The unexpected effectiveness of Python in science

Posted Jun 5, 2017 0:26 UTC (Mon) by gdt (subscriber, #6284)
In reply to: The unexpected effectiveness of Python in science by mikapfl
Parent article: The unexpected effectiveness of Python in science

"Free" also goes a some way to explaining the rising dominance of R in statistical computing. Even Microsoft has seen the way the wind is blowing and put substantial resources into improving the language's performance (on its own operating system, but source available for sharing with others thanks to the GPL).

Part of the other reason for the rise of Python is that it is a "real" programming language. The current generation of scientists are applying large-scale computing to their investigations and are aware that this is best done through extending an existing well-regarded language rather than constructing a domain-specific language which may then lack the essentials for programming-in-the-large. Especially as writing a Python module, or even a Python C API module, is a small fraction of the work of writing a domain-specific language. Moreover there's a huge network effect: in pulling OTDR data into Python I got all the file parsing, statistical analysis, graphing, and archiving at low cost. These are also present in domain-specific languages, but usually your use-case has to exactly match the intended use of the language, and OTDR data is just outside the norm of a statistical dataset (with the world's worst file format, with the statistics of interest being very different to population statistics, etc).