Considering USB type C has officially adopted the "We accept and endorse all topologies, no matter how crazy!" philosophy (also known as the "Holy fscking %#&@,
I am seriously confused about this thing! How is Joe Consumer going to react to this???" philosophy or the "Didn't you miss the days when you had a crapton of very similar connectors that said absolutely nothing about what they did? Let's do that again!" school of engineering), I'm guessing someone will make a cable that has no chance of working, because fsck this, it's just as likely to work on a random USB type C port as a DisplayPort adapter, a 100W consumer or a thunderbolt 3 device so the consumer will go "oh well, I guess this won't work".
I mean, tunneling PCI-e over the USB type C physical layer and implementing a PCI-e USB 3.1 controller on the device side, while also carrying optional DisplayPort data which is incompatible with the
other DisplayPort over USB type C standard? Seriously, I get a headache bigger than most Samba-induced headaches just by thinking about the figurative headaches this is going to cause.
Someone at Intel must've been smoking some nasty stuff...
... idly wonders how
@depasseg is hooking up USB drives to multiple machines using a clustering filesystem ...
ATA over Ethernet solves the physical layer problem. I imagine the interposer's software could be hacked enough to accept ATA commands from multiple clients...