-
@ mleku
2025-05-16 07:22:16
i could draw a graph to illustrate my point, but i'll just try doing it with text:
each algorithm you understand intimately puts you at a certain level, and a given task has a minimum level you must be at in order to pass quickly to a new level by figuring out how to write a new algorithm.
when you rely on LLM to do the algorithms for you, you are down here, and wherever you can get to with the LLM, is a higher level than you can reach based on your knowledge.
if the target algorithm is at a greater height than the LLM has in its memory, you can't reach it, it's like walking up to a cliff without climbing gear and with sheer, smooth faces in front of you.
the person who has mastered algorithms to within reaching distance of the target's height, has all the tools required to scale this cliff and reach the target, but you, dear vibe coder, do not, and there is no way for you to get there without having learned how to use all the tools. the tools are freely available, and even, LLMs as search engines can give you the description of the function, the tool, but you can't use it without practising its use.
so until the LLM can reach that height for you, it might as well be on the other side of the universe, you can't get there by a shortcut, if the path to it has never been plotted.
then what is going to happen is that you just give up, because you didn't develop your toolkit in your brain, and you will never become the advanced coder you might have wanted to be.
if all you want to do is just build a new mousetrap with shiny chrome plating, that's fine, but you will never build a new kind of mousetrap.