Inherent in this question seems to be the idea of posting a job somewhere, see who applies, then use a filter of "author of open source project we use."
My take was more that they look at the open source software they like and use, then go out and try to charm, bribe, seduce and otherwise persuade the authors into working for them.
I don't know how many employees GitHub has, but if it is in the < 10 range and they are only looking to add 1 employee at any given time, and are willing to stand pat if they don't find an ideal candidate, a total pool of 100 candidates might be enough.
And yes, your interpretation definitely seems to be more of what they're talking about.
You clearly don't have a reliable idea about how much open source software you use, and how many people are behind this code.
For instance you are probably using Firefox, some torrent client, Apache/Linux/Python/Ruby/PHP/MySQL and all the libs associated with this languages. Even if I stop now there are already a ton of guys involved. Even alone, the Ruby community, and just considering people "well known" inside the community, is a lot bigger than 100 individuals.
Even if you are using Windows or Mac OS X there is a lot of BSD code inside. You DSL modem is built on open source software. Your world is full of open source.
It's not that not participating in the open source community means that you're a bad programmer, it's just that it's easy to judge the quality of work from someone who publishes their work regularly. Easier than evaluating the typical corporate applicant (with "15 years of PERL") that can't show you his code because it's owned by someone else, anyway.
I use about 15 open source tools. In my world, there are perhaps 100 projects that are somewhat important, and could possibly be used. So your pool of candidates reduces to 100?
I use 100s of libraries for developing my applications, and I know most of the authors pretty well. Big projects, like emacs, firefox, etc. are not the entirety of the open source community.
(I think my "Why I Stick with Perl" talk is relevant: http://www.jrock.us/why-i-stick-with-perl.esl although I am not sure my slides without my commentary actually makes sense.)
If only my own blog were so effective ;)
I've found that hobby code I've worked on is much better quality than the stuff done to deadline during working hours. Its a good to be able to say 'see my code at myproject.blah.org' and give prospective employers a chance to see what you're capable of. I'm impressed that you've written a CD burner even if I've never used it.
Your comment on starting a new company versus joining an existing one is well noted. The shift is taking place already, many startups are popping up with a developer or three that hacked their way into something valuable. With hard work and iterative releases driven by customer/user info new products can become the baseline for a new business.
Thousands or potentially millions of micro businesses are the future job.
Right on Max, great stuff.
If I wanted it to be popular, I would have to go through an entire different process - I'd have to pick something in an area that is already popular (like rails), I'd have to maintain a blog, I'd have to have a persona, I'd have to basically market the project.
Only then do I become the dude who wrote x. And my reward is to go work for a tiny company? Would it not make more sense to just use that energy and found my own company? Would I not learn a lot more in the process?
I use about 15 open source tools. In my world, there are perhaps 100 projects that are somewhat important, and could possibly be used. So your pool of candidates reduces to 100?