The team does this optically - rather than mathematically, which at these data rates would be impossible - by splitting the incoming beam into different paths that arrive at different times, recombining them on a detector."
This is cool, and makes sense.
No, the "trick" is decades of hard work and research in materials, electronics, photonics, signal processing and more.
Journalists like to believe you can sum up any scientific result in one sentence understandable by mortals. It turns out it's not the case.
The particular discovery that is the core of this, is to do this transform. That doesn't imply that (a) doing that is simple (b) getting to the point where you can do the trick doesn't build on decades worth of science.
And, just out of curiosity, what's the alternative to simplification? "It's fast. There's no way you'd be able to grasp why. kthxbye."
Also, consider they did add that touch you were missing at the end:
"Think of all the tremendous progress in silicon photonics," he said. "Nobody could have imagined 10 years ago that nowadays it would be so common to integrate relatively complicated optical circuits on to a silicon chip."
And since the scientist they interviewed was German, you should also factor in certain translation issues like a German scientist explaining Fourier transformation and splitting of light beams to a British news reporter.
I'd be interested to hear how close this record gets to the theoretical maximum, and where the intrinsic limit comes from.
The wavelength range currently standardized for telecommunication is 1260nm-1675nm [1], which corresponds to 59 THz as far as I can tell. I don't know which wavelength range Vint Cerf had in mind.
[1] http://en.wikipedia.org/wiki/Fiber-optic_communication#Trans...
There's a diagram available in the abstract that compares OFDM (the technique they use) with WDM.
Apparently OFDM only uses a single light-source, versus (C|D)WDM, which multiplex multiple discrete wavelengths onto the same fiber.
