If your definition of "Linux on the desktop" is "I see it everywhere I go" then your metric is well behind the early adopter curve.
If, however, your metric is to look at influencers and early adopters, it's clearly happening.
I think back to 04 when I switched to a Mac (from Windows). The idea that Macs were "happening" was inconceivable to many. Forget finding mainstream software. Forget drivers for hw support. Forget file exchange. Forget networking with anyone doing real work at your company. It was all uphill and hard. That was just 2004.
Back then it was generally either hold outs from the dark, pre OS X days or early adopters switching over (or back) to a Mac. If you had your eye on the market, I think in hindsight it's clear that there was a groundswell of interest in the platform (from passionate early adopter switchers, etc). But at the time, nothing was obvious and early adopter advocacy to just "buy a mac" fell on usually deaf ears.
We're in a not dissimilar situation today with Linux. The early adopter segment is different and much more dev oriented, but that's not a bad thing. Add to this that there are some fundamental and very important technology shifts happening that I think are going to make a positive impact on the desktop linux experience over the next several years (wayland, systemd, container/virtualization improvements, packaging and distribution rethinks).
It's happening because people like me are using Linux on the desktop every day for real work, and for many of us it's eminently preferable to any other OS out there. It's happening because we, the early adopters, are not "on the fence" about Linux: I no longer think "Linux is nice but OS X just works". It's happening because a shift to the web has made app lock in on a specific platform less of an issue. It's happening because vendors like Dell are throwing actual resources at the market. Small though it may be, this market is made up of influencers that are the thin edge of the adoption wedge.