PE> What makes you think that its not the other way around? The
PE> stats are accurate when there's a Netcomm online, and stuffed
PE> when there's another USR online?  Or indeed, stuffed for both?

RS> Because we have an independent measure of your line, the calls by
RS> Russ to Bill and you with a USR on BOTH ends, which shows Bills is
RS> significantly worse than yours. Bill and you calling the bulldog
RS> too for that matter. The USR/USR stats do correlate well with total
RS> thruput, particularly when you include the Russ/Bulldog calls.

PE> The independent measure being another presumed-buggy USR?

Which correlates well with actual thruput when the
measurement is done with a Courier at BOTH ends of the line.

PE> I'll let you guys fight that one out amongst yourselves, because I didn't
PE> keep track of dB levels myself because I didn't know what they meant.

Dunno if you have seen Frankowitz reply to Bill in USR_Modems
on that, its basically waffle which doesnt add a damned thing.

PE> And with the fucked results I've been seeing with them,
PE> I'm glad I didn't pollute my mind with them at all!

One day you might eventually notice that its quite possible to use less
than ideal data to STILL say something useful about whats being observed.
Which can come in rather handy when you dont normally have the level of
analysis available that one Paul Edwards proclaims is all thats useful.

PE> I reckon that's a tool binned for a long time.

Yes, but you always do bin useful data.
@EOT:

---