From reid@glacier.ARPA (Brian Reid) Sat Mar 8 14:46:58 1986
Path: seismo!ut-sally!pyramid!decwrl!glacier!reid
From: reid@glacier.ARPA (Brian Reid)
Newsgroups: net.news.adm,net.news.group
Subject: who reads USENET, anyhow? Worldwide survey.
Message-ID: <5108@glacier.ARPA>
Date: 8 Mar 86 19:46:58 GMT
Organization: Stanford University, Computer Systems Lab
Lines: 169
Keywords: please run this program on your site
I'm tired of speculation about how many people read USENET, and how big the
audiences are. Here is a program that does a pretty good job of determining
that information. Please install this program on your site, and run it with
the -m option that will mail the results back to glacier. I will tally and
total the results and post them. We've been running this program at about
20 sites in the San Francisco area for 6 months now, and the results are
remarkably counterintuitive.
This shell script is named "arbitron". Edit the "configuration information"
to work at your site, then type
arbitron -m netsurvey@glacier.ARPA
to mail the results to me. If you want to keep a copy of the results, type
arbitron -m "netsurvey@glacier myname"
where "myname" is your login name.
On Glacier, which is a Vax 750 with 250 user accounts, this program takes
about 5 minutes to run on a lightly-loaded system.
#! /bin/sh
# arbitron -- this program produces rating sweeps for USENET.
#
# Usage: arbitron > filename for local data
# arbitron -p net.foo to post to USENET
# arbitron -m person to mail results to an accumulator
#
# Run "arbitron -p" at the end of each month, which will post your site's
# arbitron report.
# To participate in the international monthly ratings sweeps,
# run "arbitron -m net-survey@glacier" every month. I combine the results
# and post the information to net.news.
#
# The names of users reading individual groups are not posted to the network,
# only the summary counts. Whether or not somebody reads a group is private
# information.
#
# By Brian Reid, Stanford.
# v1.2 September 18, 1985
# v1.3 March 8, 1986
#
# copied to a certain extent from the "subscribers"
# script posted by Blonder, McCreery, and Herron. The awk script to format
# the report was especially stolen from them.
############################################################################
# Configuration information:
tmpdir=/tmp
news=/usr/lib/news
# For uucp, try {sun, pyramid, decwrl, hplabs, bellcore}!glacier
summarypath="net-survey@glacier.ARPA"
hostname=`hostname`
PATH=$news:/usr/local/bin:/usr/ucb:/usr/bin:/bin
############################################################################
export PATH
# ---------------------------------------------------------------------------
trap "rm -f $tmpdir/arb.sel.$$ $tmpdir/arb.fmt.$$ $tmpdir/arb.tmp.$$" 0 1 2 15
opt1=${1-0}
case $opt1 in
-p) opt2=${2-ba.news.ratings};;
-m) opt2=${2-netsurvey@glacier};;
esac
set `date`
dat="$2 $6"
moption=0
case $opt1 in
-p) destination="$news/inews -t Monthly arbitron ratings for $hostname ($dat) -n $opt2"
poption=1;;
-m) destination="mail $opt2"
moption=1;
poption=1;;
*) destination="cat"
poption=0
esac
################################
# Here are several expressions, each of which figures out approximately how
# many people use this machine. Comment out all but 1 of them; pick the one
# you like best
########## fast but usually returns too big a number
lowUID=5
highUID=999
nusers=`awk -F: "BEGIN {N=0}\\$3>=$lowUID && \\$3<=$highUID{N=N+1}END{print N}" </etc/passwd`
########## slow but accurate if /usr/adm/wtmp has enough data
## wc is a Berkeley-ism, I think. We just want to count the lines in the
## standard output of sort -u
# set `last | colrm 9 | sort -u | wc`
# nusers=$1
################################
cat > $tmpdir/arb.sel.$$ << 'CAT'
/^net\..*: *[0-9].*$/ { nn=split($0,n,":"); print n[1], n[2] }
/^mod\..*: *[0-9].*$/ { nn=split($0,n,":"); print n[1], n[2] }
NF == 1 && $1 ~ /^[a-z]*$/{ print $1 }
CAT
sed -e "s/POPTION/$poption/" -e "s/MOPTION/$moption/" -e "s/NUSERS/$nusers/" -e "s/HOSTNAME/$hostname/" > $tmpdir/arb.fmt.$$ << 'DOG'
# makereport -- utility for "arbitron". Shamelessly copied from the
# similar script distributed with "subscribers.sh" by Blonder, McCreery, and
# Herron.
#
BEGIN { rdrcount = 0 ; reader = "" ; grpcount = 0 ; realusers = 0}
NF == 4 { # 4 fields means it is a /usr/lib/news/active entry
grpcount++
grpname[grpcount] = $1
grpnumber[$1] = grpcount
grplast[grpcount] = $2
grpfirst[grpcount] = $3
grpcounts[grpcount] = 0
grpreaders[grpcount] = ""
}
# 1 field means it's a user name
NF == 1 { rdrcount++; rdrname[rdrcount] = $1; rdrnumber[$1] = rdrcount
reader = $1}
# 2 fields means it's a .newsrc line
NF == 2 { gnum = grpnumber[$1]
n1 = split($2, n2, "-")
n3 = split(n2[n1], n4, ",")
lastread = n4[n3]
if (lastread >= grpfirst[gnum]) {
grpcounts[gnum]++
grpreaders[gnum] = (grpreaders[gnum] " " reader)
if (realuser[rdrcount] == 0) {
realuser[rdrcount]=1
realusers++
}
}
}
END {bigblanks = " "
printf("9999 Host\t\t%s\n","HOSTNAME")
printf("9998 Users\t\t%d\n",NUSERS)
printf("9997 NetReaders\t%d\n",realusers)
if (0 == MOPTION) printf("9996 \n9995 rdrs rating share traffic M/R newsgroup\n")
for (i=0; i < grpcount; i++) {
if (grpcounts[i] > 0) {
rating = (100*grpcounts[i])/NUSERS
share = (100*grpcounts[i])/realusers
traffic = grplast[i]-grpfirst[i]
if (grpcounts[i] != 0) ratio = traffic/grpcounts[i]
else ratio = 0
if (0 == POPTION) {
obuf = sprintf("%4d %5d %s", grpcounts[i], traffic, grpname[i])
nf = split(grpreaders[i], rdrs, " ")
obuf = substr ((obuf bigblanks),1,35)
width = 35
for (j=1; j <= nf; j++) {
nwidth = length(rdrs[j])
obuf = (obuf rdrs[j] " ")
width = length(obuf)
}
} else if (0 == MOPTION) {
obuf = sprintf("%4d %5.1f%% %4d%% %5d %5.1f %s", grpcounts[i], rating, share, traffic, ratio, grpname[i])
} else {
obuf = sprintf("%d %s\n",grpcounts[i], grpname[i])
}
printf("%s\n",obuf)
}
}
}
DOG
awk -F: '{printf "echo %s\negrep : %s/.newsrc\n",$1,$6}' \
</etc/passwd | sh 2>/dev/null | awk -f $tmpdir/arb.sel.$$ >$tmpdir/arb.tmp.$$
egrep '^net\.|^mod\.' $news/active | sort | \
awk -f $tmpdir/arb.fmt.$$ - $tmpdir/arb.tmp.$$ | sort -nr | \
sed -e 's/^999[0-9] //' | $destination
--
Brian Reid decwrl!glacier!reid
Stanford reid@SU-Glacier.ARPA