Hackaday Podcast · Hands-On Engineering Podcasts · November 2026
In the Supercon wrapup, Elliot says many badge hacks were vibe coded or heavily LLM-assisted. He weighs the cost to readability against how fast people could build things.
…engineering a human. You're reverse engineering something else that wrote the code. And so it's been a really wild ride. And sometimes I'd ask people, you know, hey, that's really cool. How'd you do that? And they're like, yeah, I don't know. JLM did it for me. Yes. And that is unsettling. And I
had that exact experience several times. They're like, well, I don't know. I don't know how it did it.
And as a consumer of the code and as a human being who thinks of code not just as instructions for a machine, but a way to convey ideas about how to do things to other humans, I find this to be a real problem with the LLM code. On the other hand, in terms of how quickly it let people iterate on ideas and how fast it let them spin things up and play around with them and find corners of the code base they wouldn't have found otherwise and et cetera, it was absolutely astounding. So, huh? I mean, I think I've just summarized where we all are right now. And I don't know that there's anything particularly deep here. I just had never been confronted with this before. And now I've seen it. And now I know what all y'all software people have been talking about the last year. Yeah,
it's hard to be mad, I guess, for lack of a better word, because it did enable such incredible stuff on a ridiculous time scale that was more obvious to us because we had just been banging our head against the wall for the preceding 48 hours trying to even get to where we were, you know? So to see…
readability? What's that?
Yeah. Like, and there was things that were like, I just could not understand what would possess the thing, but nothing possesses it, right? Like, that's the answer. Like, there's that, you're trying to find logic in something that doesn't operate on the same parameters as you do. It's not, its goal…