Publicity (was: Re: Why is 2.0 in the works already?)

[Bruno Medeiros]

Don Clugston wrote:
> David B. Held wrote:
>> Lars Ivar Igesund wrote:
>>> Anders Bergh wrote:
>>>
>>>> On 6/19/07, Lars Ivar Igesund <larsivar at igesund.net> wrote:
>>>>> The only reason to worry about TIOBE, is that high rankings may boost
>>>>> knowledge about D. As is rather obvious by looking at the list, D's
>>>>> numbers are most likely highly inflated, and a reason for this is
>>>>> suggested at
>>>>>
>>>>> http://cdsmith.wordpress.com/2007/06/18/is-tiobe-fatally-flawed/
>>>> I just read that post, and scrolling down to the comments makes the
>>>> post less notable. Google apparently cuts results at 1000 results, and
>>>> removes duplicates, shrinking the numbers which caused his search
>>>> results to be even more flawed than TIOBE's.
>>>
>>> Yes, his ranking was definately wrong too, don't necessarily make 
>>> TIOBE's
>>> more correct though :) If TIOBE use Google, the argumentation would 
>>> affect
>>> them too in some form.
>>
>> TIOBE's rankings are certainly suspect, but all the hoopla about 
>> Google is just wrong.  Google does not remove "duplicate" hits, 
>> because it does not index duplicate hits.  That would be a stupid 
>> search engine. Instead, it removes pages that look like they came from 
>> the same site and possibly the same area of a site, and thus, may not 
>> present interesting new information to the user.  A simple example is 
>> searching for a term that happens to be on the footer of a bunch of 
>> pages on a site.  The hits are not "dupes", but they aren't 
>> interesting, either.
>>
>> Clearly, assuming that all result sets are < 1000 is just silly, and 
>> the blogger should have known better.  The estimated Google hit counts 
>> are probably accurate within an order of magnitude, based on various 
>> searches I've done where I compared the initial hit count to what 
>> Google  says after I've forced it to do an exact count (by visiting 
>> all the pages).  So the TIOBE page counts are probably fairly 
>> reasonable.  What is not reasonable is any interpretation of those 
>> results that mentions "popularity", "buzz", "community", or 
>> "zeitgeist".  Even less reasonable is any assumption that languages 
>> near the top of the list are "better" than those not near the top for 
>> anything but a narrow and specific definition of "better".
>>
>> Dave
> 
> And despite what Tiobe says, it's certainly not an indication of 
> "mainstreamness".
> Just look at the graph for Fortran shows how meaningless the Tiobe 
> ranking is.
> It ought to be one of the most stable languages on the list. Yet 
> according to Tiobe, it's dropped by a factor of 3 since 2003. It's 
> nonsense.

If I understood the rating correctly, that happens because the scoring 
is relative and not absolute. So even if the "popularity" (hit count) of 
a language stays constant (which should be the case for Fortran), it's 
rating can change simply if other languages become more "popular".

-- 
Bruno Medeiros - MSc in CS/E student
http://www.prowiki.org/wiki4d/wiki.cgi?BrunoMedeiros#D