TIOBE index

TIOBE index

[Gautier write-only]

Hello,

From time to time I'm looking at the TIOBE Programming Community Index
  http://www.tiobe.com/index.php/content/paperinfo/tpci/index.html
The results, for long, are not very glorious for Ada, but I told
myself if was like it was.
For instance the latest ranking for Pascal, Fortran and Ada are:

  Tiobe ranking
  --------------
  Pascal  0.603%
  Fortran 0.563%
  Ada     0.376%

Now, I spent a few minutes to look at the figures myself, following
the famous citation "The only statistics you can trust are those you
falsified yourself".
So I made the search engine queries as described there:
  http://www.tiobe.com/index.php/content/paperinfo/tpci/tpci_definition.htm
including 1-year time windowing when available.
Here are the results (the simple queries without time windowing give
similar results):

        Google     Google Blogs  MSN        Yahoo!      Wikipedia
YouTube
 
------------------------------------------------------------------
Pascal  2'020'000  25'027        3'170'000  25'200'000  1'664      232
Fortran 1'510'000  11'768        1'640'000  7'900'000   949        113
Ada     2'140'000  110'310       2'670'000  49'300'000  1'764      70

With the weights being 23% for each engine, except YouTube with 7%, it
is obviously impossible to have Ada's ranking that is lower than
Fortran's and even than Pascal's.
Funny thing: the weights sum up to more than 100%, so don't forget to
divide by the sum of weights!
If you take for instance Pascal's ranking as a reference, the rankings
would be actually the following:
  Pascal  0.603%
  Fortran 0.294%
  Ada     0.668% <- uh-oh, almost the "A" category...

Of course the reference (here Pascal) would need as well a re-ranking,
up or down, which would influence Fortran's and Ada's - this would
mean to gather the full index data!

Did I miss something ? Some "soft factors" ?
I have an Excel sheet for those interested.
Or should we bug TIOBE about their statistics' quality ?
Or publish a fair, transparent index based on the TIOBE idea, but with
the publication of the figures ?
______________________________________________________________
Gautier's Ada programming -- http://gautiersblog.blogspot.com/
NB: For a direct answer, e-mail address on the following web site:
http://www.fechtenafz.ethz.ch/wm_email.htm

Re: TIOBE index

[Martin Krischik]

Am 08.03.2010, 12:53 Uhr, schrieb Gautier write-only  
<gautier_niouzes@hotmail.com>:


> Did I miss something ? Some "soft factors" ?
> I have an Excel sheet for those interested.
> Or should we bug TIOBE about their statistics' quality ?
> Or publish a fair, transparent index based on the TIOBE idea, but with
> the publication of the figures ?

Well, ask them - they usually answer.

Apart from that they lost my personal credibility whey they dropped usenet  
if favour of hipper stuff like YouTube.  And when all first 20 languages  
where in "danger" to be "A" rank they changed something else. I wrote a  
blog entry about that:  
http://ada-programming.blogspot.com/2008/01/tiobe-january-2008.html.

Also I noticed: "The first 100 pages per search engine are checked for  
possible false positives and this is used to define the confidence factor"  
- the confidence factor for Ada is not publicised.

Martin

-- 
Martin Krischik

Re: TIOBE index

[Gautier write-only]

Well, newsgroups contents are visible, copied several times, on
various search engines...
YouTube has a normalized weight of only 5.73%.

> Also I noticed: "The first 100 pages per search engine are checked for  
> possible false positives and this is used to define the confidence factor"  
> - the confidence factor for Ada is not publicised.

That could be the explanation...

Anyway, the idea of a "transparent" index, with the same rules, but
with publication of all figures, could be a nice project.
Of course automatized...
How complicated would it be to parse the HTML page returned by search
engines ?
For instance, the Google request gives somewhere in the HTML code:
*<p id=resultStats>*&nbsp;Results <b>1</b> - <b>10</b> of about
<b>1,440,000</b> for <b>ada programming</b>
Matching the above tag, the result is at reach of the hand.
Maybe some HTTP request to do with AWS ?
Any hint is welcome, I'm doing the rest of the code :-)

Gautier

Re: TIOBE index

[Tero Koskinen]

On Mon, 8 Mar 2010 03:53:34 -0800 (PST) Gautier write-only wrote:
> Or publish a fair, transparent index based on the TIOBE idea, but with
> the publication of the figures ?

There is also http://langpop.com/ which I find slightly more realistic
based on my own experiences, although that doesn't show Ada in very
popular light either.

-- 
Tero Koskinen - http://iki.fi/tero.koskinen/

Re: TIOBE index

[Gautier write-only]

> Maybe some HTTP request to do with AWS ?
> Any hint is welcome, I'm doing the rest of the code :-)

OK, found it: http://rosettacode.org/wiki/HTTP#Ada
G.

