Steve Yegge on Writing a Quarter Million Lines of Code in Six Days Using Llms
Posted4 months agoActive4 months ago
twitter.comTechstory
skepticalmixed
Debate
60/100
LlmsCode GenerationSoftware Development
Key topics
Llms
Code Generation
Software Development
Steve Yegge claims to have written 250k lines of code in 6 days using LLMs, sparking debate about the quality and implications of AI-generated code.
Snapshot generated from the HN discussion
Discussion Activity
Light discussionFirst comment
26m
Peak period
4
0-1h
Avg / period
2
Key moments
- 01Story posted
Sep 11, 2025 at 1:27 AM EDT
4 months ago
Step 01 - 02First comment
Sep 11, 2025 at 1:53 AM EDT
26m after posting
Step 02 - 03Peak activity
4 comments in 0-1h
Hottest window of the conversation
Step 03 - 04Latest activity
Sep 11, 2025 at 3:33 AM EDT
4 months ago
Step 04
Generating AI Summary...
Analyzing up to 500 comments to identify key contributors and discussion patterns
ID: 45208113Type: storyLast synced: 11/17/2025, 6:13:10 PM
Want the full context?
Jump to the original sources
Read the primary article or dive into the live Hacker News thread when you're ready.
What even is that?
I guess we all wait for the output, which is the only true proof. Right now, it's a loon getting excited about llms competing with each other to produce what is quite likely way too much code, within which lurks way too many bugs. Any seasoned programmer will tell you less code is better, both for performance and for reliability.
I'd say I can't wait to see the output, but at this rate, it will likely be 1m loc when done, and if I was going to review a million lines of code, I'd be looking at the Linux kernel - a carefully crafted bit of software with actual review processes in place, that's used in millions upon millions of devices. I can pretty-much guarantee that whatever slop is finally served up, it won't be anywhere near the same league.