What I'm saying is that it doesn't seem reliable or scalable. Looking at my own network, the one's I've kept up with[1], which I think is pretty average, there are only a handful of people who actually work at a company I'd one day be interested in joining. I know I would easily exhaust the whole pool in a single job search (ask me how I know this). I'm not going to rely on that. To sum up the reliability aspect: I can't foresee a point in my career when I'll ever be able to say "I can rely entirely on my network for my next job search." If you can, congratulations, that's an enviable position.
To me, networking is like that great tool in your toolbox that _really works_ when it works. I think the network approach is most appropriate as a "fallback" when you are desperate for a job, would take anything, and can literally tap everyone you know in hopes of finding _something_.
1: That's another not-so-scalable aspect of networking: Maintenance. I would never out of the blue E-mail a former colleague with "Hey, I know I haven't talked to you even once in 20 years, but, HOW'S IT GOING?!? :) I noticed you work at a company I'm interested in [...]" Awkward! You have to actually keep up with everyone like you would a friend. I've seen people who can do it well, but jeez it must be positively exhausting staying in touch like that.
I'm with you, but I think I see where others are coming from. I had job referrals before, and it can work. The dynamics of this are complicated, but the bottom line seems to be that networking works when it works, but not always or for everyone or in every job market. The reality is that lots of jobs get filled every day with a process very much like the one caricaturised in the piece.
Job-Worker matching as it's done today is haphazard, weird, biased... and almost certainly extremely inefficient.
But... I don't know that we have better examples matchmaking processes at scale. Do we? University selection? Dating? What's the scalable alternative?