[e2e] Open the floodgate

[Craig Partridge]

In message <40883051.EBB34585 at cup.hp.com>, Rick Jones writes:

>...in the context of TCP VJ congestion control, wasn't one of the stated
>assumptions that networks didn't have much in the way of error rates in them
>other than drops resulting from congestion?

My recollection (and I hope Van chimes in) was that he argued that we
had no information as to whether it was congestion or error, but (a)
errors were relatively rare [true at the time]; and (b) if we ignored
real congestion, we were in deep trouble.  Most (all?) of the research
work I've seen since is in general accord with those statements, even if
errors are more common.  Restating this a bit more crisply:

    Mistaking an error-induced loss for congestion loss hurts, but
    mistaking congestion loss for error-induced loss hurts much much
    worse.  (Even if we're talking congestion loss on a link with lots
    of errors).

Craig