I feel like this pattern is a great example of how mathematicians, scientists, and technologists should focus their attentions on things that are economical and 'good enough' rather than fetishizishing ever-increasing levels of precision.

While, say, calculating fundamental constants to the nth degree of precision is fun and intellectually satisfying in many respect, the marginal utility of each additional digit of precision asymptotically approaches zero so it's really a self-indulgent luxury. But this fetishizing of precision also has drawbacks; it misdirects resources and effort into improvements whose benefits fall far short of the costs.

You wouldn't use a laser pointer to explore a room. Of course you could; but mapping a topological space using LIDAR is only possible because we have computers to do the vast amount of number crunching required, and it's only very recently that we've gone from very sparse representations to ones of sufficiently high resolution to give us representations that are immediately recognizable as familiar spaces.

I will take fast proximity over slow exactitude any day of the week; I want models that get me close enough with very few parameters rather than more perfect ones that require a thick book's worth of specifications to articulate.