Last week I had a discussion that’s still echoing in my head.

It was with some people who work at four-year colleges in the area. We were discussing various measures we had taken to improve student success and retention rates: different approaches to academic advising, tweaks to new student orientation, early warning systems, that sort of thing. At which point one of them, from a tuition-driven college, said:

“And of course, you have to identify upfront the students who no amount of help will save. Target the resources where they’ll actually make a difference.”

Oooof.

In a strictly numerical sense, she’s right. If “bang for the buck” is the relevant measure, then yes, some students will not repay the investment.

But the logic leads to some very ugly places.

Anyone with passing familiarity with the sociology of education can rattle off the demographics of the students most at risk: young men of color from shaky school districts. Their rates of academic success are dispiriting at best.

But how you choose to respond to that reveals a lot.

Yes, resources are limited, and yes, college is supposed to be hard. But there’s also a basic moral obligation to make opportunity real, and to try to avoid simply piling on. Profiling the least likely to succeed, and then writing them off, winds up looking a hell of a lot like racism.

Rather than looking at which students are likeliest to succeed, I’d much rather look at which interventions are likeliest to succeed. National data suggests that Upward Bound, for instance, accomplishes very little. But streamlining developmental course sequences accomplishes a lot. (The CCRC studies on this over the last year or so have been revelatory.) To the extent that vulnerable populations are overrepresented in the developmental sequences, they’ll benefit disproportionately from any improvements in it.

The work that needs to be done, and that really has only started, is deliberate experimentation to allow for evidence-based decisions about what works. My image of a really great college is one that incorporates experimentation -- documented, deliberate, scalable experimentation -- into the course of doing business. A college that teaches itself about its students over time -- and that faces the reality of unexpected results when they happen -- is the kind of college that will make real headway in both quality and fairness. Evidence has a way of defeating stereotypes.

The whole idea of writing off entire sections of the student population misses the point of realism. Yes, some students face more challenges than others. But the point of public higher education is not to dignify class divides with a patina of merit. It’s to give everyone a real shot. The place for realism is in evaluating the success of various measures to help those who need it. If one measure doesn’t work, try another. But don’t give up trying. The right place for idealism is in affirming the mission; the right place for realism is in evaluating the methods. When the methods displace the mission, something is fundamentally wrong.

So no, thanks. I won’t triage students when they walk in. The world does too much of that already. If that means I wind up wasting some second chances on some folks who were doomed to fail anyway, well, there are worse crimes. I’ll take that one.