This news is interesting
You do need a hype filter when reading this, and many claims are extropolation-gone-wrong, but the base idea already exists in ring osc designs inside FPGA now.
Seems ( with the right tools ) you could extend this inside a FPGA, by creating a large physical ring (long routes), with the sprinkled buffers. The physical delays would reduce the process variations in the clock, and you get the phase taps 'for free'. - but the tools _will_ need to co-operate :)
We have done this inside CPLDs, and get appx 1.3ns granularity.
With FPGAs the buffer delays are much lower, and the routing can be made to dominate.
Sounds like a project for Antti :)
-jg