On your first point: the fact that there are false positives in the admission process doesn't imply that there are no false negatives. The dropout rate would pr

On your first point: the fact that there are false positives in the admission process doesn't imply that there are no false negatives. The dropout rate would presumably increase as they admitted more students, but I don't think it's plausible that the dropout rate for the marginal admitted student is 100%. On your second point: a Berkeley diploma signals that a graduate (1) is above some minimum (but still very high) threshold of competence, and (2) has a level of competence governed by some probability distribution that's above that minimum threshold. I assume that as Berkeley admitted more students, the minimum standard would stay the same but the average of that competence distribution would decrease toward that minimum. This is subjective and I'd understand if you disagree but that seems like a worthwhile tradeoff to me. I think there are pretty few jobs for which a bottom-of-their-class Berkeley CS grad is underqualified, and obviously those jobs have their own evaluation/screening processes on top of looking at the Berkeley diploma anyway. I suppose an employer looking for the "superstars" in the top 1% (by some immeasurable criteria) of their class at Berkeley would have to identify the top 0.1% instead (if we generously assume that Berkeley's current admissions process never turns down those students), but I'm less concerned about those dozens of students than about the tens of thousands more who could benefit from both a more rigorous curriculum and the ability to demonstrate to employers that they're above that minimum threshold.