> This year's most noticeable changes were a 300% increase in Objective-C submissions, a 100% surge in C#, as well as a 33% increase in Javascript submissions while PHP lost -55%, Perl dropped -16%, and Java shrank -14%
Changes of this scale would lead me to question the methodology. There may well be some major trends at work, but it seems tremendously unlikely they are at this order of magnitude.
The oscillations mostly occur in languages with low market share, while the error of measurement is probably fairly constant as a percent regardless of how widely the language is used! Also...from the TFA:
> based on thousands of data points we've collected by processing over 100,000+ coding tests and challenges by over 2,000+ employers.
They are measuring what 2000+ employers decided to test, which seems to me could be quite noisy anyways. Take the study for what it is...its still useful, just not really universal in its conclusion.
I think they would have the same problem to a lesser degree if they sampled 20,000 employers. At the low end, you would still get a lot of fluctuation year-on-year because their simply are not enough data points.
Sadly the guy I sold that on to hasn't been doing as much with it as he could, like adding new data sources. I still think it's one of the better efforts out there, though, although of course I'm biased :-)
You should really check Fortran 2003/2008 :). The new Fortran is actually a pretty modern language - modules, coroutines, functions in functions etc ...
14 comments
[ 2.9 ms ] story [ 44.4 ms ] threadChanges of this scale would lead me to question the methodology. There may well be some major trends at work, but it seems tremendously unlikely they are at this order of magnitude.
> based on thousands of data points we've collected by processing over 100,000+ coding tests and challenges by over 2,000+ employers.
They are measuring what 2000+ employers decided to test, which seems to me could be quite noisy anyways. Take the study for what it is...its still useful, just not really universal in its conclusion.
In this regard, although TIOBE could have the same problem, it seems to use a greater number of different sources to compose their index.
Some context in the title would have been really helpful...
http://langpop.com/
That's better. PHP is not yet the new FORTRAN.
Fails basic sanity checks.