Is that the takeaway? When they say, "I cloned the backend for Truco and gave Claude a long prompt explaining the rules of Escoba and asking it to refactor the code to implement it", that doesn't really make it sound like a good heads-up comparison from which we could then say, "the coding part was not the most significant part of the problem".

I mean, the entire article is problematic as proof of anything. For starters, they didn't go through a design process for a game at all, they copied existing games. Then there are all these weird technical rabbit holes they went down that really weren't anywhere near "simplest path to MVP".

I just don't think there is anything to glean from this article. Like most posts about individual experiences with AI, it's functionally equivalent to, "I had a weird dream last night".