Ron Maimon is entirely correct when he says that GA is precisely Clifford algebra from a mathematical perspective, as any book or paper using the phrase "Geometric Algebra" is sure to say. But I think he misses both the point of the question and the point of "GA" — which is different from Clifford algebra from a pedagogical perspective.
The question
I'll paraphrase the question as:
Is GA a good, pedagogical way to introduce the mathematical side of physics to undergraduates?
My answer is: in the future, definitely yes because GA is obviously vastly superior for pedagogy; for now, probably not unless a large fraction of the physics department is willing to get behind the effort.
Eventually, for any student who will need to learn basic div, grad, curl — and maybe even just cross products — I think GA will be the standard approach. My first reason for saying that it's not necessarily a great idea right now is that no good textbook currently exists for this. I'm sure Quirino has inspired many talented undergrads, but I would also worry about losing quite a few. There are many different learning styles, and most of them benefit greatly from having a good, thoroughly refined text — or even a text that has grown out of previous generations of texts. All the books I've seen suffer from their constant attempts to explain why GA is better than standard treatments, rather than just being good, self-contained introductory texts with a focus on actual physics applications.
My second reason for suggesting that it's not a great idea right now is this: unless students can use GA in all (or at least most) of their courses, they'll have to constantly code-switch — going back and forth between GA and traditional treatments. In principle, that could actually be beneficial in getting some students to understand the concepts more deeply, but in practice it's far more likely to just lead to greater confusion. Only after you're already proficient with both GA and traditional methods is it easy or beneficial to code-switch. But this isn't likely to happen right now, because — though it is rapidly gaining in popularity — there still aren't very many professors who know GA well enough that they would teach a class using it. And even though there are instances where that could work for one class, as soon as the student moves on to a non-GA class, much of that learning is useless. Since physics is such a cumulative process — where techniques from an earlier class form the basis for future classes — this is a huge problem.
Now, it's entirely conceivable that you could get enough professors in one department who are sufficiently motivated to actually try a four-year GA curriculum (and sufficiently experienced with GA to be able to teach it well). Together, they could overcome the lack of existing textbooks by getting to the students early on — for example using a standard textbook for first-year mechanics, but insisting that all cross products be done as wedge products (which is code-switching, but easy enough that it would work). The intro E&M course is where I think this would get hardest for the teachers, and would require particularly concerted and sustained cooperation to supply the students with enough materials — extensive notes plus homework, quiz, and exam questions — to replace standard materials. And all that work can't just go by the wayside as students move on to more advanced courses. I think an advanced mechanics course would actually be easier to teach with GA than with the old methods used currently. And by the time students get to advanced E&M, they may be able to handle a standard textbook again and be able to fill in the gaps with their own superior GA techniques (with some help from the teachers). The first round of students might have a harder time, but if the teachers are conscientious about recording and analyzing how to handle the challenges, it would quickly become a successful curriculum. In the five years since I first wrote this answer, GA has become much more mainstream so that this possibility has gone from being absurd to conceivable — but I'm not holding my breath.
"Geometric" algebra
Because there seems to be a lot of confusion on this page about what GA is and how it differs from Clifford algebra, I'll back up and describe the distinctions and why using GA might be a good objective.
Geometric algebra is just a name for a pedagogical approach to teaching physics that introduces Clifford algebra (usually over the reals) in a simple way, emphasizing the geometric nature of the elements and operations, and using Clifford algebra as the fundamental tool for basically all calculations — though without mentioning many of the more formal constructions of Clifford algebra as it is usually presented, or resorting to the stupid complex-matrix approach that physicists usually see. Just as Penrose (usually) makes no mention of left minimal ideals or fiber bundles when he discusses spinors, a simpler approach to Clifford algebra is reasonable — and would be vastly preferable for first-year physics students.
Now, the contention is that Clifford algebra is under-utilized in basic physics. Every problem in rigid-body dynamics is at least as easy when using Clifford algebra as anything else — and most are far easier — which is why you see quaternions being used so frequently. Orbital dynamics (especially eccentricity!) is practically trivial. Relativistic dynamics is simple. Moreover, once you've gotten practice with Clifford algebra in the basics, extending to electrodynamics and the Dirac equation are really simple steps. So I think there's a strong case to be made that this would be a useful approach for undergrads. This could all be done using different tools, of course — that's how most of us learned them. But maybe we could do it better, and more consistently.
No one is claiming that Clifford algebra is fundamentally new; just that it could be bundled into a neater package, making for easier learning. Try teaching a kid who is struggling with the direction of the curl vector that s/he should really be thinking in terms of the algebra generated by the (recently introduced) vector space, subject only to the condition that the product of a vector with itself is equal to the quadratic form. Or a kid who can't understand Euler angles that this rotation is better understood as a transformation generated (under a two-fold covering) by the even subalgebra of $\mathcal{Cl}_{3,0}(\mathbb{R})$. No one here is arguing that that should happen. GA is just a name for a pedagogical approach that makes these lessons a whole lot easier than they would be if you sent the student off to read Bourbaki. Starting off with GA may be slightly harder at the beginning, but pays enormous dividends once you get to harder problems. And once teachers and textbooks get good at explaining GA, even the introduction will be easier.
I used to use ordinary tensor algebra for relativistic dynamics, differential forms and/or tensors for electrodynamics, and spinors for analyzing gravitational radiation. These days, I mostly just use GA for all of that. I like to make use of the manipulations that are so much simpler with Clifford's geometric product, the conciseness of objects that encode multiple grades at the same time, and the consistency of using the same exact formalism in all these domains. Why shouldn't my undergrads get the same advantages?
Should Ron and Luboš abandon their mathematical upbringing and start speaking Hestenes's language? Probably not. Might it be useful to other, younger students of physics? Definitely.
As for why you would use the name Geometric algebra (GA) instead of Clifford algebra (CA): typically, GA is restricted to vector spaces over the reals, so it's just a shorthand. Also, apparently Clifford himself called it Geometric Algebra — so it's not a sin.
I'll also point out that tensors do fit nicely within CA/GA, in the form of linear functions, for which CA/GA provides new interpretations, without making the usual manipulations harder. Also, those nice features that Penrose mentions when he talks about spinors and twistors are present naturally in GA (I'm specifically referring to CA over the reals). He talks about geometric interpretations and calculational efficiencies, which GA shares. But GA goes further. Penrose et al. (Penrose-Newman, Geroch-Held-Penrose, etc.) use complex numbers to combine contractions that really deserve to be separate-grade objects, with accompanying geometrical interpretations. The optical scalars are a prime example, where the standard treatment drops geometrically meaningful elements so that the familiar product of complex numbers can be used. GA can replace this with Clifford's geometric product while retaining the geometry, and using the exact same unified formalism used for many other types of problems.