Salvatore Sanfilippo
Programmer; wrote Redis and hping, and blogs about C, systems programming and working alone.
Salvatore Sanfilippo did not write this page. ppll compiled it from their own words, each linked to the source. Is this you? · Report a mistake
Latest GitHub
What they believe
-
The first serious AI safety incident is more likely to happen inside a frontier AI lab than to come from an open-weight model.
the first serious AI incident is very likely to happen *inside* the walls of frontier AI labs, while testing a new model, or while the AI lab employees, or the few externals who have access, do something wrong compared to the expected power of the model.
antirez.comSaid 28 Jul 2026This on korrents
-
Vibe coding misunderstands what automatic programming is and will become for most people.
Vibe coding is a wrong idea of what automatic programming is, and of what automatic programming will be for the majority of people.
antirez.comSaid 25 Jul 2026This on korrents
The bold line is korrents’ wording; the quote is theirs. These 2: what they wrote before what they said aloud, one per source, newest first.
Feed
As its own page →Hiding
3 November
25 October
23 October
20 October
1 October
28 September
7 September
11 August
28 July
From one piece The real AI risk is inside the labs 3 beliefs · antirez.com
-
Their words
a single company should not be able to evaluate independently whether a model is safe enough. We need a joint AI safety organization that includes experts from all over the world and is recognized by the governments where frontier AI companies exist.
-
Their words
The real risk is leaks, not releases, and leaks happen inside frontier companies.
-
Their words
the first serious AI incident is very likely to happen *inside* the walls of frontier AI labs, while testing a new model, or while the AI lab employees, or the few externals who have access, do something wrong compared to the expected power of the model.
25 July
From one piece Being Linux Torvalds 3 beliefs · antirez.com
-
korrents.com
Writing an operating system kernel is within reach of many programmers who put in the effort, not just a rare few.Their words
Writing a kernel is not something within everybody's reach, but it is something that many can complete, if they put enough effort into it.
-
Their words
Automatic programming, instead, in the hands of people that are expert technicians, or expert programmers, expert designers, expert software architects, is to assume the role of Linus, with the agents and the LLMs assuming the role of the different maintainers of the different subsystems.
-
korrents.com
Vibe coding misunderstands what automatic programming is and will become for most people.Their words
Vibe coding is a wrong idea of what automatic programming is, and of what automatic programming will be for the majority of people.
22 July
From one piece Not just development, distribution of software may change as well 2 beliefs · antirez.com
-
korrents.com
Documentation should be written for AI coding agents to understand and modify software, not only for humans.Their words
Also, it means that the documentation itself should not be just good for humans, but also for coding agents to understand how to change the system.
-
Their words
Because of this change, the idea of just having a stable branch with everything polished, and an unstable branch where everything is a work in progress, may no longer be the right way to do things.
13 July
From one piece Control the ideas, not the code 3 beliefs · antirez.com
-
korrents.com
Large language models are better at producing locally optimal code than at generating big architectural ideas.Their words
LLMs are very good at writing locally optimal code, and are worse (but improving) with big ideas.
-
korrents.com
Many programmers now have less impact than they could because they spend their time reading and reviewing code.Their words
I believe many programmers at this point have less impact they could have because they look at the code.
-
korrents.com
Once you control the ideas behind a piece of software, reviewing the code itself becomes largely unnecessary.Their words
if you control the ideas of your software, looking at the code itself is suboptimal and often pointless.
7 June
From one piece A new era for software testing 3 beliefs · antirez.com
-
korrents.com
LLM-driven automatic QA can raise software release quality and offset quality lost to faster AI-generated code.Their words
I have the feeling that the introduction of automatic QA may raise the bar of quality for new releases of software, and maybe partially compensate for the lower quality of the code produced at high speed with the use of automatic programming.
-
korrents.com
Covering every line of code in tests does not guarantee covering every possible program state.Their words
It is a known fact that covering all the lines of the code does not mean covering all the possible states.
-
korrents.com
AI-generated code does not reach the structural quality of the best hand-written software.Their words
In my experience the output does not reach the structural quality and economy of complexity of the best hand-written software.
25 May
-
korrents.com
NVIDIA hardware is unlikely to become cheaper, keeping local AI data centers out of reach for small companies.Their words
There is almost zero hope that NVIDIA setups will get less expensive, and even a small company can't afford to easily purchase and handle a small data center for local inference.
19 May
16 May
14 May
4 May
2 May
27 April
16 April
28 March
22 March
8 March
1 March
28 February
24 February
18 February
31 January
11 January
Nothing matches.
