Hacker News: dewarrn1

New comment by dewarrn1 in "Anonymous request-token comparisons from Opus 4.6 and Opus 4.7"

dewarrn1 — Sat, 18 Apr 2026 17:59:42 +0000

I'm hopeful that new efficiencies in training (Deepseek et al.), the impressive performance of smaller models enhanced through distillation, and a glut of past-their-prime-but-functioning GPUs all converge make good-enough open/libre models cheap, ubiquitous, and less resource-intensive to train and run.

New comment by dewarrn1 in "Alzheimer's disease mortality among taxi and ambulance drivers (2024)"

dewarrn1 — Sun, 29 Mar 2026 14:52:08 +0000

The premise for this research question is related in part to work from Dr. Eleanor Maguire (who I just learned died last year at only 55) and her team on hippocampal anatomy and careers involving spatial navigation, including taxi drivers. A connection that may be of interest to HN readers: Demis Hassabis was one of Maguire's doctoral students, although he does not appear to have worked on the project most relevant to this study.

New comment by dewarrn1 in "Why I love FreeBSD"

dewarrn1 — Mon, 16 Mar 2026 17:54:10 +0000

Hail u/cperciva.

New comment by dewarrn1 in "Lego's 0.002mm specification and its implications for manufacturing (2025)"

dewarrn1 — Thu, 12 Mar 2026 03:47:23 +0000

If you're not interested in the new sets, the core product is readily available. Moreover, enjoy the fact that you can get that bucket of bricks for cheap partly because the expensive shiny high-margin SKUs provide a subsidy.

New comment by dewarrn1 in "MacBook Neo"

dewarrn1 — Wed, 04 Mar 2026 23:39:41 +0000

$499 through Apple Edu: https://www.apple.com/us-edu/shop/buy-mac/macbook-neo.

New comment by dewarrn1 in "I am directing the Department of War to designate Anthropic a supply-chain risk"

dewarrn1 — Sat, 28 Feb 2026 02:58:59 +0000

The real question: did he have Claude write this for him?

New comment by dewarrn1 in "NASA announces overhaul of Artemis program amid safety concerns, delays"

dewarrn1 — Sat, 28 Feb 2026 02:03:34 +0000

Had the same thought. NASA already cracked this nut with Apollo; if you’re gonna crack it again and differently, be real sure your solution is better.

New comment by dewarrn1 in "Hackers (1995) Animated Experience"

dewarrn1 — Sat, 07 Feb 2026 02:11:37 +0000

Tangential, but Darknet Diaries has been running a "hacker history" series to kick off 2026, starting with this one: https://darknetdiaries.com/episode/168/. The "Hackers" movie gets a few call-outs.

New comment by dewarrn1 in "Flux 2 Klein pure C inference"

dewarrn1 — Mon, 19 Jan 2026 04:08:14 +0000

Ah, the other AI: Actual Italian. (w/apologies to Davie504)

New comment by dewarrn1 in "Google is dead. Where do we go now?"

dewarrn1 — Tue, 30 Dec 2025 00:31:07 +0000

Presumably with apologies to Steinbeck: “Everyone was a temporarily embarrassed capitalist.”

New comment by dewarrn1 in "Pop_OS 24.04 LTS with COSMIC desktop environment"

dewarrn1 — Thu, 11 Dec 2025 19:58:01 +0000

Not your question, but 4xxx-series work seamlessly.

New comment by dewarrn1 in "AWS multiple services outage in us-east-1"

dewarrn1 — Mon, 20 Oct 2025 21:13:22 +0000

This is a fun example, but now you've got me wondering: has anyone checked on folks who might have been in an Amazon Go store during the outage?

New comment by dewarrn1 in "Qwen-Image: Crafting with native text rendering"

dewarrn1 — Mon, 04 Aug 2025 23:16:51 +0000

Slightly hyperbolic, gpt-image-1 is better on at least a couple of the text metrics.

New comment by dewarrn1 in "Claude Code is a slot machine"

dewarrn1 — Sun, 27 Jul 2025 18:55:46 +0000

Claude Code is^W^W^WEvery LLM is a slot machine.

New comment by dewarrn1 in "How to Firefox"

dewarrn1 — Tue, 22 Jul 2025 17:42:56 +0000

Yes, it's the Firefox version of Python's GIL.

New comment by dewarrn1 in "Preliminary report into Air India crash released"

dewarrn1 — Sun, 13 Jul 2025 02:01:13 +0000

A “dead airspace”, perhaps?

New comment by dewarrn1 in "Zuckerberg's Grand Vision: Most of Your Friends Will Be AI"

dewarrn1 — Wed, 07 May 2025 14:28:25 +0000

It's not my area of expertise, but I have seen other estimates that American adults, especially men, are likely to have and report numbers of friends such that the median is in the single digits.

New comment by dewarrn1 in "A.I. Is Getting More Powerful, but Its Hallucinations Are Getting Worse"

dewarrn1 — Mon, 05 May 2025 19:41:06 +0000

So, in reference to the "reasoning" models that the article references, is it possible that the increased error rate of those models vs. non-reasoning models is simply a function of the reasoning process introducing more tokens into context, and that because each such token may itself introduce wrong information, the risk of error is compounded? Or rather, generating more tokens with a fixed error rate must, on average, necessarily produce more errors?

A.I. Is Getting More Powerful, but Its Hallucinations Are Getting Worse

dewarrn1 — Mon, 05 May 2025 19:33:53 +0000

Article URL: https://www.nytimes.com/2025/05/05/technology/ai-hallucinations-chatgpt-google.html

Comments URL: https://news.ycombinator.com/item?id=43898618

Points: 35

# Comments: 25

New comment by dewarrn1 in "108B Pixel Scan of Johannes Vermeer's Girl with a Pearl Earring"

dewarrn1 — Thu, 01 May 2025 23:14:58 +0000

It sounds like a compelling film. Seeing now that it's produced by Penn & Teller's team, it makes me wonder about this fits into their "juggler vs. magician" dichotomy. The implication of that duo producing the film suggests that they might believe that Vermeer's reputation as an extraordinarily skilled artist ("juggler") might in fact be the result of Vermeer's use of a sophisticated apparatus that tricked patrons and viewers into thinking that he had extraordinary abilities, thus making Vermeer a master faker ("magician"). Or maybe they simply wanted to spur debate.