Over on
this thread, there's an interesting read. The
Thrown for a Loop story has some analysis, including words from Mike Gavin, that gives the aims a little more flavour.
I too read the comments elsewhere indicating that they'll be looking at deploying G.fast from existing cabinets - but I'd think that was an important aspect to try out in the trial, while you are only targeting a couple of hundred Mbps. I suspect (heavily) that the first generations of G.fast silicon won't be able to hit the top speeds, so any rollout will be for interim hardware. I wonder if Supervectoring might get something of a call too.
There's also a little said about vectoring in
this document; unfortunately, it still doesn't give us a date, but it does put the words "vectoring" and "rollout" next to each other.
One snippet: With 80Mbps caps still in place, vectoring raised the average speed from 54Mbps to 64Mbps. On first glance, that doesn't look too great, but I think the fact the speeds were capped to 80Mbps limits the average upside.