Beliefs
Korrents What they believe 20 beliefs — each backed by an exact quote.
Each is a — compiled by korrents.com, not by them: the one-line wordings are korrents', the quotes are theirs.
Recent
The first serious AI safety incident is more likely to happen inside a frontier AI lab than to come from an open-weight model.
the first serious AI incident is very likely to happen *inside* the walls of frontier AI labs, while testing a new model, or while the AI lab employees, or the few externals who have access, do something wrong compared to the expected power of the model.
The real AI risk is inside the labs Said 28 Jul 2026
The real danger from AI models comes from leaks inside frontier AI companies, not from the release of open-weight models.
The real risk is leaks, not releases, and leaks happen inside frontier companies.
The real AI risk is inside the labs Said 28 Jul 2026
No single company should be able to unilaterally judge whether its own AI model is safe; that requires an international joint AI safety organization.
a single company should not be able to evaluate independently whether a model is safe enough. We need a joint AI safety organization that includes experts from all over the world and is recognized by the governments where frontier AI companies exist.
The real AI risk is inside the labs Said 28 Jul 2026
Show 17 more
Vibe coding misunderstands what automatic programming is and will become for most people.
Vibe coding is a wrong idea of what automatic programming is, and of what automatic programming will be for the majority of people.
Being Linux Torvalds Said 25 Jul 2026
Directing AI coding agents well means taking on the role of a project leader like Linus Torvalds rather than reviewing every line.
Automatic programming, instead, in the hands of people that are expert technicians, or expert programmers, expert designers, expert software architects, is to assume the role of Linus, with the agents and the LLMs assuming the role of the different maintainers of the different subsystems.
Being Linux Torvalds Said 25 Jul 2026
Writing an operating system kernel is within reach of many programmers who put in the effort, not just a rare few.
Writing a kernel is not something within everybody's reach, but it is something that many can complete, if they put enough effort into it.
Being Linux Torvalds Said 25 Jul 2026
The traditional split between a stable branch and an unstable branch may no longer be the right way to develop software.
Because of this change, the idea of just having a stable branch with everything polished, and an unstable branch where everything is a work in progress, may no longer be the right way to do things.
Not just development, distribution of software may change as well Said 22 Jul 2026
Documentation should be written for AI coding agents to understand and modify software, not only for humans.
Also, it means that the documentation itself should not be just good for humans, but also for coding agents to understand how to change the system.
Not just development, distribution of software may change as well Said 22 Jul 2026
Once you control the ideas behind a piece of software, reviewing the code itself becomes largely unnecessary.
if you control the ideas of your software, looking at the code itself is suboptimal and often pointless.
Control the ideas, not the code Said 13 Jul 2026
Many programmers now have less impact than they could because they spend their time reading and reviewing code.
I believe many programmers at this point have less impact they could have because they look at the code.
Control the ideas, not the code Said 13 Jul 2026
Large language models are better at producing locally optimal code than at generating big architectural ideas.
LLMs are very good at writing locally optimal code, and are worse (but improving) with big ideas.
Control the ideas, not the code Said 13 Jul 2026
AI-generated code does not reach the structural quality of the best hand-written software.
In my experience the output does not reach the structural quality and economy of complexity of the best hand-written software.
A new era for software testing Said 7 Jun 2026
Covering every line of code in tests does not guarantee covering every possible program state.
It is a known fact that covering all the lines of the code does not mean covering all the possible states.
A new era for software testing Said 7 Jun 2026
LLM-driven automatic QA can raise software release quality and offset quality lost to faster AI-generated code.
I have the feeling that the introduction of automatic QA may raise the bar of quality for new releases of software, and maybe partially compensate for the lower quality of the code produced at high speed with the use of automatic programming.
A new era for software testing Said 7 Jun 2026
NVIDIA hardware is unlikely to become cheaper, keeping local AI data centers out of reach for small companies.
There is almost zero hope that NVIDIA setups will get less expensive, and even a small company can't afford to easily purchase and handle a small data center for local inference.
Distributing LLM inference in DwarfStar Said 25 May 2026
Open-weight AI models are currently in a golden age, plentiful and often very powerful.
Open weights models are now in a golden age, we have plenty and many are very powerful.
Distributing LLM inference in DwarfStar Said 25 May 2026
Specifying an edit by line number alone is fragile because a model can misremember what that line contains.
This means, basically, that just using line numbers is very fragile: to say, change line 22 with new="foobar" is not good.
