
Tweets reference an ETH Zurich test of 100 students building software with AI assistance. One post states that computer science knowledge proved the strongest predictor of producing working code. Other developers share accounts of vibe coding projects that later stalled over logic questions the coder could not answer. Several contributors argue that understanding code remains essential regardless of AI tools, while one suggests the term itself should be replaced by design and development with agents. The conversation centers on limits observed when projects grow beyond initial prototypes.
"Software engineering is dead" until u have to build software that works... ETH Zurich tested 100 students in vibe coding and measured what actually predicts who builds software that works. Guess what... CS knowledge was the most important factor...
6.2K
8 posts, first seen 13h ago
Tweets discuss an ETH Zurich study on whether CS knowledge predicts success when students build software using AI agents.

Tweets reference an ETH Zurich test of 100 students building software with AI assistance. One post states that computer science knowledge proved the strongest predictor of producing working code. Other developers share accounts of vibe coding projects that later stalled over logic questions the coder could not answer. Several contributors argue that understanding code remains essential regardless of AI tools, while one suggests the term itself should be replaced by design and development with agents. The conversation centers on limits observed when projects grow beyond initial prototypes.
"Software engineering is dead" until u have to build software that works... ETH Zurich tested 100 students in vibe coding and measured what actually predicts who builds software that works. Guess what... CS knowledge was the most important factor...
Replies on vibe coding largely agreed good product ideas will matter more than technical skills, while others argued execution, selling, and marketing matter even more.
Based on 28 sentiment-bearing replies from 26 accounts across 2 conversations.
Replies on vibe coding largely agreed good product ideas will matter more than technical skills, while others argued execution, selling, and marketing matter even more.
Based on 28 sentiment-bearing replies from 26 accounts across 2 conversations.