Try to avoid getting in this jam; make that an explicit goal of your recruiting (and, more importantly, business planning) process. You may find that Fried's "chuck the resumes" advice starts making a whole hell of a lot more sense.
I've helped my friend do recruiting at his startup (albeit, not in the US). There are hundreds and hundreds of applications, and the constraint I pointed out becomes evident. Is it really cost-effective to personally interview and understand the context of every single resume-submitter, including those with absolutely no experience? How about someone without even a compsci degree and no experience to make up for it?
To be in a position to blow off resumes, you need to run a company like 37signals does. You need to chill out, do excellent work, collect fans, and be open to interesting people who go out of their way to bug you for jobs.
The very best people in the industry don't give a shit about your job reqs. The minute they actually get on the market, they're snatched up, because their friends track their every motion and are alert to the slightest sign of unhappiness in their current job. You think Mark Dowd, Zed Shaw, or Tom Preston-Werner are going to fill out job reqs and send out resumes? That's naive.
By running extremely lean, rocking out the business, and building stuff people like in an environment that builders like, 37signals can drag in A+ players when they become available, because they aren't clogging up the decks with people they fished out of a pile of resumes. Your friend, maybe not so much?
That’s where the cover letter comes in, right? If B’s cover letter says “I see from your Web site that your business is involved with X, Y, and Z; here are some examples of projects that I recently worked on that are directly relevant to those needs”, and A’s cover letter is all buzzword buzzword buzzword, I would take B a lot more seriously as a candidate.
But if you have a very small amount of information and time to make a decision, where would you place a bet? It's certainly very unfair for the many talented graduates from lesser known universities. However, I find the idea of recruiting based partially on these bullet points valid from a expected proficiency optimizing perspective.
The talented guy doing IT at a law firm has some options to signal his talent. For example, he can create or become a top contributor in an open source project. I am sure many talented people who feel unfulfilled in their jobs do this.