievement levels, should we teach them separately or together? I have spent my career studying this question and, although the logic of separate classes seems strong, evidence leads us in a different direction.

Related: Ending racial inequality in gifted education

For instance, after San Francisco Unified de-tracked math, the proportion of students failing algebra fell from 40 percent to 8 percent and the proportion of students taking advanced classes rose to a third, the highest percentage in district history. Until 10th grade, students take the same mathematics classes. From 11th grade on, students can choose different pathways.

Eight Bay Area school districts found similar results when they de-tracked middle-school mathematics and provided professional development to teachers. In 2014, 63 percent of students were in advanced classes, whereas in 2015 only 12 percent were in advanced classes and everyone else was taking Math 8. The overall achievement of the students significantly increased after de-tracking. The cohort of students in eighth-grade mathematics in 2015 were 15 months ahead of the previous cohort of students who were mainly in advanced classes.

...