Hacker Newsnew | past | comments | ask | show | jobs | submit | aetherspawn's commentslogin

Australia copied the Swiss model and in a very short period of time we went from 2Mbps flaky copper to now you can upgrade most properties to 2Gbps fiber for around $300 one-off fee.

I hear 10Gbps is coming soon. The only annoying thing is that ours, despite being terminated the Swiss way, isn’t symmetrical, I think due to congestion on the sea cables?


I highly doubt that. As a German who lived in Asia and now Australia, this is the most incompetent country. They can't get anything right. I live in the city of Sydney and can only get 1 Gbps down and 200 mbps up for 109 AUD a month. I lived in 2017 in Singapore and the internet was already better than that for 1/3 of the price.

Australia seems to be pretty backwards in general though.


In Belgium you get 500/100 for 65 euro, which is even worse.

Can you use the smaller Gemma 4B model as speculative decoding for the larger 31B model?

Why/why not?


Interesting, maybe you can run the output through a 2B model to uncompress it.

Which harness (IDE) works with this if any? Can I use it for local coding right now?

Yes, you can use it for local coding. Most harnesses can be pointed at a local endpoint which provides an OpenAI compatible API, though I've had some trouble using recent versions of Codex with llama.cpp due to an API incompatibility (Codex uses the newer "responses" API, but in a way that llama.cpp hasn't fully supported).

I personally prefer Pi as I like the fact that it's minimalist and extensible. But some people just use Claude Code, some OpenCode, there are a ton of options out there and most of them can be used with local models.


It needs to support tool calling and many of the quantized ggufs don't so you have to check.

I've got a workaround for that called petsitter where it sits as a proxy between the harness and inference engine and emulates additional capabilities through clever prompt engineering and various algorithms.

They're abstractly called "tricks" and you can stack them as you please.

https://github.com/day50-dev/Petsitter

You can run the quantized model on ollama, put petsitter in front of it, put the agent harness in front of that and you're good to go

If you have trouble, file bugs. Please!

Thank you

edit: just checked, the ollama version supports everything

    $ llcat -u http://localhost:11434 -m gemma4:latest --info
    ["completion", "vision", "audio", "tools", "thinking"]
so you can just use that.

This is really buggy and janky… try it in mobile, the navigation doesn’t even work. UX elements move around the place when you hover or click them.

Anything built in Svelte can be deployed to workers easily, and it’s a very good platform.

Just missing compartmentalisation features between prod and dev environments.


Not really an issue though right because virtually none of these have lasted more than 1-2 days before being discovered?

Is the profession cached in the data when they leave the job? And does the data attribute 2 entries for someone with 2 careers. That’s the question I think

They explain it in the article. Someone, often the funeral director filling out the death certificate, asks what the deceased did for most of their working life.

I’m a little skeptical of the category “ambulance drivers; not emergency medical technicians” as reliably coded, because people will often say so-and-so “drove an ambulance” when they were actually an EMT or paramedic. But it’s also not clear to me that would invalidate the findings.


Thank you, as much as a 160 page book about fonts is probably thrilling, I probably won’t get around to it for a while so was going to ask for the tl;dr

There is no compelling evidence that san-serif fonts are less readable than serif fonts under any circumstance, despite the oft-repeated lore that typographers consider serif fonts to be more readable than sans-serif fonts.

A few months ago, on a price hike announcement for Office365 posted here to YC HN, I made a comment that MDM is expensive, had high MOQs (Mosyle, Jamf) and fundamentally still doesn’t work as well as Windows and Intune. I also lamented that Microsoft keeps hiking prices and that it’s silly we’re normalising $20+ per user per month when we used to pay once for these things.

I lamented how Apple hardware is now the same price as the other vendors, yet best in class for quality and how Dell and HP are hiking their laptop pricing lately due to supply shortages. Especially on their pro lines, which have been quoted to me as twice the price of equivalent MacBooks.

I mentioned Apple would be silly not to make a further global move into MDM and email hosting territory. Particularly for small business owners: 1-10 person shops and retail who use mostly cloud based POS applications.

Others responded at the time, and I agreed with it, that it seems unlikely Apple would make a business move. After all, they don’t have much history with business, or perhaps they did but they didn’t like the market and wrapped it up.

Well, with this announcement, and with the confirmation that *Apple native email hosting is coming* I am very excited to trial it when it lands in April. Over the last few months, our small business has already cracked it and downgraded most of our email hosting to Exchange Plan 1 and dropped the desktop Office suite in favour of Pages and Numbers, which are both free and absolutely working fine. In fact, I’ve found Pages to be less laggy and more stable than Word in very large documents such as 300+ pages. The logical next step for us is to fully drop our third-party MDM and review whether Apple’s native MDM, email and identity systems are adequate for transition. We have saved thousands of $$ so far and stand to save a lot more!


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: