Quite frankly, word-based approaches miss the point of making a usable tool for the task, as well. Or in other words: it is justified to think of them as learni

Quite frankly, word-based approaches miss the point of making a usable tool for the task, as well. Or in other words: it is justified to think of them as learning tools while transitioning from natural language to the terse regex syntax. The terseness of regexes isn't the culprit though. It fits the task well - regexes could be quick to read and people can learn to remember the meanings of directives just like they learn everything - in repeated usage. There are just two necessary conditions for this learning to happen: 1. The syntax doesn't change in subtle ways every time you switch languages. 2. The syntax does not have nonsense context-dependent grammar rules, "just because".