Of course. This is the same as saying that a CS student can better understand the reasoning of a program than the reasoning of a human being. As a rule programs are, in general, more predictable.
I will disagree with the idea that it is impossible to achieve high accuracy in predicting grades on paper. It's anecdotal evidence only, but I've never been off by more than half a letter grade on any paper in the last three years of taking English classes (it's my minor). That's 6 classes with 5 different professors. So probably 20 to 30 papers? Usually I get the exact grade right.
Might I be too bold in suggesting that CS students prefer the determinism of a grading script because they are worried how their code will hold up in the real world? Given the quality of code most CS students produce, that's actually not that bold of an assumption. Seems to me that CS students that do not favor how grading is done in English have that opinion because they aren't used to thorough assessment in CS [1].
[1] The notable exception being theory classes. Go theory! But in those classes you've got a living, breathing grader as well. :-p
EDIT: I realize I didn't really address your point. Yes, that is a fair characterization of the quote in the Times. The problem lies in assuming the grading script is doing as good of a job as your English professor in determining correctness. I bet if you increased the accuracy of the grading method for programs you would sacrifice on precision and students wouldn't be as confident about their programs. If you graded on coding style and design decisions, for example. So yes, CS students can be more confident. But confident of what? The fact that your program conforms to a script? That's not the same as being confidant that your program actually works.
/happy he's got a degree in philosophy and didn't do CS at Stanford.