Re: TIOBE index

[Gautier write-only]

On Mar 8, 8:39 pm, Tero Koskinen <tero.koski...@iki.fi> wrote:

> There is alsohttp://langpop.com/which I find slightly more realistic
> based on my own experiences, although that doesn't show Ada in very
> popular light either.

That would not be a big issue for me - provided the conclusions can be
reproduced.
I only have something against dubious statistics...
G.

Re: TIOBE index

[jonathan]

On Mar 8, 11:53 am, Gautier write-only <gautier_niou...@hotmail.com>
wrote:
> Hello,
>
> From time to time I'm looking at the TIOBE Programming Community Index
>  http://www.tiobe.com/index.php/content/paperinfo/tpci/index.html
> The results, for long, are not very glorious for Ada, but I told
> myself if was like it was.
> For instance the latest ranking for Pascal, Fortran and Ada are:
>
>   Tiobe ranking
>   --------------
>   Pascal  0.603%
>   Fortran 0.563%
>   Ada     0.376%
>
> Now, I spent a few minutes to look at the figures myself, following
> the famous citation "The only statistics you can trust are those you
> falsified yourself".
> So I made the search engine queries as described there:
>  http://www.tiobe.com/index.php/content/paperinfo/tpci/tpci_definition...
> including 1-year time windowing when available.
> Here are the results (the simple queries without time windowing give
> similar results):
>
>         Google     Google Blogs  MSN        Yahoo!      Wikipedia
> YouTube
>
> ------------------------------------------------------------------
> Pascal  2'020'000  25'027        3'170'000  25'200'000  1'664      232
> Fortran 1'510'000  11'768        1'640'000  7'900'000   949        113
> Ada     2'140'000  110'310       2'670'000  49'300'000  1'764      70
>
> With the weights being 23% for each engine, except YouTube with 7%, it
> is obviously impossible to have Ada's ranking that is lower than
> Fortran's and even than Pascal's.
> Funny thing: the weights sum up to more than 100%, so don't forget to
> divide by the sum of weights!
> If you take for instance Pascal's ranking as a reference, the rankings
> would be actually the following:
>   Pascal  0.603%
>   Fortran 0.294%
>   Ada     0.668% <- uh-oh, almost the "A" category...
>

Thanks Gautier ... I hadn't realized how silly
the TIOBE index is until you explained how it works.
Another data point:

http://www.blackducksoftware.com/oss/projects#languageos

Still can't draw any real conclusions about the software
industry from this data point, but it least it is not
meaningless.

Jonathan

Re: TIOBE index

[Gautier write-only]

OK, first embryo here:
  http://sf.net/projects/lang-index/
Already working for some engines :-)
G.

Re: TIOBE index

[Gautier write-only]

> Did I miss something ?

Yes I did!
The request was: x+programming instead of: +"x programming"
There are of course fewer results (except for MSN!)

        Google  Google  MSN / Bing  Yahoo!   Wiki-  YouTube
        |       Blogs   |           |        -pedia |
        ---------------------------------------------------
Pascal  97'100  1'388   25'200'000  341'000  87     27
Fortran 38'100    458    3'830'000  147'000  50     3
Ada     30'700    338   29'100'000  124'000  97     5

... and the ranking, assuming Pascal is "correct", is

Pascal  0.603%
Fortran 0.218% (from 0.563%, way off!)
Ada     0.369% (from 0.376%, very close)

Probably a lesson of this story is that it is a very uncertain
exercise.

Gautier

Re: TIOBE index

[Marco]

On Mar 8, 3:04 pm, jonathan <johns...@googlemail.com> wrote:
> On Mar 8, 11:53 am, Gautier write-only <gautier_niou...@hotmail.com>
> wrote:
>
>
>
> > Hello,
>
> > From time to time I'm looking at the TIOBE Programming Community Index
> >  http://www.tiobe.com/index.php/content/paperinfo/tpci/index.html
> > The results, for long, are not very glorious for Ada, but I told
> > myself if was like it was.
> > For instance the latest ranking for Pascal, Fortran and Ada are:
>
> >   Tiobe ranking
> >   --------------
> >   Pascal  0.603%
> >   Fortran 0.563%
> >   Ada     0.376%
>
> > Now, I spent a few minutes to look at the figures myself, following
> > the famous citation "The only statistics you can trust are those you
> > falsified yourself".
> > So I made the search engine queries as described there:
> >  http://www.tiobe.com/index.php/content/paperinfo/tpci/tpci_definition...
> > including 1-year time windowing when available.
> > Here are the results (the simple queries without time windowing give
> > similar results):
>
> >         Google     Google Blogs  MSN        Yahoo!      Wikipedia
> > YouTube
>
> > ------------------------------------------------------------------
> > Pascal  2'020'000  25'027        3'170'000  25'200'000  1'664      232
> > Fortran 1'510'000  11'768        1'640'000  7'900'000   949        113
> > Ada     2'140'000  110'310       2'670'000  49'300'000  1'764      70
>
> > With the weights being 23% for each engine, except YouTube with 7%, it
> > is obviously impossible to have Ada's ranking that is lower than
> > Fortran's and even than Pascal's.
> > Funny thing: the weights sum up to more than 100%, so don't forget to
> > divide by the sum of weights!
> > If you take for instance Pascal's ranking as a reference, the rankings
> > would be actually the following:
> >   Pascal  0.603%
> >   Fortran 0.294%
> >   Ada     0.668% <- uh-oh, almost the "A" category...
>
> Thanks Gautier ... I hadn't realized how silly
> the TIOBE index is until you explained how it works.
> Another data point:
>
> http://www.blackducksoftware.com/oss/projects#languageos

