Top all-time · #6424
Lennart Finke
Personal blog · fi-le.net · on HN as fi-leaiprogrammingphilosophy
Lennart Finke writes The Fiefdom of Files, a personal blog at fi-le.net about ai, programming and philosophy. Since 2025, 12 different Hacker News readers have linked it 46 times, most often “What GPT-OSS leaks about OpenAI's training data”. It is still writing: the last post was on 2026-10-05, and you can follow it by feed.
Too few sampled comments (5) to say how often HN recommends it.?
Why HN links it
HN readers link Lennart Finke's posts on LLM quirks and experiments, such as the seahorse emoji token glitch, gpt-oss bad tokens, and the Baba is Eval test of models playing video games.
Summarised by Claude from a sample of 5 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
By Lennart Finke.
Most-linked posts?
- What GPT-OSS leaks about OpenAI's training data4 people · 4 mentions · first linked 2025-09-20
- How Much of AI Labs' Research Is Safety?3 people · 3 mentions · first linked 2026-01-13
- Baba Is Eval2 people · 2 mentions · first linked 2025-07-03
- Measuring Is Not Enough Anymore2 people · 2 mentions · first linked 2026-06-25
- More Targets of the OpenAI Agent Swarm2 people · 2 mentions · first linked 2026-09-04
What Hacker News says?
Oh, and more stuff yet, they get around! https://fi-le.net/vanderbilt/ (via https://news.ycombinator.com/item?id=49569146 )
Comment linking /vanderbilt/
on “Discovery of a new OpenAI agent message board”Submitted What GPT-OSS leaks about OpenAI's training data as a story · 348 points
HN discussionIt looks like like the link shortener vanderbi.lt, operated by Vanderbilt University, was compromised in some form, too: https://fi-le.net/vanderbilt
Comment linking /vanderbilt
on “Discovery of a new OpenAI agent message board”Submitted Operating Margins as a story · 289 points
HN discussiongpt-oss has similar bad tokens. https://fi-le.net/oss/
Comment linking /oss/
on “Why do LLMs freak out over the seahorse emoji?”Submitted Baba Is Eval as a story · 254 points
HN discussionNot to out myself as old, but I learned to program before ChatGPT before stack overflow and before Google. There are some here that are even older. There were paper books with indexes. Indexes! That way of life is gone for me. I've got a smartphone and I doomscroll to my detriment. What's new and fascinating to me is the models themselves. https://fi-le.net/oss/ currently trending here is the tip of a whole new area…
Comment linking /oss/
on “The deadline isn't when AI outsmarts us – it's when we stop using our own minds”Submitted A Map of Python as a story · 175 points
HN discussionright click -> inspect HTML looks pretty handwritten (lack of crazy css classes and simple structure) <link rel="stylesheet" href=" https://fi-le.net/css/tufte.css "> They are using tufte.css https://edwardtufte.github.io/tufte-css/
Comment linking /css/tufte.css
on “Operating Margins”Submitted Why Does Claude Speak Byzantine Music Notation? as a story · 139 points
HN discussionWe're doing a successor to this, working hard and going public in month or so, hopefully. But HN gets a preview of course: https://huggingface.co/datasets/lennart-finke/SimpleStories And here's a more interactive explorer: https://fi-le.net/simplestories
Comment linking /simplestories
on “TinyStories: How Small Can Language Models Be and Still Speak Coherent English? (2023)”Submitted Lessons from That 1834 Landscape Gardening Guidebook as a story · 136 points
HN discussionNice! I remember "Baba is Eval" ( https://fi-le.net/baba/ ), released 11 months ago, back when Claude Opus 4 was the strongest model. Back then, I was surprised how poor was it even at the first level. I am happy to see an another approach - and indeed, with much stronger results.
Comment linking /baba/
on “Why are large language models so terrible at video games?”Submitted The Whimsical Investor as a story · 97 points
HN discussionSubmitted Where Is GPT in the Chomsky Hierarchy? as a story · 55 points
HN discussionSubmitted Towards the Blank Search Bar as a story · 22 points
HN discussionSubmitted More Targets of the OpenAI Agent Swarm as a story · 20 points
HN discussionSubmitted That Squiggly, Treacherous Line as a story · 8 points
HN discussionSubmitted Who Gets the German Federal Cross of Merit? as a story · 4 points
HN discussionSubmitted Coding Intelligence Asymptotics as a story · 4 points
HN discussionSubmitted Abundance in Europe as a story · 3 points
HN discussionSubmitted Casuistic Alignment as a story · 3 points
HN discussionSubmitted Applied Keynesian Psychology as a story · 3 points
HN discussionSubmitted What GPT-OSS leaks about OpenAI's training data as a story · 3 points
HN discussionSubmitted Measuring Is Not Enough Anymore as a story · 2 points
HN discussionSubmitted Measuring Is Not Enough Anymore as a story · 2 points
HN discussionSubmitted The Harvard Library Passport as a story · 2 points
HN discussionSubmitted How Much of AI Labs' Research Is Safety? as a story · 2 points
HN discussionSubmitted Hilbert's Power as a story · 2 points
HN discussionSubmitted Handling the Shoes as a story · 2 points
HN discussion
