The bar is then probably: "can be ugly if it's too much work to do it by the state of the art, but robust enough not to (unexpectedly) break on you at the worst time"
Maybe it's a matter of handling nominal cases, shipping/releasing and only then handling the cases more at the edge (just) after, but with nominal cases out of the way with someone waiting for results and putting pressure while you are working on your stuff. You win the occasion on working on your stuff will less stress, which might as well end up in a more robust result.
For some problems, maybe it's acceptable to handle them when they occur. It could have allowed someone who rely on a feature to use it earlier and be more productive despite this problem, instead of waiting for the perfect solution. It may allow more communication, and therefore more insight, which may help producing better results too.
> If something needs to be improved it will. Or it won't, but then maybe it's not your problem anymore.
I had my situation in mind, where I would often find myself wanting something to be improved but unfortunately I'm not the one deciding what gets done. And if I were, I still would not make the decision of doing so because of priorities.
Maybe it's a matter of handling nominal cases, shipping/releasing and only then handling the cases more at the edge (just) after, but with nominal cases out of the way with someone waiting for results and putting pressure while you are working on your stuff. You win the occasion on working on your stuff will less stress, which might as well end up in a more robust result.
For some problems, maybe it's acceptable to handle them when they occur. It could have allowed someone who rely on a feature to use it earlier and be more productive despite this problem, instead of waiting for the perfect solution. It may allow more communication, and therefore more insight, which may help producing better results too.