nateb2022

> In comparison: license plates, when in public, are always visible, and very easy to discern from one-another (different state-unique numbers); so in my mind the expectation of privacy is far lower.

According to Carpenter:

  A person does not surrender all Fourth Amendment protection by venturing into the public sphere. To the contrary, 'what [a person] seeks to preserve as private, even in an area accessible to the public, may be constitutionally protected.'

Being present in plain view isn't equivalent to a total surrender of privacy.

Curious if bunny.net offers any measurable DNS performance improvements over Cloudflare or vice versa.

[dupe] https://news.ycombinator.com/item?id=48648039 (2 hours ago, 58 comments)

I applaud this. Assuming this is to combat the rampant LLM posts and comments that irk me whenever I visit.

Feels like a solution in search of a problem; a reinvention of https://docs.docker.com/ai/sandboxes/

opencode (https://opencode.ai/docs/permissions/#defaults) already forbids access to .env by default.

On Amazon I see a "Orange Pi 6 Plus 32GB RAM 12 Core 64 Bit LPDDR5 Single Board Computer, CIX SoC 45TOPS AI NPU Mini PC Run Linux, Android, Windows, ROS2 OS with Heat Dissipation Assembly with Cooling Fan" listed for $734.99 USD.

You can get a mini PC for less with a modern Ryzen 8845HS/Core Ultra processor with a SOC that pushes far past 45 TOPS. And you'll get standard x86 architecture so you won't have to deal with ARM oddities or fight with non mainline kernels.

> Mozilla’s fast release cycle for Firefox is a minor irritation, yes. (Of course, there’s always the Extended Support Release channel, if you want to hop off the treadmill.)

I am unaware of anyone irritated by Firefox's release cycle.

> Seems very unethical, no? Who uses service providers like this? The whole point of anti-bot measures is to get rid of bots - you are not wanted there.

I'm familiar with companies automating access to software only accessible via the web with poor/no API support. This is software they pay (usually a lot of money) for, and usually has built in captchas to guard logins. They aren't a large enough customer to ask the removal of these captchas or whitelabelled (just one out of many SaaS tenants), so they simply work around that restriction.

> With Qwen3.6-35B-A3B-MTP-UD-Q4_K_XL.gguf?

Ahh no I'm using the MLX version, it's about 5-10% faster than GGUFs in my experience.

I get over 100 tok/s sustained on my M4 Max and M5 Max, in MacBook Pro's. LM Studio + MLX.

> As long as you use proprietary OSes, do not ever update past what has come with the hardware(reinstall back to it if needed). That's why there are still machines using windows xp still in use today. That's where you will get the fastest/best experience.

This is terrible advice that would cause a TON of people to become exposed to patched vulnerabilities.

How's your battery health? A severely degraded battery will not be able to deliver power at the rate your system may be demanding it. I'd replace it asap.

What sites already are using JPEG-XL en masse?

Not 100% sure but I believe it uses gnome-keyring under the hood, I'd install that and start it with

  gnome-keyring-daemon --start

I think an important caveat is that LLMs are prone to writing unnecessary code, i.e. there's almost a superfluity to it, that at the same time makes it less straightforward and more prone to unnecessary side effects, which it does catch and handle, but that again expands code further.

Imo the article is completely copy/pasted from Claude. The figures, which couldn't be copy/pasted verbatim, were screenshotted/pasted.

Quoting from the article.

> The Town of Atherton's lawsuit against Caltrain electrification is the clearest case study we have of how a tiny, wealthy minority can hijack a regional public project — and stick everyone else with the bill.

The unnecessary emdash-for-effect is a huge LLM-ism. No human I know uses such a device in writing.

> Here is the short version. In 2012, Caltrain budgeted its electrification project — the backbone of the Peninsula's transit future and a prerequisite for high-speed rail to ever reach San Francisco — at roughly $1.5 billion. By 2017 that number had ballooned to $1.9 billion. In between, the Town of Atherton sued.

"Here's the short version" along with the bolding of the numbers tells us something about the "author's" prompt. He prompted it for a short version, and told it to include those two numbers.

I challenge any commenter to name one time they began a paragraph with the meager "Here's the short version."

> One veto point, closed.

Why is the comma even there? Classic LLM melodrama.

A broken clock is correct twice a day.

I agree with the sentiment although the post seems unnecessarily vitriolic. It's AI generated however, and I don't believe AI generated slop posts deserve consideration, either on HN or Reddit. If they rather not take the time to write something that long, they're free to write a prompt-length post instead of forcing us to read the AI equivalent of a zipbomb.

This is a completely AI generated Reddit slop post. Most of the Reddit OP's comments are also AI generated. Lately I've seen a lot of AI-esque activity on Reddit, especially especially in Codex/Claude Code subs. Which is not surprising, but I find it hard to believe that anyone would take the time to copy/paste LLM responses that consistently. I think it's bots.

Mado is older (~April 2024) vs rumdl (~March 2025). It's also faster, per rumdl's own benchmark graph.

Mado is purely a linter; rumdl is also a partial formatter, and offers a LSP.

Panache has a well written and broader comparison between these, and other tools here: https://panache.bz/guide/comparison.html

Have you noticed a gap between 8bit and 4bit quant? I've always ran 4bit quant cause less memory required

Comparing to Opus is a little unfair, a comparison against Haiku would be more fair. And for a really fast cloud model, I'd be interested to see how latency stacks up against GPT-5.3 Instant or Gemini 3.1 Flash Lite.

Pricing is fair! And I like the design, props to you

Might be worthwhile to note that Google Docs now supports Export to Markdown

Have you seen https://www.daytona.io/ ?

This is awesome! Looks well built and the value prop is reasonable.

Your pricing and "what it does" navbar links aren't working

GitHub is immune

Given the sensitive nature of these filings, I'd consider SOC 2 certification to be a minimum.

Congrats, that's a huge achievement!! Curious where your ARR is approximately right now, and how long it's taken you to get to this point? What's your marketing strategy looked like, and is it just you running things?