Vibe coding: the end of coding as we know it

Key Takeaways

the code you write in the next 12 months may be some of the last code you ever write. Now, I think that's a big bold statement, but I think it's directionally correct. Like the amount of code people are going to be writing is just gonna, you know, continue to go down or the inverse of that is you just have more people coding and like, you know, true software engineers are just going to, you know, move to a different part of the stack, but like, you know, they're still going to be doing that same amount of coding. Like, it could easily go either way or both ways at once. Yeah. I mean, I think that's definitely an overstatement. Like I don't think people will not write code again after this year after the next 12 months. I mean that's like the Silicon Valley CEO thing. I think that like the amount of incremental change we're going to see with like each new model release is going to be smaller and smaller. And so when it comes to actually building some of these like complex softwares or complex systems, I think there's going to be a line in which it starts to get a bit blurry where like below this line is going to be very easy to create and then above this line is actually going to be still hard to create. maybe easier than it was before, but you're still going to need extra time and resources to go and like make that thing. Yeah. I I think um I think what's deceiving a lot of people like you know I try to build stuff on weekends and like test out new softwares and all that and it can handle a small codebase really well like vibe coding you know you you can get to like 80% of a pretty complex project super quickly. the last 20% bug fixes, adding tests, like it is painful. And so I think you're you're probably right that the floor is going to continue to move up slowly though. Um, but it gives people this feeling of like, hey, we're going to do a build vers and we're going to choose to build it ourselves because look, like engineer X put together a prototype that's like half of what we want in like an hour. So imagine what we can do in the next 5 hours, right? But then you realize there's actually it's like a lo like a logarithmic slope and that you make a lot of progress and then it's just painful the more progress you try to make and at a certain point there's just this like there's this plane of complexity that the AI cannot handle and that's actually you know the whole reason that solutions exist.