Code is not the only element in the process generating GHG. The coder's way of life, the coder's tool, the resources needed should all be taken into account. What matters in the end is the global footprint of the code.

Writing quickly a very inefficient code that will run a handful of time is much better than spending 2 months full time optimizing it. And straightforward codes are also easier to debug. It may be a concept quite linked to Python, be premature optimization is where the crux of the problem is. Except we are now talking about an optimisation in processing power with an impact that is much wider, and thus with a more complex link.