A Brief Commentary on the March Tiobe Index

The March Tiobe Index is out. I follow Java (1, up short term), Python (3, continual rise), R (11, steady mid-term) MATLAB (15, down mid-term), SAS (21, steady-ish) and Julia (49, wildly down?) as my interests tend towards deployment of quant and data science models into production architectures. Some thoughts on these platforms are below. What do you think?

Full disclosure. I work for a JDK/JVM (Java) provider. I used to work at MathWorks (MATLAB) and after that a Python-using geoscience firm. Friends & former colleagues use Python, R and a few experiment with Julia for data science, quant, analytics and other projects. I also track SAS where I have former colleagues and since they embed Java as part of their software delivery, our paths have intersected on several occasions of late. 

I see three factors driving Java's 1 and 2 year rise, though the language has featured at the top of the index for decades: i) new-found positivity among the OpenJDK community after Oracle’s bombshell; ii) distributed data-oriented container workflows (aids Python too) iii) some see increasing interest for Java in the embedded markets given the interconnected data flows between device, gateway and cloud.

Python is up from 2014, with a notable self-perpetuating rise from 2018 driven by data science, quant, also many university science & engineering departments facilitated by Jupyter-led notebook collaboration. It has become a general-purpose tool of choice, extremely popular on the blogosphere. The Python community has overcome the difficult Python 2 to Python 3 transition pretty quite well. On the flip side, corporate management and even some on the blogosphere are now less forgiving of Python hype, some users falling by the wayside not quite having reinvented the wheel “for free”, occasionally leaving problematic legacy stacks.

Rumours of R’s demise at the expense of Python are exaggerated, though it has retrenched to being stats-first rather than a general-purpose behemoth. R’s Tiobe rating is gently oscillating after a dramatic fall from 2017/18 Peak R. SAS too is gently oscillating over the medium term, but unlikely to reach the heights of its late 2000s ratings.

My alma mater MATLAB did well in December, in February reverting to its mid-term (from 2017) descent. Does this add impetus to make core MATLAB “free”? Were MATLAB free, the pace of development in science and engineering would surely look very different very quickly. Still, MATLAB today holds a higher Tiobe rating than during its supposed pre-Python 2005-2015 peak. Also, its new [commercial software-focussed] Gartner Data Science rating stands it in good stead, emphasizing completeness of vision, in particular crediting embedded code generation capabilities, i.e. making data science useful for the many and not just the few. Perhaps it is not so bad being Python’s favorite +1 and occasional Pythonista punchbag.

To my surprise, Julia has dropped like a ton of bricks and is in danger of dropping out of the Tiobe 50, just above the largely legacy S language. Marketing noise from the summer of 2019 and JuliaCons perhaps not backed up by substantial use? I’d be really surprised given what I hear, but who knows?