Top all-time · #12894
Diogo Almeida
Personal blog · completeskeptic.comaiprogramming
Diogo Almeida writes Complete Skeptic, a personal blog at completeskeptic.com about ai and programming. Since 2026, 6 different Hacker News readers have linked it 10 times, most often “(KV) Cache Rules Everything Around Me”. It is still writing: the last post was on 2026-09-11, and you can follow it by feed.
Too few sampled comments (3) to say how often HN recommends it.?
Why HN links it
Linked posts cover the 'jagged frontier' of LLM benchmark performance and why data may matter more than architecture ('the bitterest lesson'). The comments are self-promotional and do not evaluate the writing.
Summarised by Claude from a sample of 3 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
Sane takes in an insane world.
Most-linked posts?
- (KV) Cache Rules Everything Around Me3 people · 3 mentions · first linked 2026-09-09
- Scaling Laws, Honestly3 people · 3 mentions · first linked 2026-07-04
- Diogo Almeida: Lies, Damned Lies, and Benchmarks2 people · 2 mentions · first linked 2026-09-15
What Hacker News says?
1. I am extremely on the same page 2. I do think that subconscious is not only much smarter than we give it credit for, but also much more robust than the "jagged frontier" of current LLMs (shilling my blog post on that jaggedness: https://www.completeskeptic.com/p/lies-damned-lies-and-bench... )
Comment linking /p/lies-damned-lies-and-benchmarks
on “Introducing System One Models and Jev”Submitted (KV) Cache Rules Everything Around Me as a story · 3 points
HN discussionarchitecture is close to the chest for now, but we have talked about writing a paper I don't want to shill my blog too much, but I will say data is probably far most interesting than architecture: https://www.completeskeptic.com/p/the-bitterest-lesson
Comment linking /p/the-bitterest-lesson
on “Introducing System One Models and Jev”Submitted (KV) Cache Rules Everything Around Me as a story · 3 points
HN discussionLetting a model manage its own context is very bitter lesson-pilled Biggest challenge is you will get a much lower cache hit rate if you frequently edit the agent's context/prefix, so this can not be implemented efficiently via e.g. the Anthropic API. This ^ can be solved in principle but likely requires modifications to the transformer architecture and definitely to serving infrastructure See related: "KV Cache Rul…
Comment linking /p/kv-cache-rules-everything-around
on “Context Language Models”Submitted Scaling Laws, Honestly as a story · 3 points
HN discussionSubmitted Scaling Laws, Honestly as a story · 2 points
HN discussionSubmitted Diogo Almeida: Lies, Damned Lies, and Benchmarks as a story · 1 points
HN discussionSubmitted Is it even possible for the Chinese Labs to distill US models? as a story · 1 points
HN discussionSubmitted Scaling Laws, Honestly as a story · 1 points
HN discussion