Top all-time · #3264
Alex L. Zhang
Personal blog · alexzhang13.github.io/blogaiprogramming
Alex L. Zhang's personal blog at alexzhang13.github.io covers ai and programming. Since 2024, 23 different Hacker News readers have linked it 29 times, most often “Recursive Language Models (RLMs)”. It is still writing: the last post was on 2026-09-26, and you can follow it by feed.
Alex Zhang's Website.
Most-linked posts?
- Recursive Language Models (RLMs)7 people · 9 mentions · first linked 2025-10-15
- The Mismanaged Geniuses Hypothesis5 people · 5 mentions · first linked 2026-04-10
- A Meticulous Guide to Advances in Deep Learning Efficiency over the Years3 people · 5 mentions · first linked 2024-10-31
- Speculative Programmatic Tool Calling3 people · 3 mentions · first linked 2026-08-24
- Highlights of NeurIPS 2023 from Reading All 3584 Abstracts2 people · 2 mentions · first linked 2024-01-21
- Language Model "Shape"2 people · 2 mentions · first linked 2026-09-27
What Hacker News says?
I think this is an unsolved problem. The most interesting thing I saw here is the Recursive Language Models paper. https://arxiv.org/abs/2512.24601 There's also a great write up here by the author: https://alexzhang13.github.io/blog/2025/rlm/
Comment linking /blog/2025/rlm/
on “Advancing the price-performance frontier with GPT‑5.6”Submitted Recursive Language Models (RLMs) as a story · 135 points
HN discussionNot exactly. See: https://alexzhang13.github.io/blog/2026/mgh/
Comment linking /blog/2026/mgh/
on “DeepSeek V4 Pro 0813”Submitted The Annotated Kolmogorov-Arnold Network (Kan) as a story · 36 points
HN discussionA write up on RLM (Recursive Language Models) by one of the authors of the RLM paper: https://alexzhang13.github.io/blog/2025/rlm/
Comment linking /blog/2025/rlm/
on “Prime Agent: A self-improving RLM agent”Submitted A Meticulous Guide to Advances in Deep Learning Efficiency over the Years as a story · 4 points
HN discussionThe whole thing was borne out of wanting to keep costs low! My favoured approach (last time I was doing this) is using only a tiny sliding context window based on message pairs, rather than tokens, and only for the agents that need it. Amending prose style, for example, shouldn't need context beyond the message it's working on, and then its system prompt. For the models that require context, I personally found combi…
Comment linking /blog/2025/rlm/
on “Kimi K3 is now live”Submitted Speculative Programmatic Tool Calling as a story · 3 points
HN discussionhere's a more readable version: https://alexzhang13.github.io/blog/2025/rlm/
Comment linking /blog/2025/rlm/
on “Recursive Language Models”Submitted Language model harnesses are compositional generalizers as a story · 3 points
HN discussionI had some ideas for extending the lem editor (emacs in common lisp) the other day and I am barely literate in Lisp. So I had Claude Code do it. Fully awesome. No problems. A few paren issues, buit it seemed to not really struggle really. It produced working code. Was also really good at analyzing the lem codebase as well. I even had it write an agentic coding tool in Common Lisp using the RLM ideas: https://alexzha…
Comment linking /blog/2025/rlm/
on “SBCL Fibers – Lightweight Cooperative Threads”Submitted A Meticulous Guide to Advances in Deep Learning Efficiency as a story · 3 points
HN discussionReasoning is recursive - you cannot isolate where is should be symbolic and where it should be llm based (fuzzy/neural). This is the idea that started https://github.com/zby/llm-do - there is also RLM: https://alexzhang13.github.io/blog/2025/rlm/ RLM is simpler - but my approach also have some advantages.
Comment linking /blog/2025/rlm/
on “Towards a science of scaling agent systems: When and why agent systems work”Submitted A Meticulous Guide to Advances in Deep Learning Efficiency over the Years as a story · 3 points
HN discussionLoopy is a filesystem sandbox whose entire state lives in a single string. It comes with a bash‑style shell so you or your agent can navigate, search, and manipulate a tree‑structured knowledge base with composable commands. The idea is to simulate a file system to build and navigate a knowledge tree with POSIX-style commands that agents are already familiar with. Why? When an agent processes information, it needs s…
Comment linking /blog/2025/rlm/
on “Loopy: An in-memory, zero-side-effect filesystem API over a single Python string”Submitted Highlights of NeurIPS 2023 from Reading All 3584 Abstracts as a story · 3 points
HN discussionWhen an agent processes information, it needs somewhere to put it - somewhere it can search, reorganize, and grow organically. For any type of knowledge base like agent memories, product taxonomies, etc the challenge was enabling recursive interaction without a pile of specialized tools (search, create, delete, etc.) that are added to context. Recursive Language Models (RLMs) introduced the idea of putting the entir…
Comment linking /blog/2025/rlm/
on “Loopy: A filesystem API over a single Python string”Submitted Highlights of NeurIPS 2023 from Reading All 3584 Abstracts as a story · 3 points
HN discussionSubmitted The Mismanaged Geniuses Hypothesis as a story · 2 points
HN discussionSubmitted Speculative Programmatic Tool Calling as a story · 2 points
HN discussionSubmitted Language Model "Shape" – Alex L. Zhang as a story · 1 points
HN discussionSubmitted Language Model "Shape" as a story · 1 points
HN discussionSubmitted Speculative Programmatic Tool Calling as a story · 1 points
HN discussionSubmitted LLMs are underutilized due to sub optimal management as a story · 1 points
HN discussionSubmitted The Mismanaged Geniuses Hypothesis as a story · 1 points
HN discussionSubmitted The Mismanaged Geniuses Hypothesis as a story · 1 points
HN discussionSubmitted Language Models will be Scaffolds as a story · 1 points
HN discussionSubmitted A Meticulous Guide to Advances in Deep Learning Efficiency over the Years as a story · 1 points
HN discussionSubmitted A Meticulous Guide to Advances in Deep Learning Efficiency over the Years (2024) as a story · 1 points
HN discussion
