And it does show that rather well.
This image shows what the impact can be from just a single well-chosen "disturber" line on the "victim" line (or badly-chosen, as the case may be). That shows how much impact can occur from just one "neighbour" taking the service.
My rule of thumb was that the Original FTTC spec would give us one-third of the ultimate speed, the 17a profile would give us the next third, and that vectoring would give us a final third. "us" being people within around 600 metres.
The latest trials suggests that Vectoring is doing even better than that, and works for people out beyond 1km. I don't know whether it will be enough for BT to increase the headline rate for FTTC though.
I also note a mention of a "G.inp" spec which replaces the FEC mechanism with a retransmission-on-error mechanism instead. That loses the fixed overhead of FEC parity bytes, and re-transmits blocks only in the event of an error... and hopefully there are fewer of those with the reduced crosstalk.