Phil's comments about changes to RSCS versus accepting any and all
DISTRIBUTE traffic are a little misleading.  There are only 41 LISTSERVs
among 1200+ nodes, and in the states, as it's been observed, many of the
high traffic nodes don't have LISTSERVs.  Now, if two copies of a file
are being distributed to our parent node, some other LISTSERV is going
to realize it's "better" to send one copy to our LISTSERV and let us
send two copies.  Therefore, we inherently accept one reception and
two sends that normally we would not see.  Granted, we save any number
of intermediate nodes a send/receive, but there is a greater cost to
this particular node.  If there was a higher concentration of LISTSERVs
it might be easier to show that this wouldn't happen, but with the
sparseness of LISTSERVs in the States, it might very well happen often.