There's a fascinating analogy between coding in python using hyper-optimized vectorized libraries (numpy, cupy, torch...), and living in Singapore.

In python, whatever you're not vectorizing takes the longest, even if it's utterly trivial. Literally picking merkle branches out of a pre-built tree is taking me longer than an entire fast fourier transform.

In Singapore, whatever you get from international brands (as opposed to eg. hawker centers) is the most expensive, even if it's utterly trivial. I've literally had meals where the starbucks green tea I had right after a hawker center meal cost more than the meal.