Hacker News: newswasboring

New comment by newswasboring in "Advertise in ChatGPT"

newswasboring — Wed, 22 Jul 2026 06:51:30 +0000

OpenAI also started as a non profit. I don't trust any of these people and I propose no one should.

New comment by newswasboring in "Qwen-Image-3.0: Rich Content, Authentic Details, Deep Knowledge"

newswasboring — Tue, 21 Jul 2026 14:35:58 +0000

I've been struggling with this question myself. But isn't this a model training/use problem (a.k.a skill issue )? Isn't there a way to make these models be faithful to how people will actually look?

New comment by newswasboring in "Claude Code uses Bun written in Rust now"

newswasboring — Sun, 19 Jul 2026 18:01:18 +0000

Anything that can be written in JavaScript will be written in JavaScript. That's just how it is it seems.

New comment by newswasboring in "The Kimi K3 Moment"

newswasboring — Sun, 19 Jul 2026 17:52:21 +0000

I'm just going by the materials you provided.

New comment by newswasboring in "The Kimi K3 Moment"

newswasboring — Sun, 19 Jul 2026 14:04:04 +0000

The china talk page points to an article about crypto currency stolen credit cards. If I believed every random X account i would have to believe too many false things. Not really convinced my guy.

New comment by newswasboring in "The Kimi K3 Moment"

newswasboring — Sun, 19 Jul 2026 13:46:01 +0000

> They often use stolen credit cards, educational discounts, or resold compute credits to acquire them at essentially zero cost. They're not always paying customers.

Yeah but where are you getting this from? I've seen this claim many places but only as pure speculation. No proof, just bold faced assertions.

New comment by newswasboring in "The Kimi K3 Moment"

newswasboring — Sun, 19 Jul 2026 13:26:49 +0000

Why should other people be responsible to make their business model work? They hold like all the money, if they can't make it work please shut down. Companies like OpenAI have already broken the public trust by breaking their non profit promises. They don't deserve any politeness at this point.

New comment by newswasboring in "The Kimi K3 Moment"

newswasboring — Sun, 19 Jul 2026 13:18:59 +0000

Where are you getting this stolen credit card thing? That's so random.

New comment by newswasboring in "The state of open source AI"

newswasboring — Sat, 18 Jul 2026 10:28:09 +0000

It wouldn't matter how much money you have if everyone around you is stressed. We are social beings, on some level most people cannot enjoy in isolation for too long.

New comment by newswasboring in "Waldi: A quiet place to write, and to be read"

newswasboring — Sat, 18 Jul 2026 10:26:47 +0000

How does one get an invite code for this?

New comment by newswasboring in "The state of open source AI"

newswasboring — Fri, 17 Jul 2026 18:20:23 +0000

> The frontier AI companies are betting "the house" on them, and if they pay off they could, hypothetically, make them financially competitive.

They are not betting the house, they are betting the American economy on it. When this crashes it will take everyone down.

New comment by newswasboring in "$100 AI Music Video: Claude Fable 5 vs. GPT-5.6 Sol"

newswasboring — Fri, 17 Jul 2026 15:36:24 +0000

If you consider microdramas as art (who knows what art means to people these days) then the Chinese duanju already has not just full shows using AI but also already have big hits using it.

Yes the people making these are frustrated by how much stress it puts on them (directors managing multiple shows at a time daily must be stressful af) but nobody (neither artists nor audience) are complaining about AI usage. As always they don't hate AI, they hate capitalism.

New comment by newswasboring in "Show HN: Clx – Compile Lua to Native Executables Through C++20"

newswasboring — Fri, 17 Jul 2026 09:20:06 +0000

I'm currently learning to make games with love2d. I'm using love.js to deploy for web. Would this have some advantage over love.js?

New comment by newswasboring in "Bonsai 27B: A 27B-Class model that runs on a phone"

newswasboring — Thu, 16 Jul 2026 00:19:54 +0000

Try these

https://gist.github.com/rick2047/113586b40c30206e45054a15080...

https://gist.github.com/rick2047/9a5544b2b9b6829f753380c6548...

New comment by newswasboring in "Bonsai 27B: A 27B-Class model that runs on a phone"

newswasboring — Wed, 15 Jul 2026 17:04:52 +0000

I edited the comment above

New comment by newswasboring in "Bonsai 27B: A 27B-Class model that runs on a phone"

newswasboring — Wed, 15 Jul 2026 15:05:40 +0000

It is just basic agent setup. When I get home in the evening I can isolate my agent files and share GitHub link. In a couple of hours.

Edit: So life got in the way and I couldn't isolate it. But I have made this project public and committed the dev and code review agents here.

https://github.com/rick2047/meuseum-game-new/tree/main/.open...

The delegate builder is too ugly to share yet. But to be honest just use the normal builder with a frontier model and instruct it to delegate to developer any developer task. That works well.

Of course you have to install the skills [1]

[1] https://github.com/mattpocock/skills

New comment by newswasboring in "Bonsai 27B: A 27B-Class model that runs on a phone"

newswasboring — Wed, 15 Jul 2026 12:54:47 +0000

That is my entire workflow in opencode through delegates. I have an "orchestrator" which uses matt pocock skills like grilling and writes specs in gh tickets. Once that is done it delegates individual tickets to deepseek v4 flash based developer subagent which executes the code pretty fast. This agent can only delegate one subagent which is a code-reviewer which uses a better model like qwen 3.7 for review. So dev does its own review loop before returning.

It's going amazingly. The orchestrator holds the big knowledge from grilling and also enables me to do more grilling to refine the specs. The trick was to check and re-align after each phase was developed. Also carefully defining each workflow step explicitly otherwise it makes mistakes like trying to self review the code. Also needed to define very explicit contracts. I chose the same surfaces as the matt skill set.

New comment by newswasboring in "Claude-real-video － any LLM can watch a video"

newswasboring — Fri, 03 Jul 2026 14:44:51 +0000

NotebookLM still uses the transcript method I think. But Gemini is wonderful. I have been using it to analyze the youtube videos of wrestling matches (trying to build a fan website for WXM, the best pro wrestling promotion to come out of India in a while). It does move by move analysis, audience reaction based match flow tracking, isolates interesting parts of the video (big moves, botches, story beats etc). I have run some experiments to get video editing plans out of it. I think I can combine it with something like remotion skill to make highlight videos.

Edit: BTW, you can analyze about 8 hours a day on free tier.

New comment by newswasboring in "Superpowers 6"

newswasboring — Fri, 03 Jul 2026 11:42:34 +0000

I am currently slowly exploring subagent based feature factories people have been talking about. I looked at superpowers and it just felt like a lot. The process seems be trying to match the complexity of a human team workflow. Seems like the wrong angle to take for this. I am getting a lot out of my simple subagent team by just clearly defining their roles and restricting what they can access. Although experimenting with orchestration policies is required. This feels like tokenmaxxing for marginal improvements.

New comment by newswasboring in "HackerRank open sourced its ATS. My resume scored 90/100. Oh wait 74. No – 88"

newswasboring — Wed, 01 Jul 2026 09:44:48 +0000

Yes I'm using one of the most popular computer peripherals ever invented, yes.