______________________
LANGUAGES POPULARITY
Nicolas Herry
______________________
2017/03/31
1 Languages popularity
======================
[An article posted on monday on Revolution Analytics] echoes the
latest report published by Redmonk, which shows which programming
languages are most popular. In this report, Python climbed to position
#3 and R registered a slight step back, to #14. So, what does it tell
us? Well, not much, I'm afraid.
[An article posted on monday on Revolution Analytics]
http://blog.revolutionanalytics.com/2017/03/redmonk-jan-2017.html
1.1 Useful information
~~~~~~~~~~~~~~~~~~~~~~
Of course, the first thing is that putting all languages in one bag
seems a bit silly, to say the least. How often do you hesitate between
C and Javascript? And why is the popularity contest run every six
months by Redmonk and every month by TIOBE? How does that relate to
anything in the real world?
But it´s not just the methodology I have a problem with. I believe the
value of any piece of information is directly proportional to the
value of the decisions you can make based on it. So, what decision can
we make based on the knowledge that [Redmonk] or [TIOBE] reports that
some language (let's call it "wiggly") that used to be #11 is now #15,
for example?
- Shut down all projects written in wiggly
- Shut down all future plans to write anything in wiggly
- Convert all projects written in wiggly to a language in the top 10
- Organise trainings for all wiggly people willing to evolve to
something more fashionable and fire all those who won't
Well, I guess we can agree all the above seems a bit over the top. But
if you were to make such a decision, what would you do if in the next
report, dear wiggly was to climb back to #8? And if you aren't to make
any such decision, exactly what kind of decision can you make based on
the report?
Well, since this is not about the technical merits of the languages
themselves, maybe it's about the market. So, what does it say about
the market? Not much. We get a picture of how much activity GitHub and
StackOverflow see for these forty languages. It doesn't say anything
about the value of mastering these languages on the market and it
doesn't tell anything about the value generated by these languages on
the market either.
So, if it's not about the technologies themselves, not about the value
of the skills, the size of the market or what technology is right,
then what is it about? I guess the answer is that it reflects trends
and fashions, and I'm not sure I want to make any decision based
solely on that.
[Redmonk] https://www.redmonk.com
[TIOBE] https://www.tiobe.com/tiobe-index/
1.2 Making decisions
~~~~~~~~~~~~~~~~~~~~
So, if these kinds of reports aren't of any help in making decision,
then what is? What should we be considering? I'm not stupid enough to
believe I hold the definitive answer to that question, but to me,
there are a few things that you must take into account, always.
1.2.1 The team
--------------
A project is first and foremost made up of the people involved in
it. Besides being all sweet and cute, this sentence should act as a
reminder that to get a project done, you need people with the right
skills. So, whenever I need to decide on some technology, I first ask
myself what these people know already. Especially when a project is
about exploring new grounds, it's better to do it with a toolset you
already master. Starting a new project, in a new domain, in a new
language and possibly with new people seems a little risky to me. This
is not to say that any novelty should be ruled out from the start:
introducing new languages and technologies is healthy and useful, but
I believe it should be done while keeping in mind a clear career path
for the people on the team. In other words, this is long-term
thinking, not something you revise entirely every month or even twice
a year. As a side effect, not jumping on everything new and shiny can
help getting a product well written, easy to maintain (or rather, not
completely horrible to maintain, since you are less likely to discover
unexpected shortcomings with the technology) and with a better time to
market as a result.
1.2.2 Community and ecosystem
-----------------------------
If the reports from Redmonk and TIOBE give some idea about the size of
the respective communities behind the different languages, they tell
nothing about the culture their programmers evolve in. Are they a good
match to yours? Is there enough documentation available on your
favourite support, be it books, forums, tutorials, mailing lists or
IRC channels? Or are there enough open source code you can study, if
this is how your team likes to learn about technology? Also, are there
enough libraries available so you won't be pioneering your field in a
language new to you? Another thing is that a large community is not
always a good thing: as a community grows, the average skill of
potential hires you can get from it gets lower and lower. This may or
may not be a problem, depending on the project or the sector, but it
is important to factor this in the equation.
1.2.3 Cost
----------
The two considerations above already cover this item to a great
extent, so no need to rehash everything I already said. The question
remains the same I ask myself with any project decision: how much does
introducing a new technology cost in short term and in the long run?
It's not much of a surprise, but no popularity contest can provide a
proper answer to that question.
So, in conclusion, I really can't see the point of these reports. In
the latest TIOBE Index for March 2017, the star is Swift, which is
entering the top 10. Who would have guessed programming on the iPhone,
with the technology stack actively pushed by Apple is a popular thing
do? I certainly didn't, but I'm not losing anymore time; I already
started rewriting all my MATLAB code in wiggly.