A bit black and white there. I generally agree that resumes should not be a top indicator, but they can be useful to screen out some people.
For example, I once posted a job that included the word "Ruby". I thought the job posting made it clear that Ruby was a very, very small part of the job. But I still got several resumes from people who were Ruby experts or at least claimed to be, and that is all the experience they had given their resumes. They had little or no non-Ruby experience. At the time, I needed someone who could write Ruby code but had much more experience elsewhere.
So the resumes helped me screen out those candidates. And I also learned that my job posting was worded incorrectly.
And you talked to none of them? You just assumed that their resume told you everything you needed to know about their capabilities? And if it didn't, too bad for them for writing a shitty resume?
In this particular case, I received ~20 resumes for 1 job and 80% were not Ruby-only resumes. So yes the Ruby-only screened them out. However I would have gone back and contacted them if none of the other 80% of candidates worked out.
You will get better results if you build a process that doesn't screen people out because of their resumes. That doesn't mean you don't screen candidates, and it doesn't mean that you don't have a funnel that knocks out a big chunk of them before you get on the phone.
It just means: don't select people for their getting-a-new-job skills. Those skills aren't relevant to software development work.
You can read about tptacek's company's process here: http://matasano.com/careers/. It really does seem like a good way to learn whether a candidate will be able to perform the job.
I wonder that a lot too. I have friends who have tried it for sales and are liking it, but it's too early to tell. (It's not my process, by the way; here's a link 'tokenadult found about it: http://mavweb.mnsu.edu/howard/Schmidt%20and%20Hunter%201998%...)
For example, I once posted a job that included the word "Ruby". I thought the job posting made it clear that Ruby was a very, very small part of the job. But I still got several resumes from people who were Ruby experts or at least claimed to be, and that is all the experience they had given their resumes. They had little or no non-Ruby experience. At the time, I needed someone who could write Ruby code but had much more experience elsewhere.
So the resumes helped me screen out those candidates. And I also learned that my job posting was worded incorrectly.