Frustration time was between 19 and 27 seconds for the 6 scanners, averaging out to 22.3 seconds. That's a pretty high number and that's why the StreamQ is B and C+. (Frustration time must be under 5 seconds for an A+).

For on-demand content, the StreamQ scale isn't demanding at all - A+'s are quite common. But live events are different story. We have yet to see what a reasonable scale is for mega-sized live events.

NYC-1 has a dedicated T1 from AT&T. In this group, it is the scanner with the most bandwidth. Yet, it had the worst connection time of all the scanners. Goes to show you that bandwidth doesn't always win the day. Could be a peering problem with AT&T's Manhattan POP.

The rebuffer times have the most variability across the different scanners. NYC-1 had only 0.8 seconds of rebuffering on average. (And you can see from the "Rebuffer Event" row that rebuffering only occured in about 1 out of every 10 checks.) Compare this with Bay-2 (an @Home cable modem). It had a rebuffer event on every other check (that's what the 0.5 means).

The interpretation here is pretty clear. Rebuffering is a result of inconsistent connection speed. For NYC-1, the time to initially reach the server was longer, but once the connection was established, it was a smooth signal (which is what you'd expect from a dedicated T1).