HNHacker News
TopNewBestAskShowJobs

rzk

20,650 karma · joined June 13, 2018

submissionscomments
rzk··on A Maoist survival guide to the Iranian energy crisis
https://archive.ph/7ZuDd
rzk··on Looks like it is happening
Here is the data since the beginning of HN, where a new submitter is defined as a user whose account was created less than three months before the submission:

Stories: https://pastebin.com/Zc4jXvp4

Comments: https://pastebin.com/cFuczTWJ

The number of new submitters has indeed increased significantly since the beginning of 2026.

rzk··on Buteyko Method
It seems Buteyko method aims to increase carbon dioxide tolerance. This blog post has an interesting discussion about that:

Carbon Dioxide as a Stimulant for Respiratory Function —https://blog.supplysideliberal.com/post/2020/7/28/carbon-dio...

rzk··on Show HN: I built an interactive HN Simulator
I'm curious, how did you come up with the archetypes and moods? Were they generated automatically, or did you think them through?
rzk··on Israeli-founded app preloaded on Samsung phones is attracting controversy
> a wikipedia page on Western values

Here you go: https://en.wikipedia.org/wiki/Western_values

Since the wikipedia page also mentions individual liberty and rule of law, let me also link to this topic:

Arrest and detention of Palestinian minors by Israel — https://en.wikipedia.org/wiki/Arrest_and_detention_of_Palest...

rzk··on Israeli-founded app preloaded on Samsung phones is attracting controversy
> Israel as a country is aligned with Western values

Western values include, among other things, a commitment to equality and human rights, not apartheid[1] and genocide[2].

[1] https://en.wikipedia.org/wiki/Israeli_apartheid

[2] https://en.wikipedia.org/wiki/Gaza_genocide

rzk··on Learning to Model the World with Language
Here is an interesting related read (author is a researcher at OpenAI): Building AGI Using Language Models (2020) — https://bmk.sh/2020/08/17/Building-AGI-Using-Language-Models...

It makes the case that perfecting language prediction might be the path to AGI — because to predict language well, you have to understand the world it describes.

rzk··on Metabolic and cellular differences between sedentary and active individuals
There's a quote in that blog that I quite liked:

We’ve engineered movement out of our lives. We sit in chairs, stare at screens, and outsource physical effort to machines. Then we try to cram all our movement into 45-minute bursts a few times a week.

This is like eating only once a week and calling it a balanced diet. Most people are malnourished, not from lack of food, but from a lack of diverse, nutrient-dense movement.

rzk··on Metabolic and cellular differences between sedentary and active individuals
Here is a study that tries to answer that question: https://link.springer.com/article/10.1007/s40279-024-02120-2

A good summary: https://bannister.coach/mitochondria-and-exercise-how-differ...

Quoting the conclusion:

> In this systematic review and meta-regression covering ~ 50 years of research data, we demonstrate that the magnitude of change in mitochondrial content, capillarization, and VO2 max to exercise training is largely determined by the initial fitness level. The ability to adapt to exercise training is maintained throughout life irrespective of sex and presence of disease. Larger training volumes (higher training frequency per week and larger number of training weeks) and higher training intensities (per hour of training, SIT > HIT > ET) are associated with greater increases in mitochondrial content and VO2 max. Therefore, training load (volume x intensity) is a robust predictor of changes in mitochondrial content and VO2 max. Increases in capillarization occur primarily in the early stages of exercise training (< 4 weeks) with ET, HIT, and SIT equally enhancing capillaries per fiber, while ET is more effective in increasing capillary density (capillaries per mm²) due to less pronounced muscle fiber hypertrophy.

rzk··on Samourai Wallet Developer Sentenced to 5 Years for Unlicensed Money Transmitting
https://archive.ph/XMUYN
rzk··on FBI tries to unmask owner of archive.is
If only archive.is would share a dump of its archives through torrents the way Anna’s Archive does — it’d make it much more resilient.
rzk··on Poker Tournament for LLMs
See also: https://nof1.ai/

Six LLMs were given $10k each to trade in real markets autonomously using only numerical market data inputs and the same prompt/harness.

rzk··on The Logarithmic Time Perception Hypothesis
https://archive.ph/7Yskr
rzk··on The murky economics of the data-centre investment boom
https://archive.is/5CY26
rzk··on How Israeli actions caused famine in Gaza, visualized
> Unfortunately those security concerns turned out to be accurate

https://en.wikipedia.org/wiki/Self-fulfilling_prophecy

rzk··on F3: Open-source data file format for the future [pdf]
> Sometimes tech doesn’t win on the merits because of how entrenched existing stuff is and the high cost of switching.

