bob is gonna correct you and tell you that you have to sketch out an idea before you start implementing, just not painfully detailed
i almost always start new projects with a day or two work on a rough spec draft for protocol or whatever the thing is meant to do, in fact, we don't even get funding as grantees without such a document, so it is actually first
The tests vs spec discussion might be interesting then.
You could write the specs in Gherkin and then it's also a test.
Josua was working on that, already.
I you want, you can also make them your public docs. But that‘s tricky. See an example here: https://rspec.info/features/3-13/rspec-core/subject/one-liner-syntax/
Ah, that's actually quite clever. Forces you to include all relevant information and structure it well.
