Real

yes, it's always better to learn by doing. and someone with 5-8 years experience will always be a better coder and vibe coder than someone without experience and only knows vibe coding. however, i think the gap is becoming tighter: how much real hands-on experience do you really need unless you're building some top security high tech complex app for the Pentagon or Palantir or FAANG? for example, many people in CS still obsesses over Big O optimization or the various sort algorithms or security. ...but these days, since AI has already been trained on those as well -- has the gap between being aware of those things and knowing them like the back of your hand been greatly reduced? how deep should my knowledge of "system architecture" or "security" be if i already know that they exist and can direct LLMs to incorporate that into my code? do i need to have a PhD in those things? or would it be enough to know about them on a shallow level such that i can at least guide the LLMs to consider those when using AI? when was the last time the average coder created something groundbreaking or novel that it didn't already exist in the AI ecosystem? now, as far as employment goes -- sure, if you can't explain your code or why something broke when you speak to your coworker or manager -- that's going to look bad (and honestly -- you probably had a very low chance of actually making it past interviews). but assuming something like a solo developer...is his inability to explain his code or debug holding him back vs. someone who is able to walk the walk and talk the talk? i would love to have a real productive discussion on my thoughts.. TLDR: some may not be able to talk it, but they can still walk it. and that might be enough for majority of coders in the upcoming future.