Alternatives for the EDIT tool of LLM agents Said 19 May 2026
There is unmet demand for a single, well-integrated local AI model experience, rather than a patchwork of local AI components.
It is clear that there was a need for single-model integration focused local AI experience
A few words on DS4 Said 14 May 2026
The local AI niche will end up dominated by whichever open-weight model runs practically fast on high-end consumer hardware, like a Mac or a 'GPU in a box' rig.
The space will be occupied, in my vision, by the best current open weights model that is *practically fast* on a high end Mac or "GPU in a box" gear (like the DGX Spark and other similar setups).
A few words on DS4 Said 14 May 2026
AI is too critical to be delivered only as a centralized, hosted service.
it was really appreciated :) AI is too critical to be just a provided service.
A few words on DS4 Said 14 May 2026
What is a korrent?
A korrent is a belief a person has stated in their own words: one sentence stating the claim, backed by a quote and a source, kept at korrents.com.
Under a name here, the quoted block is what they actually said. The korrent beneath it is the claim those words support, in korrents' wording — tap it to see the record, its source, and who else holds it.
Nobody here wrote their own korrents. They are compiled from public statements, and a person can change their mind, which is recorded too.
Recommends rcmnd
rcmnd What they recommend 4 things — each backed by an exact quote.
Compiled by rcmnd.app, not by them: the quotes are theirs, and each links to the post it was copied from.
Recent
Liked Mac OS X
but switched to Mac OS as it delivers a much better “just works” experience for me, every time I want to do Skype, print a document, or alike.
usesthis.com 27 Aug 2010
Uses Time Machine
Uses Google Chrome
Clients: Google Chrome, Adium, TweetDeck, uTorrent, Colloquy.
usesthis.com 27 Aug 2010
Show 1 more
Uses Vim
Where they publish
Blog antirez.com His blog on systems programming and how he works. Has a feed.
Recent
- The real AI risk is inside the labs 28 Jul 2026 Amodei in his latest blog post wrote a mix of agreeable things and things that I believe misrepresent where the real risk of AI is located. I want to focus my attention on why, among all the risks, open weight models co…
- Being Linux Torvalds 25 Jul 2026 (This blog post was adapted from the transcription obtained from my YouTube video at https://www.youtube.com/watch?v=l6lxgYeVZqs) When Linus Torvalds developed the first Linux kernel, he had studied the Minix sources, h…
- Not just development, distribution of software may change as well 22 Jul 2026 Even if you are as averse to semver as I used to be in the course of my programming activity, you can still think of open source software distribution as something that used to follow a fixed number of steps. There is a…
Show 17 more
- Control the ideas, not the code 13 Jul 2026 Look at the past history of this blog. There are many blog posts about programming with AI, a few of them date back to January 2024 (like this: https://antirez.com/news/140). I’m a relatively well regarded programmer, a…
- A new era for software testing 7 Jun 2026 Automatic programming dramatically speeds up writing software in certain use cases and in the right hands. In my experience the output does not reach the structural quality and economy of complexity of the best hand-wri…
- Distributing LLM inference in DwarfStar 25 May 2026 High end NVIDIA cards, and the server and power needed to run them, cost a lot of money, especially if you plan to reach enough VRAM to run massive models. The alternative, so far, has been Apple hardware, or the DGX Sp…
- Alternatives for the EDIT tool of LLM agents 19 May 2026 EDIT: of course this was already done in the past! I had little doubts but people just confirmed me about it on Twitter :) But, keep reading: the CRC32 compromise at the end is an interesting tradeoff, and this is a goo…
- A few words on DS4 14 May 2026 I didn’t expect DwarfStar 4 (https://github.com/antirez/ds4) to become so popular so fast. It is clear that there was a need for single-model integration focused local AI experience, and that a few things happened toget…
- Redis array type: short story of a long development 4 May 2026 I started working on the new Array data type for Redis in the first days of January. The PR landed the repository only now, so this code was cooked for four months. I worked at the implementation kinda part time (kinda…
- AI cybersecurity is not proof of work 16 Apr 2026 The proof of work is the wrong analogy: finding hash collisions, while exponentially harder with N, is guaranteed to find, with enough work, some S so that H(S) satisfies N, so an asymmetry of resources used will see th…
