The comparison between different languages gets tiring when it focuses on making a black-and-white statement like "Julia is better" or "Python is better" and "x is never going to overtake y". Yes, Python has many more libraries thanks to it being much older than Julia, same for R. But at the same time, Julia can be used for impressive work that R/Python struggle with and which only seem solvable in these languages because of large investments into certain packages by big companies.
So I find the fact that many hard problems can be solved very generically and performant with small libraries written in Base Julia much more interesting than countering that much larger and older Python packages with millions of developer hours poured into them are currently more feature-complete. Yes, they are, right now. Why wouldn't they be. But does what is being done in Julia with much fewer resources not point to an impressive ability of the language to facilitate such development?
While I generally agree with your argument, it's worth noting that the median Julia programmer is probably more invested in the language/ecosystem than the median R/Python programmer.
Back in the mid-90's Java was the new hotness, and it probably made problems that required 100+ lines of C easier, but it's not still full of above-average programmers, as any language/ecosystem that achieves success will inevitably regress to the mean.
That's true. I am of course biased in talking mostly to people on the Julia Slack etc. who enjoy the language a lot and do interesting things with it.
That's one of the reasons, though, why I never find the "how many people are using it" argument the most convincing when talking about the merits of a language. Because most people I've seen using R, Matlab and Python, at university or work for example, used it really superficially, and therefore wouldn't have any interesting things to say about it. Neither do they add anything interesting to the respective ecosystems. I don't think it's the first interest of a new language to get this type of user, although of course in the long term you want to build tools that are easy to be picked up and used by a wide audience, and number of users is some indicator of that.
So I find the fact that many hard problems can be solved very generically and performant with small libraries written in Base Julia much more interesting than countering that much larger and older Python packages with millions of developer hours poured into them are currently more feature-complete. Yes, they are, right now. Why wouldn't they be. But does what is being done in Julia with much fewer resources not point to an impressive ability of the language to facilitate such development?