So... what?
Carrying multiple serial protocols (USB, Firewire, RS232, Ethernet- even more parallel stuff like IEEE 1284 and video if you've got the relevant encoder chip) is a piece of piss- you take the signal in from the device and get an ASIC to slap some header info on the data and then fire it out to a laser transceiver. At the other end you reverse the process. With, say, an established premium market to provide "early adopters" and your own silicon design/fab people- funnily enough, that would be exactly who did the presentation- then you'd be able to very easily fit that into a single chip at a price point people would be willing to pay. So it's not exactly "paradigm-shifting" technology...
In case that^^ sounds like bull, in my last job I was responsible for the design of a system with 16 unidirectional video channels, 2x bidirectional Gigabit ethernet links and a couple of megabits worth of serial/DIO/AIO comms- all over a single fiber. So I'm not exactly new to this...
I don't see what advantage this will have for the majority of users, except it's not utterly retarded like the USB3 "Yeah, so we'll take the old convenient, compact(ish) connector and... we'll put a bloody great blob on the top." approach but provides a higher bandwidth and should have a connector that looks pretty sleek. Looks like they're using a 4-channel CWDM to get the 4 fibers, so that'll be one primary and one redundant fiber in each direction- so these things should be sufficiently robust for home use.
Definately not a fail, though- it's a useful, important evolution. But it's not, as presented, the revolution many will undoubtedly make it out to be.
ROTM as more and more optical systems means less and less effective EMP weaponry when the machines rise up...