anton@mips.complang.tuwien.ac.at (anton

Now and then I post some data on programming language popularity,
based on the number of postings in the comp.lang.* groups. I just
took another sample, and below you can find the results. For more
results and the data, on which they are based, see
<http://www.complang.tuwien.ac.at/anton/comp.lang-statistics/>.

Feb 2006 (100d) Mar 2005 (185d) Dec 2003 (160 d) Feb 2003 (133 d)
news.tuwien.ac.at |---- news.individual.net ----|
posts language posts language posts language posts language
1 23622 java 39240 java 46418 java 59814 java
2 19346 c++ 36763 c++ 36633 c++ 44242 c++
3 17502 python 31491 python 28469 python 27054 c
4 15486 c 22787 c 26673 c 24438 python
5 14788 ruby 22115 ruby 21106 perl 23590 perl
6 10569 javascript 20733 perl 15889 clipper 18993 javascript
7 9374 perl 15874 javascript 15245 javascript 17271 clipper
8 8505 lisp 14157 lisp 13752 lisp 17131 basic
9 7963 php 13786 php 12930 basic 14921 ruby
10 6093 labview 12591 clipper 12852 ruby 13174 php
11 5488 tcl 11150 basic 12243 clarion 12383 clarion
12 5014 basic 8099 tcl 10468 pascal 12219 lisp
13 4696 clipper 7782 cobol 9472 labview 11312 tcl
14 4039 fortran 7225 fortran 9312 tcl 8823 pascal
15 3621 cobol 6714 pascal 6479 ada 7430 labview
16 2768 scheme 6574 labview 6064 cobol 7428 smalltalk
17 2180 smalltalk 6062 forth 5618 fortran 5872 fortran
18 1717 pascal 5798 ada 5593 functional 5189 ada
19 1634 forth 4245 smalltalk 3976 scheme 4718 forth
20 1419 vhdl 3035 scheme 3864 smalltalk 4348 cobol
21 1354 ada 2349 vhdl 3823 forth 3416 scheme
22 1347 idl-pvwave 2301 asm 3072 asm 3103 asm
23 1005 asm 2217 idl-pvwave 2236 vhdl 2423 vhdl
24 900 pl1 1785 rexx 1764 postscript 2207 functional
25 796 postscript 1725 awk 1703 objective-c 1947 idl-pvwave
26 762 awk 1605 functional 1693 idl-pvwave 1930 postscript
27 687 functional 1423 prolog 1078 pl1 1872 objective-c
28 637 verilog 1390 verilog 1013 verilog 1760 awk
29 551 apl 1344 postscript 1007 awk 1636 prolog
30 435 misc 1144 pl1 912 prolog 1218 pl1
31 428 prolog 1084 clarion 758 rexx 1204 verilog
32 316 logo 718 eiffel 755 misc 1188 rexx
33 312 clarion 703 misc 724 vrml 1184 vrml
34 305 rexx 644 vrml 658 eiffel 952 apl
35 302 eiffel 603 objective-c 657 logo 807 misc
36 204 vrml 602 apl 508 mumps 748 eiffel
37 198 objective-c 465 logo 443 apl 594 logo
38 187 mumps 371 ml 358 ml 549 mumps
39 101 asm370 353 mumps 272 modula2 464 ml
40 93 oberon 300 modula2 215 asm370 381 asm370
41 83 ml 251 pop 190 pop 246 dylan
42 79 modula2 181 asm370 147 oberon 153 modula2
43 47 dylan 115 oberon 98 modula3 152 modula3
44 30 pop 69 icon 63 dylan 105 pop
45 14 icon 68 dylan 39 idl 105 oberon
46 11 modula3 46 modula3 29 beta 93 icon
47 11 idl 40 idl 24 icon 21 idl
48 3 clos 15 hermes 21 sather 13 limbo
49 2 beta 9 beta 19 clos 13 clos
50 1 sather 8 clos 7 clu 12 sather
51 0 visual 7 prograph 6 prograph 9 beta
52 0 prograph 6 clu 6 limbo 4 clu
53 0 limbo 2 sather 3 visual 2 prograph
54 0 hermes 2 limbo 1 php
55 0 clu 0 visual 1 hermes

