IMO if the industry isn't careful, we're going to end up needing a load of different interfaces on desktop boards 'just in case'.
Most users don't need 12Gbps storage yet. I bet that if I connected my Samsung 840 PRO to a SATA 3Gbps system, I couldn't tell the difference in performance unless I did something that really did exploit the bandwidth potential. As it is, I've got a new WDB 1TB and that SSD in my system, so if I'm transferring between the two, the maximum throughput I'm likely to get is 180MB/sec. I don't often do large file copies that are purely on the SSD.
IMO, get SATA 12Gbps on the table with the usual backwards compatibility, then spend time coming up with an elegant replacement for SATA that has some decent evolutionary headroom.
According to wikipedia, PATA was around (ie. 'current') for 17 years. If SATA was phased out in 2014, that would be nine years in total. I still have some PATA drives knocking around. Boards had PATA sockets until about 2-4 years ago.
If the successor to SATA isn't properly planned out, then we could end up replacing it in half the time SATA has been around. It would be a royal PITA for everyone involved. AFAIK no version of Windows supports booting off of a PCIe SSD yet, let alone what is dreamed up after SATA or the successor to that.
Furthermore, getting companies to work within constraints is sometimes helpful - before netbooks came along, who would have thought that Windows's memory footprint would actually DROP with future versions? Similar things have happened with regard to power usage and processor design due to practical constraints.