People have and did make that claim about Python. The creator of this very website wrote exactly that, in 2004: http://www.paulgraham.com/pypar.html It is one of the most famous and popular things he has ever written.

It turns out that people write posts about things they like. Sometimes they like things for good reasons, sometimes they like things for bad reasons. But that has no direct bearing on how good a tool is.

I really don't relate to this. When I was in high school and college and for the first few years of my career, Python was not "a very good tool for that particular job". A number of the Ruby programmers I first knew and respected were using Ruby for doing computational / scientific computing with more iteration and velocity than they could achieve with Fortran, C, or C++ (the common scientific languages at the time). There was no reason back then to think that Python as a language was any better suited for this purpose than Ruby or Perl or any other language really. (And what we now call data scientists were already mostly doing this with R.)

It was indeed through a significant amount of writing and advocacy and persuasion that Python - through numpy and scipy and pandas - won out here.

For a time it looked like it was going to fade, as it was too slow and all the "big" tooling was built on the jvm instead. But it got faster and people made more interfaces to underlying faster components, and people did more writing and advocacy and persuasion, and now it's mostly python still, with a bunch of native or jvm code under the hood sometimes.

But none of this happened just through python being clearly better for this niche. People always write and discuss and advocate for the tools they like. It's normal. It's same as it ever was.

Well said. Now replace "Rust" with "Julia" in your first sentence for a moment of enlightenment for everyone.

Fair enough.

Until now I've convinced that people don't code Rust for higher performance, it's a cult and people use that to be looked cool. Rust are not some kind of black wand that can zap your 100k lines of C/C++ code to be 100 lines of code with the same speed. If you want high performance then it's not just the language which is fast, it's the way you implement your idea.

People are always talking shit about things to make people think they are smart, like people around me think they are Larry Wall, bashed Java for being slow, but they don't know that a lot of important, performant thing are coded in Java.

If you want raw performance, use C/C++. If you want simplicity, then use Python. If you want reliability and safety, then use Java/Go.