What is it about Cucumber that is so attractive to early in career engineers (myself included)? Is it the naive belief that others, non-engineers even, will actually want to read or write your tests? It's great in theory, but like many ideologies that attract young people, political or otherwise, its promise of egalitarian test authorship rarely holds up in practice. My experience is that it ends up as an extra abstraction layer to maintain.
I spent a lot of time tryngito figure them out mid carreer before I realized that high priced consultants made a lot of money selling training me on it to management. The same salesmanship worked on me. Perhaps in part because they were also teaching unit testing which did turn out useful - once I got rid of the bad ideas they taught since they didn't know what worked in the real world but had a lot of money to make by sounding confident when asked questions.
I find it sad that cucumber managed to pour so much cold water on the idea of readable spectests. It's like if we gave up on programming because COBOL was bad.