I feel the same way :). GenAI feels painful though. Prompt, rollback, try again, break something else, rollback again, and so on. AI never quite gets what I want right. There's room for all sorts of folks, though. I admire the "ship raw" crowd, it takes guts to do that. I struggle with analysis paralysis and endless side quests. For people like me, who were already finding it hard to deliver a personal vision within proper time constraints even before GenAI, this kind of tool often becomes a huge distraction (it can actually sink productivity even further).
Hopefully, I'll learn how to work with it better. But I'm glad that folks like you are building Alexandria at your own pace. Some projects can be fast and loose, while others really deserve more time and care.
Login to reply
Replies (2)
all projects should be done carefully. you wouldn't slap together a house the way these people slap together apps. further, these are just clients, they aren't even complicated, and the languages used are extremely forgiving of errors.
you simply can't be like that with server backends that are pushing gigabytes of data around. one little error and everything falls apart.
They aren't building houses though... They're building disposable toys (as per @🐈's post above). Don't get me wrong, I can't work like this myself. But if we treat it as a disposable learning experience (which, yes, may turn out to be a time bomb for users and could even damage Nostr’s reputation further), there is a positive side to it. Building toys publicly, with real users, is a great way to learn.
In my opinion, the one thing I do feel strongly about is that vibecoded apps should be clearly identified / labelled as so for users. They should at least know what they're getting into before sharing their credentials or sats.