HNHacker News
TopNewBestAskShowJobs

scrlk

8,635 karma · joined March 24, 2022

submissionscomments
scrlk··on EU commissioners shut down air conditioning for employees, leave theirs on
As a side note, it's nearly impossible to buy a dual-hose portable AC in the UK and Europe. For whatever reason, the market has converged on inefficient single-hose portable ACs.
scrlk··on DSpark: Speculative decoding accelerates LLM inference [pdf]
See "exabox" from George Hotz: https://tinycorp.myshopify.com/products/exabox-preorder
scrlk··on Previewing GPT‑5.6 Sol: a next-generation model
> Sol, Terra and Luna

So the next naming scheme might be FTX, Madoff and Enron? :^)

scrlk··on U.S. government will decide who gets to use GPT-5.6
The common denominator between "GPT-2 is too dangerous to release" and Anthropic is Dario.
scrlk··on IBM debuts sub-1 nanometer chip technology
Cymer builds the EUV light source, but the biggest enabler for High NA EUV is using anamorphic optics (ie asymmetric horizontal and vertical magnification) from Zeiss: https://www.asml.com/en/news/stories/2024/5-things-high-na-e...
scrlk··on Apple to skip high-end M6 Mac chips in favor of AI-focused M7 line
Thanks for catching that - unfortunately I can't edit the submission URL, but I've emailed hn@ycombinator.com to see if the mods can fix it.

EDIT: gift link if paywalled (archive.is capture is truncated): https://www.bloomberg.com/news/articles/2026-06-25/apple-to-...

scrlk··on GLM-5.2 is the new leading open weights model on Artificial Analysis
I've seen a few articles from providers talking about KV cache quantisation, but it's not something they explicitly point out like they do with weights.

So you could end up paying more for unquantised weights, only to get silently hit with a quantised KV cache...

scrlk··on GLM-5.2 is the new leading open weights model on Artificial Analysis
IME, unquantised -> FP8 is pretty much lossless. What matters more is having an unquantized KV cache - using an FP8 KV cache can result in a significant drop in quality.
scrlk··on UK police officer under criminal investigation over alleged use of AI
https://archive.ph/BIOej
scrlk··on Anthropic's model naming, extrapolated
And Sony as well - AI-LLM1000XM5 Mark II.
scrlk··on I Hate (Most) Keyboard 'Fn' Keys
The rationale behind the Fn-Ctrl layout used on ThinkPads until very recently:

> The Fn key was originally placed by the ThinkPad designers in the lower left hand corner to make the key easier to locate when using the keystroke combinations. There was a rationale. This is especially handy for turning on the ThinkLight in the dark. Aim for the two extreme corners.

https://web.archive.org/web/20111115202457/http://www.lenovo...

IMO, putting Ctrl in the bottom left still isn't great for ergonomics: Ctrl should be where Caps Lock is, in line with the home row. This was the case on the original PC keyboard (IBM Model F) before it was moved to the bottom left in the Model M.

One of the only mass-produced keyboards that has Ctrl in this position today is the Apple Japanese keyboard: https://www.apple.com/shop/product/mxcl3j/a/magic-keyboard-u...

scrlk··on Microsoft builds MacBook Pro rival with NVIDIA-powered Surface Laptop Ultra
A bug affecting M5 MBAs & MBPs made it impossible to connect to a WiFi network on the activation screen. The workaround was to use an Ethernet dongle: https://x.com/TheBobPony/status/2033958150936703363
scrlk··on Uber blows through its AI budget in 1 quarter
> after incentivizing employees to adopt the technology through an internal leaderboard ranking teams by total AI tool usage.

When you pay by the dead cobra, don't be surprised when people start breeding cobras.

scrlk··on Classic 7 is a Windows 10 LTSC mod to look 1:1 to Windows 7
> I also recall this 3D shell where your desktop was basically like an first-person shooter, where there would be a literal desk with files that you could click on, a media wall that would display your photos and so on.

Was it Task Gallery from Microsoft Research?

https://www.microsoft.com/en-us/research/publication/the-tas...

https://dl.acm.org/doi/10.1145/332040.332482

scrlk··on Improving ICU handovers by learning from Scuderia Ferrari F1 team
Next up: learning effective radio communication from Ferrari F1 race engineers ;)
scrlk··on CATL claims 6-minute charge and 1,500km range for new electric vehicle batteries
https://archive.ph/D5IJd
scrlk··on An Interview with Pat Gelsinger
Sadly nothing about why Pat was forced out of Intel. Panther Lake is quite a good CPU and 14A looks like it'll be a competitive process. IMO, it vindicates the decisions that Pat made a few years ago and wasn't able to see through to completion.

"Designing microprocessors is like playing Russian roulette. You put a gun to your head, pull the trigger, and find out four years later if you blew your brains out." (Robert Palmer, former DEC CEO)

scrlk··on Intel 486 CPU announced April 10, 1989
More specifically, it was the end of Dennard scaling [0] that killed off the growth in clock speeds in the mid-2000s.

[0] https://en.wikipedia.org/wiki/Dennard_scaling

scrlk··on Teardown of unreleased LG Rollable shows why rollable phones aren't a thing
LG never really recovered after releasing several phones with bootlooping issues between 2015-16: https://en.wikipedia.org/wiki/LG_smartphone_bootloop_issues

Plus their software support was poor, even for the era.

scrlk··on Google releases Gemma 4 open models
I transposed the table so that it's readable on mobile devices.