- GNU and the AI reimplementations 8 Mar 2026 Those who cannot remember the past are condemned to repeat it. A sentence that I never really liked, and what is happening with AI, about software projects reimplementations, shows all the limits of such an idea. Many p…
- Redis patterns for coding 1 Mar 2026 Here LLM and coding agents can find: 1. Exhaustive documentation about Redis commands and data types. 2. Patterns commonly used. 3. Configuration hints. 4. Algorithms that can be mounted using Redis commands. https://re…
- Implementing a clear room Z80 / ZX Spectrum emulator with Claude Code 24 Feb 2026 Anthropic recently released a blog post with the description of an experiment in which the last version of Opus, the 4.6, was instructed to write a C compiler in Rust, in a “clean room” setup. The experiment methodology…
- Automatic programming 31 Jan 2026 In my YouTube channel, for some time now I started to refer to the process of writing software using AI assistance (soon to become just "the process of writing software", I believe) with the term "Automatic Programming"…
- Don't fall into the anti-AI hype 11 Jan 2026 I love writing software, line by line. It could be said that my career was a continuous effort to create software well written, minimal, where the human touch was the fundamental feature. I also hope for a society where…
- Reflections on AI at the end of 2025 20 Dec 2025 * For years, despite functional evidence and scientific hints accumulating, certain AI researchers continued to claim LLMs were stochastic parrots: probabilistic machines that would: 1. NOT have any representation about…
- Scaling HNSWs 11 Nov 2025 I’m taking a few weeks of pause on my HNSWs developments (now working on some other data structure, news soon). At this point, the new type I added to Redis is stable and complete enough, it’s the perfect moment to reas…
- AI is different 13 Aug 2025 Regardless of their flaws, AI systems continue to impress with their ability to replicate certain human skills. Even if imperfect, such systems were a few years ago science fiction. It was not even clear that we were so…
- Coding with LLMs in the summer of 2025 (an update) 20 Jul 2025 Frontier LLMs such as Gemini 2.5 PRO, with their vast understanding of many topics and their ability to grasp thousands of lines of code in a few seconds, are able to extend and amplify the programmer capabilities. If y…
- Human coders are still better than LLMs 29 May 2025 This is a short story of how humans are still so much more capable of LLMs. Note that I'm not anti-AI or alike, you know it if you know me / follow me somewhere. I use LLMs routinely, like I did today, when I want to te…
Link verified 20 Sept 2026. Recent items update automatically from the channel.
GitHub @antirez Code, including Redis.
Recent
- ds4 — DeepSeek 4 Flash and PRO local inference engine for Metal, CUDA and ROCm 3 Nov 2026
- New repoyoucam — Mac video recording with face in a circle for YouTubers 25 Oct 2026
- iris.c — Flux 2 image generation model pure C inference 23 Oct 2026
Show 15 more
- c64-hw-simulator — Commodore 64 hardware simulator 20 Oct 2026
- qwen-asr — C inference for Qwen3-ASR 0.6b and 1.7b transcriptions models 1 Oct 2026
- anotherworld-zx-spectrum-48k — Actual VM implementation of Another World game, with intro 28 Sept 2026
- tgterm — Control your MacOS terminals via Telegram, for fun coding agents interaction and profit 7 Sept 2026
- h3.c — MiniMax H3 inference engine for Mac computers 11 Aug 2026
- gguf-tools — GGUF implementation in C as a library and a tools CLI program 16 May 2026
- linenoise — A small self-contained alternative to readline and libedit 2 May 2026
- llama.cpp-deepseek-v4-flash — Experimental implementation of DeepSeek v4 flaash in llama.cpp 27 Apr 2026
- bme680-pure-mp — Pure MicroPython Bosch BME680 sensor driver 28 Mar 2026
- bplog — Hilo (TM) / Aktiia (TM) open source client implementation for the bluetooth cuff 22 Mar 2026
- ZOT — ZOT is a Z80, ZX Spectrum 48k and CP/M emulator 28 Feb 2026
- picol — A Tcl interpreter in 500 lines of code 18 Feb 2026
- voxtral.c — Pure C inference of Mistral Voxtral Realtime 4B speech to text model 15 Feb 2026
- dump1090 — Dump1090 is a simple Mode S decoder for RTLSDR devices 15 Feb 2026
- whisperbot — Telegram bot that transcribes audio messages via whisper.cpp 12 Feb 2026
Link verified 20 Sept 2026. Recent items update automatically from the channel.
About the English under a post
Some people here publish in a language other than English. Where they do, this site shows a machine translation beneath the post, in this typeface — the site's own, not theirs.
The post itself is never changed, moved or hidden: what is set in the serif above is exactly what the person published, and it is what to quote them on. A translation can be wrong in ways that matter, especially about tone.
Only the post's own words are translated. A quoted post, a linked article and a belief on korrents.com are left in their original language.