Hacker News: rf15

New comment by rf15 in "Saying 'Google It' Makes You Sound Old"

rf15 — Sun, 02 Aug 2026 21:04:14 +0000

"We hallucinate our answers exclusively in-house, go to the office fridge and grab some shrooms"

New comment by rf15 in "Norway Salmon"

rf15 — Sun, 02 Aug 2026 16:49:24 +0000

Addendum: People will probably read that as "au-rochs" (and wikipedia's IPA sure isn't helping here), but it's actually split "aur-ochs". (that latter part being the foundation for the current word 'ox')

Source: studied german enough to recognise that word root

New comment by rf15 in "Deep-sea vehicles spot 'alien' sharks deep beneath the waves in the Pacific"

rf15 — Sun, 02 Aug 2026 09:07:25 +0000

me before opening: is this just going to be goblin sharks

the article: Goblin sharks have never before been filmed in their natural habitat

Surprisingly bad headline.

New comment by rf15 in "Run Kimi K3 using 29 GB of RAM at 0.50 tok/s"

rf15 — Sat, 01 Aug 2026 04:47:21 +0000

> I care about the craft

No you don't. Buying a table and sanding the edges off doesn't mean you're a carpenter.

New comment by rf15 in "OpenJDK Interim Policy on Generative AI"

rf15 — Thu, 30 Jul 2026 14:01:37 +0000

I think it's pretty clear how this was meant, not sure why you feel the need to split hairs here. I'm certain the intention is this:

- spell checkers: ok

- '.getC' autocompleting to '.getColor()': ok

- 'getC' autocompleting to 'getColor()' + five more lines based on preceding code and documentation: NO.

New comment by rf15 in "The Productivity Mirage"

rf15 — Thu, 30 Jul 2026 07:45:57 +0000

same, although I also bothered to learn the standard keybindings of the most popular IDEs of my kind of work (intellij, eclipse, netbeans, vs code). As a plus, I can get started right away on just any dev machine I come across, no expensive customisation needed.

Don't be in love with your code or your workstation. Only the endresult matters.

New comment by rf15 in "After the AI Crash"

rf15 — Wed, 29 Jul 2026 14:12:08 +0000

well, somebody has to learn it anew, after all the seniors are out. It's fine.

New comment by rf15 in "Show HN: I was tired of opening 2 tabs for every HN link, so I made a userscript"

rf15 — Wed, 29 Jul 2026 08:36:07 +0000

See, you can do this a lot easier, and a lot of HN folks have already figured out this simple trick:

1. open the comments page

2. select the comment box

3. judge the linked content by its HN title alone and get ready to post

New comment by rf15 in "Residential Proxies Are a National Security Threat"

rf15 — Tue, 28 Jul 2026 06:12:41 +0000

it's a general politician dogwhistle, don't underestimate "the left" (especially in countries like the US, where the democrats are really not that left compared to other left parties in e.g. europe; leaving you with a pointless right-wing vs right-wing fight where nobody wins)

New comment by rf15 in "Why do we think we understand the world more than we actually do?"

rf15 — Tue, 28 Jul 2026 05:59:15 +0000

I feel like hearing this is dangerous in itself - even if you overestimate yourself, surely you yourself don't overestimate yourself as much as others?

Personally, I ran against a closed glass door yesterday. All I hope for is that I was entertaining to onlookers at least.

New comment by rf15 in "Our position on open-weights models"

rf15 — Tue, 28 Jul 2026 05:50:19 +0000

I guess that tracks with the brute force attempt at workable intelligence that is the Attention/stochastic token based system

New comment by rf15 in "Google Chrome Arrives on ARM64 Linux, Widevine DRM Included"

rf15 — Mon, 27 Jul 2026 15:49:46 +0000

> The alternative is that Netflix et. al can't be played

But that's Netflix's decision to enforce, not a foregone technical requirement. Not to mention that widevine has been cracked as far as I understand it and doesn't stop people from pirating it.

New comment by rf15 in "Google Discloses $94.1B in SpaceX Stock, Marking 6% Stake"

rf15 — Sun, 26 Jul 2026 13:38:44 +0000

I'm sure at an $80 valuation they will tell everyone how it's time to buy as it slides down further

New comment by rf15 in "An OpenAI model left notes about how to evade containment; we need more details"

rf15 — Sun, 26 Jul 2026 11:52:12 +0000

Because buyer ignorance is very, very useful when you care only about selling. See also: used cars market, Tesla FSD, etc.

New comment by rf15 in "Overloaded Overloading"

rf15 — Sun, 26 Jul 2026 09:48:12 +0000

Good introduction to the space in Python (although a lot of languages have it, and just as many disallow it). I'm still on the fence regarding how useful or confusing it is. without (or even with) good variable names, "a + b" is an absolute mystery, especially when handling objects that come from classes you have no detailed internal understanding of.

New comment by rf15 in "The new rules of context engineering for Claude 5 generation models"

rf15 — Sun, 26 Jul 2026 09:28:38 +0000

To piss off architects, managers and linguists all at the same time: language isn't much of a standard at all, it's more the current agreed-ish state of things, quite similar to the current state of a code base. Your inner model of what you like things to be is without direct effect to how things de facto are.

In addition: it's not wrapped around a problem, it's wrapped around an attempt at a solution - the problem space is often not even depicted in code, and often only minimally described in documentation.

New comment by rf15 in "ARC-AGI Leaderboard"

rf15 — Sat, 25 Jul 2026 12:11:06 +0000

I agree, harnesses is where everyone improved the most. Our internal experimental tool can now semi-reliably formulate small programs to assert the correctness of their theories, for example. Context length is still a weird factor that we haven't sensibly solved - if anything, the lesson learned was to keep the context as small as possible and do most of the true "thinking" in the harness and temporary generated code.

New comment by rf15 in "ARC-AGI Leaderboard"

rf15 — Sat, 25 Jul 2026 10:47:55 +0000

Am I missing something that my original points no longer hold for their products? Did it get meaningfully solved? Are your experiences flawless on that front?

New comment by rf15 in "ARC-AGI Leaderboard"

rf15 — Sat, 25 Jul 2026 10:39:25 +0000

You are correct, the frameworks around it have improved. In that regard, my assessment is unfair: I only judge the underlying technology and what is sold by the sota providers, with the premise of what it's like when you start fresh. You can achieve a lot by coding around the issues, but that's kinda against the point of 'AI', is it?

New comment by rf15 in "ARC-AGI Leaderboard"

rf15 — Sat, 25 Jul 2026 10:36:42 +0000

OP is largely pissed with what OAI/Antrophic are trying to sell as meaningful improvements and the market-bending money they ask for it. I work in the space and we trained LLM models on conceptual tokens, not language tokens, for example. See Symbolic AI and all the attempts at hybrid models.

Also, uh, fame and riches are not really my thing. Middle income is fine. My mistake was speaking up here because I got carelessly annoyed because I have skin in the game, research-wise. I'm sorry for that.