You can find lots of benchmarks of Julia if you want that. However as you should know, benchmarks tend to be microbenchmarks as it is really hard to make controlled comparisons of language implementations of larger projects.

I think you will agree with me that making a controlled experiment with writing whole climate models in Fortran and Julia would be extremely hard, expensive and impractical. As software developers we don't have the luxury of objective scientific analysis for issues like this.

Most aspects of different programming languages cannot be discussed in an easily quantifiable way, but rather has to be dealt with in a qualitative way.

I approach these sorts of topics with a Baysian mindset. You don't know anything in life with absolute certainty. However we can make numerous observations and assign probabilities. If you find numerous stories of people switching to Julia and experiencing major speed boosts, then there are many possible explanations/hypothesis for that:

- It was all a fluke. A random occurence.

- It is a conspiracy. The Julia creators are paying developers to give fake testemonies.

- It is a biased press or media, only showing good results for Julia and cherry picking bad for Fortran, Python etc.

- Julia actually is faster and better, which is why these stories keep popping up.

If you don't know what to believe, then the best approach is to read more testemonies and then give it a try yourself. If you keep having good results then would you attribute that to simply luck every time?