I should have mentioned that the Qwen 3.5 benchmarks were from the Qwen3.5-122B-A10B model card (which includes GPT-5-mini and GPT-OSS-120B); apologies for not including the smaller Qwen 3.5 models.

scrlk··on Google releases Gemma 4 open models
Comparison of Gemma 4 vs. Qwen 3.5 benchmarks, consolidated from their respective Hugging Face model cards:

    | Model          | MMLUP | GPQA  | LCB   | ELO  | TAU2  | MMMLU | HLE-n | HLE-t |
    |----------------|-------|-------|-------|------|-------|-------|-------|-------|
    | G4 31B         | 85.2% | 84.3% | 80.0% | 2150 | 76.9% | 88.4% | 19.5% | 26.5% |
    | G4 26B A4B     | 82.6% | 82.3% | 77.1% | 1718 | 68.2% | 86.3% |  8.7% | 17.2% |
    | G4 E4B         | 69.4% | 58.6% | 52.0% |  940 | 42.2% | 76.6% |   -   |   -   |
    | G4 E2B         | 60.0% | 43.4% | 44.0% |  633 | 24.5% | 67.4% |   -   |   -   |
    | G3 27B no-T    | 67.6% | 42.4% | 29.1% |  110 | 16.2% | 70.7% |   -   |   -   |
    | GPT-5-mini     | 83.7% | 82.8% | 80.5% | 2160 | 69.8% | 86.2% | 19.4% | 35.8% |
    | GPT-OSS-120B   | 80.8% | 80.1% | 82.7% | 2157 |  --   | 78.2% | 14.9% | 19.0% |
    | Q3-235B-A22B   | 84.4% | 81.1% | 75.1% | 2146 | 58.5% | 83.4% | 18.2% |  --   |
    | Q3.5-122B-A10B | 86.7% | 86.6% | 78.9% | 2100 | 79.5% | 86.7% | 25.3% | 47.5% |
    | Q3.5-27B       | 86.1% | 85.5% | 80.7% | 1899 | 79.0% | 85.9% | 24.3% | 48.5% |
    | Q3.5-35B-A3B   | 85.3% | 84.2% | 74.6% | 2028 | 81.2% | 85.2% | 22.4% | 47.4% |

    MMLUP: MMLU-Pro
    GPQA: GPQA Diamond
    LCB: LiveCodeBench v6
    ELO: Codeforces ELO
    TAU2: TAU2-Bench
    MMMLU: MMMLU
    HLE-n: Humanity's Last Exam (no tools / CoT)
    HLE-t: Humanity's Last Exam (with search / tool)
    no-T: no think
scrlk··on First Western Digital, now Sony: The tech giant suspends SD card sales
TI started production at their SM1 fab back in December 2025, which focuses on 28 nm to 130 nm.
scrlk··on Excel incorrectly assumes that the year 1900 is a leap year
An interesting read related to this bug from Joel Spolsky - My First BillG Review: https://www.joelonsoftware.com/2006/06/16/my-first-billg-rev...
scrlk··on Crypto investor turns $50M into $36,000 in one botched move
Plus, having $50M sitting in a mobile hot wallet.
scrlk··on Britain is ejecting hereditary nobles from Parliament after 700 years
As Walter Bagehot wrote in The English Constitution: "An ancient and ever-altering constitution is like an old man who still wears with attached fondness clothes in the fashion of his youth: what you see of him is the same; what you do not see is wholly altered."

Absent ideological capture, it is perhaps one of the best forms of government ever created due to its pragmatic nature and its Lindyness is proof.

scrlk··on Britain is ejecting hereditary nobles from Parliament after 700 years
The irony is that, on a technicality, the hereditary peers were the only members of the Lords who had to win an election to get their seats.

> Under the reforms of the House of Lords Act 1999, the majority of hereditary peers lost the right to sit as members of the House of Lords, the upper house of the Parliament of the United Kingdom. Section 2 of the Act, however, provides an exception from this general exclusion of membership for up to 92 hereditary peers: 90 to be elected by the House, as well as the holders of two royal offices, the Earl Marshal and the Lord Great Chamberlain, who sit as ex officio members.

https://en.wikipedia.org/wiki/List_of_excepted_hereditary_pe...

scrlk··on MacBook Pro with M5 Pro and M5 Max
Incremental. If you want a major change, the M6 MBP is rumoured to launch towards the end of the year. It's expected to bring a new design and an OLED touchscreen.

https://www.bloomberg.com/news/articles/2026-02-24/apple-s-t... (https://archive.ph/qT3QV)

scrlk··on Gemini 3.1 Pro
IME, they definitely nerf models. gemini-2.5-pro-exp-03-25 through AI Studio was amazing at release and steadily degraded. The quality started tanking around the time they hid CoT.
scrlk··on Pentagon used Claude in Maduro Venezuela raid
> Claude was used to do things their guidelines prohibit (facilitate violence, develop weapons or conduct surveillance)

There's Claude Gov models for this:

> U.S. national security customers may choose to use our AI systems for a wide range of applications from strategic planning and operational support to intelligence analysis and threat assessment. Claude Gov models deliver enhanced performance for critical government needs and specialized tasks. This includes:

> * Improved handling of classified materials, as the models refuse less when engaging with classified information

> * Greater understanding of documents and information within the intelligence and defense contexts

> * Enhanced proficiency in languages and dialects critical to national security operations

> * Improved understanding and interpretation of complex cybersecurity data for intelligence analysis

https://www.anthropic.com/news/claude-gov-models-for-u-s-nat...

scrlk··on Claude Code is suddenly everywhere inside Microsoft
Cf. -2000 Lines Of Code:

https://www.folklore.org/Negative_2000_Lines_Of_Code.html

← PreviousPage 3 of 24Next →