So what do we see? The top 9 are the same as last time, with the same
ranks (except for a switch between Perl and Javascript). Also, most
other ranks are roughly the same, including Forth's. Some prominent
languages that seem to be in decline are Basic and Pascal (and, less
prominent, Clipper). Ruby rose quickly, but now seems to settle, and
I don't see any other consistently rising languages in the top 20
(maybe Scheme).

- anton
--
M. Anton Ertl http://www.complang.tuwien.ac.at/anton/home.html
comp.lang.forth FAQs: http://www.complang.tuwien.ac.at/forth/faq/toc.html
New standard: http://www.complang.tuwien.ac.at/for...forth200x.html

andrew haley

I suspect, but cannot prove, that this total is more closely
correlated with the popularity of programming languages amongst
inexperienced programmers, especially students.

If you're going to claim that the number of postings in the
comp.lang.* groups measures something other than the number of
postings in the comp.lang.* groups (such as programming language
popularity) you should provide evidence.

Andrew.

anton@mips.complang.tuwien.ac.at (anton

Andrew Haley <andrew29@littlepinkcloud.invalid> writes:


That would certainly explain the peak that Java had some time ago.
But is it true in general? And even if it is, if a language is
popular among inexperienced programmers, it is probably also popular in general.


If by that you mean that courses using a language determine the
rankings in this list, I doubt that very much. E.g., Scheme is used
in more courses (especially introductory programming courses) than
Python or Ruby, yet Scheme has rank 16, Ruby has rank 5 (and >5 times
as much traffic) and Python has rank 3 (and >6 times as much traffic).

If you mean that students who pick up a language on their own are
responsible for a lot of traffic in the language's newsgroup, that may
be the case. In that case I would expect these numbers to be an
indicator of future popularity in the more general programmer population.

I think we had that already, didn't we? Wasn't it that the ranking
disagrees with your beliefs about programming language popularity, but
you don't have any evidence for your beliefs whatsoever?

The evidence I have is that it does not disagree too much with other
ways of estimating programming language popularity.

Another argument, of course, is that the amount of discussion about a
programming language is a better indicator of popularity than, e.g.,
the number of lines written in the programming language (even if that
could be measured); just because someone churns out Cobol or Forth
code because he is paid for it does not mean he is actually interested
in the language and likes it.

- anton
--
M. Anton Ertl http://www.complang.tuwien.ac.at/anton/home.html
comp.lang.forth FAQs: http://www.complang.tuwien.ac.at/forth/faq/toc.html
New standard: http://www.complang.tuwien.ac.at/for...forth200x.html

andrew haley

Sure, but that does not mean that popularity dominates the number of
postings. I'm sure it correlates in some vague way, though.


Yes, we did, which is why I'm astonished you're posting this again.


As I said before, I don't know how popular these programming languages
are, but I am skeptical that this measurement is valid.

Which may be equally bogus, no?

There may have been some real surveys, and perhaps even ones that have
been properly conducted. That would be far more credible.

I don't think so. It correlates directly, I suspect, with the amount
of trouble a person is having when programming in a language. Now, it
is very likely that this is correlated in some way with a programming
language's popularity, but there may be other factors. And it is
quite possible that other factors might be significant.


No, and just because someone churns out news postings does not mean he
is actually interested in the language and likes it either! Many of
the newsgroup postings I've read in the comp.lang.* groups are of the
form "I have this homework/job to do but I am too dim/tired/lazy to
figure it out for myself."

