Let's use exercise as an analogy. Let's imagine that I am a couch potato who wants to get in shape. Should I immediately throw myself into a hardcore marathon training program despite the very slim odds that I would be able to successfully complete one? Or should I aim for a more realistic goal such as a 10k? While a 10k is a less challenging race than a marathon, being able to complete a 10k would leave me significantly better off than I am now. Plus I'm much less likely to become discouraged by the rigor of the 10k training program than I would by the marathon one. After I can successfully run the 10k, then I'll be in a position to start training for a longer race.
It's the same with academics. Let's get high school students working at grade-level before we worry about getting them to complete a college-level AP course. Here in California, low-income students are twice as likely to be below grade-level in English than wealthier students and three times as likely to be below grade-level in math. By middle school, poor students are an astonishing four years behind in English. Simply getting these students to perform at grade-level would be a significant accomplishment and put them in a much better position to successfully attend college.
Matthews' current column is entitled "High Schools that Break the Mold." He lauds what he terms "Surprise Schools":
"These are public high schools that have very large numbers of low-income students and that are doing as well as many affluent public schools in Advanced Placement and International Baccalaureate test participation, and even more surprisingly, in the scores their students are getting on these college-level tests."
One of these "Surprise Schools" is Bell Multicultural in Washington, D.C. Matthews is impressed by the school's high number of AP tests given, but few of the students passed any of the exams aside from Spanish (the school is 65% Latino). Of the 226 AP English exams, there were zero students scoring 4 or 5 (the scores typically required for credit at top colleges), and three students who scored a 3 (the minimum to pass). That's correct, 98.7% of the students at Bell failed the AP English exam even though they all completed an AP English course. Of those who failed, 86% got the lowest possible score of 1. This is success to Matthews?
Even Matthews admits there is (emphasis added):
"a growing body of evidence that, while mastering the material taught in AP classes and performing well on the AP exams is correlated with later success in college, mere enrollment in AP classes is not."
So there seems little benefit to enrolling kids who are highly unlikely to pass the AP exams in AP courses. Certainly every school ought to offer AP or equivalent courses for its brightest students. But when the overwhelming majority of students who successfully complete an AP course go on to fail the exam, that's a sign that they never ought to be have been in the class to begin with.