Netscan and Python

[Paul Prescod]

According to Netscan,

comp.lang.python was the 19th most popular Usenet newsgroup in 1999
It rose to 8 in 2000
It stayed in 8th place in 2001 (I don't know why that year was flat)
It rose to 6 in 2002
It jumped to 2 in 2003 (comp.lang.c is still almost twice the size
despite the fact that it shrunk in both 2002 and 2003)

It would be crazy to say that this proves that Python is the second most
popular programming language, but does show how astonishingly quickly
Python's popularity has been growing.

According to Netscan, comp.lang.python is the only comp.lang. group in
the top 5 to experience positive growth in 2003.

The drop-offs were as follows:

* c: -17%
* Python +13%
* java.programmer: -35%
* perl.misc: -17%
* javascript: -16%

Rounding out the top 10 we have:
* clarion: +35% (what's going on here??)
* Ruby -4%
* PHP: +70% (comp.lang.php was only created mid-2002)
* clipper: -12%
* Lisp: 0%

I would say:

a) Usenet itself is not very healthy

b) Usenet does not very directly reflect usage patterns

c) Nevertheless, Python's consistent growth across several years does
seem to reflect other indicators.

Paul Prescod

 

[Aahz]

According to Netscan,
comp.lang.python was the 19th most popular Usenet newsgroup in 1999

Based on the rest of the post, I'm assuming you meant to say
"comp.lang.*" instead of "Usenet". There's also the question of whether
popularity is best measured by number of posts -- I'm sure the
readership of rec.humor.funny still far outweighs comp.lang.python.

 

[Nelson Minar]

According to Netscan, comp.lang.python is the only comp.lang. group in
the top 5 to experience positive growth in 2003.

Coincidentally, I just saw Marc Smith talk about Netscan at ETech:
http://conferences.oreillynet.com/cs/et2004/view/e_sess/4797

The thing he said that caught my attention is that comp.* is slowly
diminishing over the years. He showed a graph that indicated
microsoft.* has just about surpassed comp.*, both in number of posts
and number of authors.

Another random thought - folks often look at the relative lack of
Python books vs. Perl books as a sign of Python's weakness. Isn't it
strength? Python is so simple and the online docs are so good I never
even thought about learning it from a book.

 

[Martin Christensen]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Nelson> Another random thought - folks often look at the relative lack
Nelson> of Python books vs. Perl books as a sign of Python's
Nelson> weakness. Isn't it strength? Python is so simple and the
Nelson> online docs are so good I never even thought about learning it
Nelson> from a book.

I wouldn't think so. The online documentation (mostly the Python
library documentation) is where I go to have most of my Python
questions answered, but that sort of documentation cannot and should
not cover everything. Consider, for instance, David Mertz' book 'Text
Processing in Python' (http://gnosis.cx/TPiP/). It uses several
hundred pages to expand on topics that in the Python library reference
is covered in only a few pages, this because they have very different
aims: the library reference exists to give a brief overview of the
capabilities of each module, and TPiP is a complete topic guide to
text processing.

I've learnt Python almost exclusively from online documentation, too.
However, for in-depth discussions of more advanced topics, I've found
Python in a Nutshell and The Python Cookbook to be invaluable, and
I've definitely gained insight from them that I would not have from
the online documentation. On the other hand, not all books are
worthwhile, and as such I've still not demanded my copy of Andre
Lessa's Python Developer's Handbook back from the friend I lent it to,
since I've found it both less informative and practical than the
library reference, and indeed it seems to be little but a reworded
version of said library reference.

So in short, books are invaluable for digging deeper than most online
documentation does. Of course, it's fortunate that we are blessed with
some of these books being also made available online.

Martin

- --
Homepage: http://www.cs.auc.dk/~factotum/
GPG public key: http://www.cs.auc.dk/~factotum/gpgkey.txt
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)
Comment: Using Mailcrypt+GnuPG <http://www.gnupg.org>

iEYEARECAAYFAkAv+c4ACgkQYu1fMmOQldVUhACg1TnmtARmboZgPLovH5WBFpJN
oxIAnRKaGU0h0c5DVJsWSp91855UfXPS
=H1CE
-----END PGP SIGNATURE-----

 

[Paul Prescod]

Based on the rest of the post, I'm assuming you meant to say
"comp.lang.*" instead of "Usenet". There's also the question of whether
popularity is best measured by number of posts -- I'm sure the
readership of rec.humor.funny still far outweighs comp.lang.python.

True. But as you surmised, rec.humor.funny is not in the domain of
newsgroups that interest me for this discussion. It seems logical to me
that popularity would follow number of readers and that number of posts
would tend to follow number of readers. The relationship is certainly
not direct: the more mainstream a language is (e.g. C++) the less likely
one is to use a newsgroup as a support channel for it. Also, in the case
of Ruby, cultural issues may be relevant. Perhaps some Japanese mailing
list is booming.

Paul Prescod

 

[Paul Prescod]

...
Another random thought - folks often look at the relative lack of
Python books vs. Perl books as a sign of Python's weakness. Isn't it
strength? Python is so simple and the online docs are so good I never
even thought about learning it from a book.

That is also true for me. Perl 4 is almost certainly the last language I
learned primarily from a book and I really wanted to hurl that Camel
book through the window.

I also think that books are a trailing indicator. The most likely person
to buy a book is someone about to be interviewed for a Python job or
someone who needs to maintain a large Python code base. If you are just
playing around with Python for your own fun (as early adopters tend to
be), you may not want to shell out.

The recent shift in this newsgroup to newbies with very specific
problems suggests to me that the market is starting to become ready for
specialized Python books (e.g. VB to Python, Perl to Python, Python for
multimedia, Python algoritms for your homework, ...).

That said, Tim O'Reilly recently told me that Python books are "merely"
flat while other book sales dropped off after the boom. Also, Learning
Python is selling very well. I am quite confident based on a variety of
anecdotal measures that 2003 was a big growth year for Python.

Given that Python's growth is not driven by nor capped by the growth of
any other technology (e.g. the Internet, the Web, Linux, etc.) I expect
continued rapid growth in 2004 and 2005.

Paul Prescod

 

[A.M. Kuchling]

The recent shift in this newsgroup to newbies with very specific
problems suggests to me that the market is starting to become ready for
specialized Python books (e.g. VB to Python, Perl to Python, Python for
multimedia, Python algoritms for your homework, ...).

Indeed; I think it's been ready for a while. We have introductory books
for programmers and for complete non-programmers, books that are concise
(consider Beazley's Essential Reference, which covers the whole language in
~20 pages) and books that are large. I don't think there's any point in
writing an introductory or a reference book any more; it would be too
difficult to improve upon the existing titles.

--amk