Technology and Trends
 USENET Archives
  
From: hta@uninett.no (Harald T. Alvestrand)
Newsgroups: comp.os.linux.announce
Subject: Counting Linux users: an attempt
Date: 30 Sep 1993 22:07:17 GMT
Approved: linux-announce@tc.cornell.edu (Matt Welsh)
Message-ID: <28flal$sm3@samba.oit.unc.edu>

Hello,

there have been many attempts at defining the size of the Linux
user base. None of them contained hard data.

Now, in an attempt to establish a lower limit on the number of
Linux users, I have put up a MAIL SERVER that does counting.

So, if you are a Linux user, and want to be counted, send an E-mail
to the address

    linux-counter@uninett.no

with the subject being one of

    I use Linux at home
    I use Linux at work
    I use Linux at home and at work
    I do not use Linux

The scanner looks for the words "home", "work" and "not" to determine
where you use Linux, and sends you back an ACK that includes the current
state of the count.

The whole subject line is kept, along with your FROM address; if you
want to change your opinion, just send another mail to linux-counter.

DO NOT SEND VOTES TO MY MAILBOX!
PUT YOUR VOTE IN THE SUBJECT LINE, NOT THE BODY OF THE MESSAGE!

The Reply-to: line of this message is set to the counter, so you may try
to just juse "reply" to this message.

The current count (after announcing it on comp.os.linux.misc for 1 day,
and on no.linux for 2 days):

Total 236 users

BY COUNTRY
  56 edu       35 com       28 no        26 de        18 uk      
  13 nl         9 se         9 gov        7 fi         6 ca      
   4 at         3 dk         2 bitnet     2 unknown    2 is      
   2 us         2 fr         2 net        2 org        1 au      
   1 jp         1 nz         1 za         1 tw         1 cz      
   1 mil        1 ch      

PROBABLE DISTRIBUTION OF PLACES WHERE LINUX IS USED
 160 at home
  55 at home and at work
  11 somewhere
   8 at work
   2 not used

DISCUSSION:

I know that any counting mechanism is biased. And given what Linux is,
no mechanism will get registrations from those who don't want to be counted.
There are also a number of ways to confuse the issue, such as:

- Mass registrations from bogus accounts (aka Usenet "votes")
- Junk registrations of "I don't care what Linux is" and such things
- Trying to estimate the number of non-"voting" persons

But after a week or five of this, I will have a number, and we can start
debating whether that number is interesting or not.

Have fun voting!
-- 
                   Harald Tveit Alvestrand
                Harald.T.Alvestrand@uninett.no
      G=Harald;I=T;S=Alvestrand;O=uninett;P=uninett;C=no
                      +47 7 59 70 94
My son's name is Torbjxrn. The letter between "j" and "r" is o with a slash.


-- 
Send submissions for comp.os.linux.announce to: linux-announce@tc.cornell.edu

From: hta@uninett.no (Harald T. Alvestrand)
Newsgroups: comp.os.linux.announce
Subject: Linux user count passes 2000 users
Date: 7 Oct 1993 21:46:20 GMT
Approved: linux-announce@tc.cornell.edu (Matt Welsh)
Message-ID: <2922nc$76o@samba.oit.unc.edu>

(To get included into the count, send an E-mail to linux-counter@uninett.no
with one of the subject lines:

  I use Linux at home
  I use Linux at work
  I use Linux at home and at work
  I don't use Linux

You will get an E-mail back saying how your vote was counted)


LINUX counter status, as of Thu Oct  7 10:31:59 MET 1993

Number of messages processed: 2162
Number of current votes     : 2064 (98 revotes)
Number of Linux users       : 2029 (35 non-users voted)


PROBABLE DISTRIBUTION OF PLACES WHERE LINUX IS USED
1300  62% at home
 505  24% at home and at work
 156   7% at work
  68   3% somewhere
  35   1% not used

