[rkangel]

Machine learning is a slightly unusual case with powerful important libraries doing the work, with the language on top just being used for orchestration, data loading etc. If that's what you're doing, then you're right, the ML framework availability is far more important.

That's a still a relatively small corner of programming in general though. For most languages when talking about libraries we talk about the 'ecosystem' - what is the availability and quality of all the bits and pieces that we can build upon. It's a question of many small things, rather than one large thing.

Ecosystem differences are less absolute than 'has tensorflow', and so can be weighed up against other language advantages and disadvantages.

Web frameworks are an interesting case because you (usually) don't use them as simple libraries. The interaction with them tends to be complex enough that good frameworks are built around what the language is good at. If the language is right for the sort of programming you want to do, then the framework will express that.