There has  much been said  but why the  heck not adding  my own opinion  to it.
I'll try to keep it short ...
There are two views to LISTSERV (and almost every other software):
a)  easier-to-use functions  -  we could  live without  LISTSERV,MAILER,NETSERV
 and all  that stuff.  you are mainly  interested in  "function", "performance"
 only a second concern.
b) saving  of resources -  this could be DISK-space,  CPU-time, line/link-load,
 and "human"-resources. But the last one is closely connected to (a).
 This point could be divided in ...
 ba) LOCAL resources
 bb) NETWORK resources
*-----*
Now  seeing it  from these  points  of view  I  want to  raise some  questions.
1) who  complaining about  the additional  "consumption" by  the LISTSERV  made
 an *overall* evaluation  of the effects ?  I.e. only one file coming  in for a
 list  of users,  savings on  the  "end-user" side  and  so on  should also  be
 taken into account.
2) what is  the real additional  load induce by the  LISTSERVer ? What  I heard
 is that  LISTSERV is  most active  when also the  users are  - but  that holds
 true for  RSCS and  other components  too. This is  general problem  but could
 be alleviated  somehow by quiescing  certain components trough  the peek-hours
 - but you have to pay in SPOOL-space for it ...
 Eric: maybe  we should think  over the  possible effects of  "working-hours" -
 a new  List-keyword like: Online= ....  with values specifying the  hours when
 the List  is to be active  (and also for  Distribute) ? Sure this  would cause
 a delay  in distribution  (and lead  to answers  arriving before  questions) -
 but if  some site feels that  its better to  quiesce during the day  they will
 do it manually anyway. With the additional Keyword you would know in
 advance which hours a certain list is served.
3) for  off-way nodes  running  a LISTSERV  the traffic  will  increase if  the
 central nodes is not running LISTSERV (correct me if I'm wrong).
 On the  other hand  central/HUB nodes and  traffic on the  main lines  will be
 reduced.  So why  not taking  some of  the work  ? With  LISTSERV we  have the
 chance  to split  the additional  CPU... between  a bigger  number of  off-way
 nodes. It  could be a  big relief  for the central  nodes and main  lines with
 only a slight increase of load in *many* others. "Share and enjoy".
4) Why bother what is  distributed to whom ? When I  see several BITNAUTS LISTs
 (only an example  - no offend :-))  crossing the Atlantic one  after the other
 and loading  the links  for several hours  I know what  DISTRIBUTE is  worth (
 or COULD  be worth). Its  a question of  "who is to  blame" - noone  would say
 "the heck  why do  you permit your  RSCS to  send such big  pieces of  junk" -
 but with LISTSERV there is always the POSTMASTER who installed it and
 maintains it.  But if there  weren't LISTSERV the  files would make  their way
 anyway ...
 I think  its more the  feeling that something is  here not totally  under your
 control - but this is true for many other components too.
*------*
The heck, I said short ...
*Eric:*
two more things to the DISTRIBUTE:
1) I hadn't  the time to  go through all  the MEMOs - did  I hear Priority  ? I
 wouldn't  like to  see it  if a  80000  bytes file  is sent  with prior=0  ...
2) what  about automatically  splitting big  JOBs (I  know of  the problems  of
 packet-sending .... it almost impossible but maybe worth thinking).
*--*
So much for  now, I think I've  repeated everything already said  - sorry ..:-)
<CR