Andrew.

marc olschok

I do not know much about any particular beliefs that Andrew might have.
While your statistics satisfy my curiosity, I am also reluctant to view
the number of postings as a reliable indicator for popularity:

(a) the number of postings cannot distinguish between "many posters"
and "few posters generating plenty of postings".

(b) the number of postings cannot distinguish between "many short threads"
and "few long threads".

(c) counting postings cannot tell about content, topic or quality of
the postings. In particular it does count OT threads.

( all this will be more obvious as this thread proceeds :-)

Now, if all those different newsgroups share the same pattern with
respect to (a)-(c) this would not affect the outcome much. But I am
not sure if this is the case.

Another point: did you count Message-IDs or the numbers of postings
within a particular group as provided by the nntp-server?
In the latter case crossposting would not have been taken into account.
This _does_ affect different NG's differently.
For example consider the cluster functional/ml/scheme/lisp.


I agree. But still I fear that at best the popularity of language newsgroups
has been measured.

Marc

julian v. noble

I'd be intrigued to know why Python is so consistently popular.

That is, some of my confreres here must know the language, and I
would value their opinions about what's to like (or dislike).

I know why _I_ like Forth, but I also edit a column in Computing in
Science and Engineering where a number of people seem to like
Python when they are not using MatLab or Mathematica. But they
have never explained it to me in a convincing way.

--
Julian V. Noble
Professor Emeritus of Physics

http://galileo.phys.virginia.edu/~jvn/

"As democracy is perfected, the office of president represents, more and
more closely, the inner soul of the people. On some great and glorious
day the plain folks of the land will reach their heart's desire at last
and the White House will be adorned by a downright moron."

--- H. L. Mencken (1880 - 1956)

andrew haley

It does all the scripty things that PERL does but it doesn't look like
line noise.

Andrew.

paul e. bennett

Number of lines of code as a measure would favour those languages that were
less expressive. So, LOC is not really a good basis for language
comparison. Now, completed function points in each language would be a
better indicator. However, this is one measure that is very difficult to
collect. Likewise, the language use statistics from the magazine surveys,
subscription cards or exhibition ticket applications restrict the set to
the well known five or six and a box for other. Forth is always in the
"other" category.

In the end, you know exactly what you can prove with statistics.

....... nothing!!. -- ************************************************** ******************
Paul E. Bennett ....................<email://peb@amleth.demon.co.uk>
Forth based HIDECS Consultancy .....<http://www.amleth.demon.co.uk/>
Mob: +44 (0)7811-639972
Tel: +44 (0)1235-811095
Going Forth Safely ..... EBA. www.electric-boat-association.org.uk..
************************************************** ******************

elizabeth d rather

There have been a lot of attempts over the years to measure language
popularity, and I'm not really impressed by any of them. For example,
Embedded Systems Programming occasionally publishes the breakdown of
languages their subscribers profess to be using, but does that say anything
beyond the behavior of ESP subscribers? In other words, I'm not sure it's
possible to provide *any* "evidence" of language popularity other than a
measurement within a particular demographic (e.g. ESP subscribers or
comp.lang.* posters). FWIW, we have a fairly large population of active
customers, the vast majority of whom never post on c.l.f. So, the c.l.f
demographic is not at all representative of our customer base. I have no
idea how many of them subscribe to ESP, although I know the majority are
professionally engaged in embedded systems programming.

All good points.

Agreed.

Cheers,
Elizabeth

--
==================================================
Elizabeth D. Rather (US & Canada) 800-55-FORTH
FORTH Inc. +1 310-491-3356
5155 W. Rosecrans Ave. #1018 Fax: +1 310-978-9454
Hawthorne, CA 90250
http://www.forth.com

"Forth-based products and Services for real-time
applications since 1973."
==================================================

paul marciano

Indeed, the number of posts in a newsgroup may not indicate anything
about the usage of the language. VHDL and Verilog and used by pretty
much all logic designers but the majority of discussions take place on
comp.arch.fpga, and even then the volume isn't that high.