Yesterday I got in touch with some people using SAS again. I last
used SAS three years ago, and many things already are happily
forgotten. Python has a community which is open, friendly and
growing. Python is easy to learn and adopted as a first programming
language in many undergraduate programs. Python serves as a lingua
franca of data science (artificial intelligence, machine learning,
choose your name) world. SAS is proprietary, with the University
Edition made available about five years ago. Before that, you needed
to buy an expensive license only to download their software. Or you
could organise a pilot project, of course, if your company was large
enough. Now there are some freely available courses of SAS Base and
even some co-programs with universities. But I can't imagine anyone
learning SAS for fun. There is no community after all. There are
certified professionals who excel in this exact technology and are
interested in preserving its usage by their employer at any cost.
Yes, SAS works well in what it is designed for. They have excellent
documentation, and their solutions are easy to use in many
scenarios. Sometimes, SAS just works, no complaints. But they feel
the need to be modern and more libre since they can't afford to be
gratis.
For example, they developed a SAS Kernel for JupyterLab and a
`saspy` package which enables writing Python code which is then
translated into SAS code and executed on SAS servers. And I confirm,
that `saspy` is written in quite a readable manner. They even use
type annotations! Whether I will use it or not is more a question of
general architecture than personal taste. But the proprietary
software must die, that's for sure. Future is for free and
open-source.
