You are assuming researchers are saintly figures dwelling in a vacuum who don't need to constantly prove to their department head or promotion evaluation committee of their worth. That is not the case. The walled gardens are desirable for some because their social functions are not easily replaceable.

One way to decouple the evaluation of scientific output from the walled garden is simply to stop using them as a gate-keeper in making hiring and research grant distribution decisions. But apart from the constant lip service, there is no momentum in doing anything concrete about this in academia.

Who do you think is supervising the high school student? Where does the idea for the project come from? Where does the money supporting the high school intern' experiments come from?

The second part of the sentence seems fictional. I don't think there is any support of that from a basic understanding of carbonate chemistry.

A statistician/data scientist wrangling data and making plots would not have cared whether loading a CSV file takes one second or one microsecond, because they may only do it a handful of times for a project.

A data engineer has different requirements and expectations. They may need to implement an operational component that process CSV files repeatedly for billions of time a day.

If your use case is the latter, then pandas is probably not for you.

> fossil-fuel-sourced CO2 and CH4 are steadily warming the planet

No climate scientist denies this. It is the trend happening since the industrial revolution. But say, if you have three months or three years of data, you are not gonna be able to fit a statistically robust trend. At best, you can do some Bayesian attribution of the likelihood that a climate event is attributable to the long-term climate change effect.

> Many of the supposed discoveries of periodic oscillations in the North Atlantic and the Pacific simply don't hold up over time, and as with ENSO, have zero predictive power over what phase/amplitude of the next 'period of the oscillation' will be.

You are jumping into conclusions too soon. The fact that a system has periodicity does not necessarily mean that it is perfectly deterministic. You may not be able to predict when the next El Niño is gonna happen, but you can say with confidence how many El Niño events you will likely see in a 100-year time span.

> These phenomena are likely mostly chaotic in nature

Again, being chaotic does not mean no predictability at all. The atmosphere and the ocean do follow the basic laws governing fluid dynamics. How much predictability you have very much depends on the spatial and temporal scale you are looking at. Being able to predict one event and being able to predict a statistical distribution or a power spectrum are two different things.

I know it is always appealing to tell a story with a grand unifying narrative. But sound research must prop it up with empirical evidence. Could there be limits to growth? Probably. But a highly simplistic model not informed by appropriate data or economic understanding is not the way to tell such limits.

Also, would you expect JAX acceleration to work well with other types of discretization, such as spectral methods?