Yes! Path dependence: https://en.wikipedia.org/wiki/Path_dependence

rzk··on Top UN legal investigators conclude Israel is guilty of genocide in Gaza
> Except no-one has ever offered any proof UN Watch is “tied to Israel”

Here is Hillel Neuer, Executive Director of UN Watch (https://x.com/HillelNeuer/status/1711844160804638980): “I just got off the phone with a third call with Prime Minister Netanyahu”

rzk··on Top UN legal investigators conclude Israel is guilty of genocide in Gaza
From https://en.wikipedia.org/wiki/UN_Watch:

> Agence France-Presse has described UN Watch as "a lobby group with strong ties to Israel" ... Primarily, UN Watch denounces what it views as anti-Israel sentiment at the UN and UN-sponsored events.

rzk··on Is documentation like pineapple on pizza?
Documentation is more like a pizza baking manual, so that if a new chef takes over, they can still make the same pizza.
rzk··on Rationale for a Large Text Compression Benchmark (2009)
Related: Building AGI Using Language Models (2020) - https://bmk.sh/2020/08/17/Building-AGI-Using-Language-Models...
rzk··on Aspiration Adaptation Theory (1998)
https://sci-hub.ru/https://www.sciencedirect.com/science/art...
rzk··on How long does it take to create a new habit? (2015)
But it doesn't necessarily take more than 21 days. According to the linked article, the average modeled time to plateau was 66 days, but the range was from 18 to 254 days.
rzk··on Experimental release of GrapheneOS for Pixel 9a
Thank you for all the hard work you do on GrapheneOS.
rzk··on Cell Phone OPSEC for Border Crossings
GrapheneOS offers good OPSEC against Cellebrite: https://discuss.grapheneos.org/d/14344-cellebrite-premium-ju...

This comment, at the end of the thread, is particularly interesting:

My Pixel 6 was confiscated by the German police after a political rally. I was recently able to pick it up again. From an inquiry with my lawyer, the following emerged: The authorities tried to read the device with both UFED4PC and Cellebrite Premium Touch. In addition, software from other forensic providers was used without success. The software did not succeed in breaking the system. The device was in BFU mode and had a 30-digit PIN. USB port was deactivated. As of March 2025, I can therefore say that it is not possible for Cellebrite to break a secured GrapheneOS.

rzk··on Heap-overflowing Llama.cpp to RCE
This is amazing—made even more impressive by the fact that the author is just 15 years old!

Also, it's nice to see this mentioned:

> For this 10k-word write-up, I spent around a month finishing up the main parts, and refining/editing it took an extra while. Writing this is indeed a painful process. I spent the entire day on the weekend and 4-5 hours during the rest of the week working on it for around two weeks.

It's the kind of behind-the-scenes effort that often goes unspoken.

rzk··on AI Blindspots – Blindspots in LLMs I've noticed while AI coding
> The LLM knows nothing about your requirements. When you ask it to do something without specifying all of the constraints, it will fill in all the blanks with the most probable answers from the universe of its training set. Maybe this is fine. But if you need something more custom, it’s up to you to actually tell the LLM about it.

Reminds of the saying:

“To replace programmers with AI, clients will have to accurately describe what they want.

We're safe.”

rzk··on Hyper-Systemizing
Related:

The Pattern Seekers: How Autism Drives Human Invention (Book review) — https://humanvarieties.org/2023/04/07/how-autism-drives-huma...

rzk··on The Evaporative Cooling Effect in Social Networks (2010)
Previous discussion:

The Evaporative Cooling Effect in Online Communities — https://news.ycombinator.com/item?id=1777665 (2010, 35 comments)

https://archive.ph/q8DlQ

rzk··on A Big Little Idea Called Legibility (2010)
Related: Book Review: Seeing Like A State — https://slatestarcodex.com/2017/03/16/book-review-seeing-lik...
rzk··on Ask HN: How do you control anger?
This might help: https://michaelsamsel.com/Content/Individuals/anger.html
← PreviousPage 2 of 3Next →