When Assembler was listed quite high (when is the last time you used a
OSS project written in assembler!) I noticed this:
"broken it down by language using source lines of code."

with SLOC it needs to be normalized between languages for example
assign the factor 1.0 to C and have other factors based from this one.

"Effective lines of code represent a metric independent of code and
language style."

I couldn't find any factor comparison tables on the WWW.

TIOBE index

[Martin Krischik]

Hi

surprise, Ada is back in the top 20:

http://www.tiobe.com/index.php/content/paperinfo/tpci/index.html

And the «Very Long Term History» holds a surprise as well: Ada  was on  
place 3 once…

Regards

Martin
-- 
Martin Krischik

Re: TIOBE index

[jonathan]

On Sep 26, 4:21 pm, "Martin Krischik" <krisc...@users.sourceforge.net>
wrote:
> Hi
>
> surprise, Ada is back in the top 20:
>
> http://www.tiobe.com/index.php/content/paperinfo/tpci/index.html

Another (rather better!) language popularity index

   http://lang-index.sourceforge.net/

also shows a large September jump, but the interesting thing to me is
the source code (follow the links). To me it's very nice example of
Ada + Ada Web Server programming, not something I would ever know how
to do myself without examples like this.

J.

Re: TIOBE index

[Martin Krischik]

Am 26.09.2010, 23:53 Uhr, schrieb jonathan <johnscpg@googlemail.com>:

> On Sep 26, 4:21 pm, "Martin Krischik" <krisc...@users.sourceforge.net>
> wrote:
>> Hi
>>
>> surprise, Ada is back in the top 20:
>>
>> http://www.tiobe.com/index.php/content/paperinfo/tpci/index.html
>
> Another (rather better!) language popularity index
>
>    http://lang-index.sourceforge.net/
>
> also shows a large September jump, but the interesting thing to me is
> the source code (follow the links). To me it's very nice example of
> Ada + Ada Web Server programming, not something I would ever know how
> to do myself without examples like this.

True the numbers are a lot better. And I like the separation between  
“general-purpose and compiled” and the rest. Just a few pretty diagrams  
are missing…

Martin

-- 
Martin Krischik
mailto://krischik@users.sourceforge.net
https://sourceforge.net/users/krischik

Re: TIOBE index

[Yannick Duchêne (Hibou57)]

Le Sun, 26 Sep 2010 23:53:50 +0200, jonathan <johnscpg@googlemail.com> a  
écrit:
> Another (rather better!) language popularity index
>
>    http://lang-index.sourceforge.net/
Yes, nice to see Ada at position #17 on 100, but just look at what is  
behind : languages like Logo, ABC, Prolog, and the likes. Well, at least  
Ada is at the front of this category. Still could be better for a general  
purpose language. Ex. on the second column from the left, Ada is at  
position #11., while ML, which is, although I like its paradigms, mostly a  
language for theories and models, is at position #14… not really a kind of  
success here for such a language as Ada.

-- 
Si les chats miaulent et font autant de vocalises bizarres, c’est pas pour  
les chiens.

Re: TIOBE index

[Yannick Duchêne (Hibou57)]

Le Tue, 26 Oct 2010 04:41:42 +0200, Yannick Duchêne (Hibou57)  
<yannick_duchene@yahoo.fr> a écrit:
> Ex. on the second column from the left, Ada is at position #11., while  
> ML, which is, although I like its paradigms, mostly a language for  
> theories and models, is at position #14… not really a kind of success  
> here for such a language as Ada.
In that same column, even Basic is far ahead, at position #4 (!)


-- 
Si les chats miaulent et font autant de vocalises bizarres, c’est pas pour  
les chiens.