Hacker News: mhjkl

New comment by mhjkl in "Data Centers Now Consume 6% of US Electricity–and the Backlash Has Begun"

mhjkl — Sun, 24 May 2026 10:40:32 +0000

During the dot com boom they said data centers consumed 8% of US electricity. It was later proven to be a lie, created with fake studies by coal companies to trick the US into building more coal power plants using environmental activist outrage to implant the idea that the internet needs coal into the mainstream. Make of that what you will

New comment by mhjkl in "Is AI Profitable Yet?"

mhjkl — Sat, 23 May 2026 03:28:00 +0000

Reminds me of the “Has The Turing Test Been Passed” website. It says no, but if you read on they cite “The relatively minimal funding allocated to AI research” as one of the reasons AI hasn’t been achieved “yet”. Website stopped being updated before it became relevant, so you will never see it say “yes”, similarly to how the Loebner prize mysteriously vaporized when GPT-2 came out, just when winning it for real started becoming an interesting possibility

New comment by mhjkl in "LLM Policy for Rust Compiler"

mhjkl — Fri, 15 May 2026 21:42:16 +0000

Nim

New comment by mhjkl in "Spinel: Ruby AOT Native Compiler"

mhjkl — Fri, 24 Apr 2026 15:19:16 +0000

Antis: "If AI is so useful, where are the AI shovelwares? Where are the AI open source contributions? It's all hype" 6 months later: Matz used Claude and now Ruby runs 86 times faster after 1 month of work

At this point it's impossible to take antis seriously at all. Every claim is disproven simply by the passage of time. History will remember them just like the dot-com antis (that is, it won't)

New comment by mhjkl in "Shuttered startups are selling old Slack chats and emails to AI companies"

mhjkl — Sat, 18 Apr 2026 17:01:49 +0000

Reminds me of the Enron dataset

New comment by mhjkl in "A macOS bug that causes TCP networking to stop working after 49.7 days"

mhjkl — Thu, 16 Apr 2026 02:14:09 +0000

They didn’t need to wrap it because it’s modular arithmetic so the result after casting to int is the same regardless of wrapping behavior. 4294990000 after wrapping is 22704 and 4294960000 - 22704 = 4294937296 which is -30000 after uint to int cast

New comment by mhjkl in "A macOS bug that causes TCP networking to stop working after 49.7 days"

mhjkl — Thu, 16 Apr 2026 01:57:52 +0000

But the TSTMP_GEQ macro casts the difference to int, so any number above the signed integer limit (about 2 billion) becomes negative and the comparison returns false as they said

New comment by mhjkl in "I won't download your app. The web version is a-ok"

mhjkl — Mon, 06 Apr 2026 16:49:06 +0000

OP had no problem with pointers prior to trying C++. I think there is a case to be made that C(++) makes pointers unnecessarily confusing and there is no real disconnect between understanding pointers in theory and in practice otherwise

New comment by mhjkl in "My MacBook keyboard is broken and it's insanely expensive to fix"

mhjkl — Mon, 30 Mar 2026 18:07:38 +0000

You can remove and replace the switches too

New comment by mhjkl in "My MacBook keyboard is broken and it's insanely expensive to fix"

mhjkl — Mon, 30 Mar 2026 18:03:51 +0000

What exactly broke? I had this problem recently and thought the keyboard was done for, but turns out you can also replace the switches (not keycaps) and that solved the issue in my case. You have to be very careful though since it is very easy to break the switch or key cap if you remove it incorrectly (this happened to me while I was trying to clean the keyboard)

New comment by mhjkl in "Arm AGI CPU"

mhjkl — Sat, 28 Mar 2026 22:48:46 +0000

Couldn't he just subcontract it out to whoever he needs to make his chips? Doesn't need to do anything himself

New comment by mhjkl in "ICAO issued new power bank restriction on flight"

mhjkl — Sat, 28 Mar 2026 22:45:42 +0000

Afaik sodium batteries are much safer than li-ion and already in mass production. Unless Donut scales up really quick I think it will be more viable to just use sodium batteries for safety in the medium-term future

New comment by mhjkl in "You Do Not, in Fact, Have to Hand It to Them"

mhjkl — Sat, 28 Mar 2026 14:06:30 +0000

There's a big difference between things like Metaverse, social media and AI. The former are mainly entertainment/communication products that rely on network effects, so if people don't like them they're bust. AI is a capital good so it doesn't need to be popular as long as it does profitable work

New comment by mhjkl in "Arm AGI CPU"

mhjkl — Tue, 24 Mar 2026 18:39:58 +0000

What RAM? OpenAI booked the silicon wafers, they can print anything they want on them. I wouldn't call them "far behind" on hardware when OpenAI are actively buying Cerebras chips.

New comment by mhjkl in "Statement from Dario Amodei on our discussions with the Department of War"

mhjkl — Fri, 27 Feb 2026 02:34:03 +0000

Because as we all know the EU would never try using AI for mass surveillance /s

New comment by mhjkl in "Aqua: A CLI message tool for AI agents"

mhjkl — Mon, 23 Feb 2026 09:18:05 +0000

Might as well just encode the Unix timestamp at creation into proquints.

New comment by mhjkl in "Microsoft May Have Created the Slowest Windows in 25 Years with Windows 11"

mhjkl — Sat, 10 Jan 2026 20:20:57 +0000

The reason is that the web empire is just better at operating systems than Microsoft. If they just had less bad development tools for native UI this would not be a problem. Look at what Google does with Android, or Apple.

Open Source Historical LLM trained exclusively on 19th century text

mhjkl — Sat, 20 Dec 2025 08:44:31 +0000

Article URL: https://github.com/haykgrigo3/TimeCapsuleLLM

Comments URL: https://news.ycombinator.com/item?id=46334562

Points: 12

# Comments: 0

New comment by mhjkl in "Linux Sandboxes and Fil-C"

mhjkl — Sun, 14 Dec 2025 18:28:28 +0000

There’s no guarantee the toolchains will support WASM “preview” forever and make the bloat optional, and even if they do you could still end up in an ecosystem where it would be unviable. At some point you’re probably better off just compiling to RISCV and using an emulator library instead.

New comment by mhjkl in "Show HN: Stun LLMs with thousands of invisible Unicode characters"

mhjkl — Mon, 24 Nov 2025 22:01:35 +0000

Afaik most LLM datasets use FastText or something similar to detect the language of the data and if it's spam, and some additional small language models to detect if text is "educational" or desirable in some other way. Often text is filtered in instead of filtered out, so anything unusual like this probably won't pass the filter, you don't need to detect it explicitly.