BY DOMAIN
 496 edu      303 com      291 de       124 uk        88 ca      
  80 nl        73 au        70 fi        51 se        51 no      
  38 net       38 gov       37 at        35 dk        27 org     
  26 fr        20 ch        19 unknown   15 za        15 uucp    
  14 be        13 mil       11 it         8 jp         8 us      
   6 bitnet     6 tw         5 pt         5 ie         5 pl      
   5 nz         5 es         4 hk         4 il         4 is      
   3 hu         3 br         3 co         3 cl         3 gr      
   2 su         2 si         2 ee         1 mx         1 cz      
   1 cs         1 th         1 hr         1 ro         1 gb      
   1 in      

BY COUNTRY
(I like sorting by #/population, because that puts Norway near the top :-)
Country                         #Linux   MPop Lin/M
============================================================
Iceland                              4    0.3  15.7 
Finland                             70    5.0  14.0 
Norway                              51    4.3  11.9 
Denmark                             35    5.1   6.9 
Sweden                              51    8.6   5.9 
Netherlands                         80   14.9   5.4 
Austria                             37    7.6   4.9 
Australia                           73   17.1   4.3 
Germany                            291   79.1   3.7 
USA                                885  249.6   3.5 edu,com,gov,mil,org,us
Canada                              88   26.6   3.3 
Switzerland                         20    6.7   3.0 
United Kingdom                     125   57.2   2.2 uk,gb
New Zealand                          5    3.4   1.5 
Ireland                              5    3.5   1.4 
Belgium                             14    9.9   1.4 
Estonia                              2    1.6   1.2 
Slovenia                             2    2.0   1.0 
Hong Kong                            4    5.9   0.7 
Israel                               4    6.3   0.6 
South Africa                        15   30.2   0.5 
Portugal                             5   10.3   0.5 
France                              26   56.2   0.5 
Greece                               3   10.1   0.3 
Taiwan                               6   20.3   0.3 
Hungaria                             3   10.5   0.3 
Chile                                3   13.5   0.2 
Croatia                              1    4.8   0.2 
Italy                               11   57.7   0.2 
Poland                               5   38.4   0.1 
Spain                                5   39.5   0.1 
Czech Rebublic                       1   10.0   0.1 
Colombia                             3   34.3   0.1 
Japan                                8  123.3   0.1 
Czechoslovakia (former)              1   15.7   0.1 
Romania                              1   23.2   0.0 
Brazil                               3  158.2   0.0 
Thailand                             1   57.6   0.0 
The World (somewhere)               78 5000.0   0.0 net,int,bitnet,uucp,unknown
Soviet Union (former)                2  147.4   0.0 
Mexico                               1   81.4   0.0 
India                                1  844.0   0.0 

Comments:

* YES, I know that this count is a severe undercount, because

  a) not all Linux users can send E-mail
  b) of those that can send E-mail, a lot don't read c.o.l.announce

  Nevertheless, this count has already passed the previous largest real count
  of Linux-interested people (the c.o.l split vote, which had 1842 votes).

* YES, the allocation of "com" to the US and "bitnet" to "the world" is
  more or less arbitary. I will think about getting a better split across
  countries in a week or two, but at least, we get an idea.

* I don't think letting people register others into the count is a good
  idea. How shall we know whether two people are registering the same friend?

* I count users, not machines. BBS users, who neither know nor care that
  their BBS is running on a Linux system, should probably not be counted,
  while "dialin UNIX" users with E-mail access probably should be. I will
  think about counting machines some time later (probably not before November).

* So far, at least 38 of the replies from the counter have bounced. This
  means that 1.7 % of the mail has problems. I consider this good!

One more word:

PLEASE crosspost the vote request to local newsgroups, BBSes and the like!
I believe many German Linux users (for instance) would read mostly their
"own" newsgroups, and would get the information by such paths.

>From X.400 networks, the addresses

S=linux-counter;O=uninett;PRMD=uninett;ADMD= ;C=no
S=linux-counter;O=uninett;PRMD=uninett;ADMD=uninett;C=no

might work.
Have fun!

-- 
                   Harald Tveit Alvestrand
                Harald.T.Alvestrand@uninett.no
      G=Harald;I=T;S=Alvestrand;O=uninett;P=uninett;C=no
                      +47 7 59 70 94
My son's name is Torbjxrn. The letter between "j" and "r" is o with a slash.

-- 
Send submissions for comp.os.linux.announce to: linux-announce@tc.cornell.edu