I wish the legislators thought about the privacy implications of this, because anyone can learn your birthday by watching when the category changes.

The brackets are a few years wide, so it could take a bit of waiting. But yeah I’d consider setting a slightly different day/month for a child if I was paranoid.

I guess you could also make the bracket selectable instead of requiring the age

> The brackets are a few years wide, so it could take a bit of waiting.

There are millions of people moving between the proposed age brackets every day. This is a DoB-gifting firehose to ad tech.

Ad tech doesn't need this feature to know roughly how old you are.

Also, various sites were already legally required to gather this information anyway to know if someone is over or under 13.

Wait, are there reasons to use omarchy beyond being a DHH fan?

Are there any other comparable options? Seems like every distro i’ve tried before this wanted to be windows or macos, and didn’t succeed at being either. I like it for not trying to be them.

That said, every time I peek under the hood (or into the omarchy git repo) i get pretty worried the whole thing seems glued together with a bunch of vibe-coded scripts

I always assumed it was basically just arch with some guy's dotfiles.

Hm, the demos all render wrong on my system (Fedora, Firefox). The torus for example is completely distorted.

Edit: example: https://files.catbox.moe/4w3um0.png

I think it goes to show, if you try to make something like this you'll be chasing a long tail of edge cases ~forever.

I've been thinking about something along these lines, but coupled with deterministic inference. At each "macro" invocation you'd also include hash-of-model, and hash-of-generated-text. (Note, determinism doesn't require temperature 0, so long as you can control the rng seed. But there are a lot of other things that make determinism hard)

You could take it a step further and have a deterministic agent inside a deterministic VM, and you can share a whole project as {model hash, vm image hash, prompt, source tree hash} and have someone else deterministically reproduce it.

Is this useful? Not sure. One use case I had in mind as a mechanism for distributing "forbidden software". You can't distribute software that violates DMCA, for example, but can you distribute a prompt?

Deterministic inference is mechanically indistinguishable from decompression or decryption, so if there's a way to one-weird-trick DMCA, it's probably not this.

You’d think that, but it sees like big business and governments are treating inference as somehow special. I dunno, maybe low temperatures can highlight this weird situation?

Temperature is an easy knob to twist, after all. Somebody (not me I’m too poor to pay the lawyers) should do a search and find where the crime starts.

What does temperature have to do with anything?

Or however deterministic inference is supposed to happen. I don’t know LLMs.

Well, it's still not deterministic even at temp 0. The tech described in my comment's parent is speculative, and technically it's not even inference, once it's perfectly reproducible.

At that point it's retrieving results from a database.

EDIT: how would OP address my main point, which is that det. inference is functionally equivalent to any arbitrary keyed data storage/retrieval system?

> The tech described in my comment's parent is speculative, and technically it's not even inference, once it's perfectly reproducible.

This is not true. Fabrice Bellard's ts_zip [0] and ts_sms [1] uses a LLM to compress text. It beats stuff like .xz etc but of course is much slower. Now.. if it were non-deterministic, you would have trouble decompressing exactly into what it compressed. So, it uses a deterministic LLM

[0] https://bellard.org/ts_zip/ https://news.ycombinator.com/item?id=37152978

[1] https://bellard.org/ts_sms/ https://lobste.rs/s/5srkwz/fabrice_bellard_s_ts_sms_short_me... (funny enough many people comment that, if it uses LLM, it must be lossy. This is not the case. It's compared to xz in the page because it's lossless)

Deterministic inference isn't speculative, it's achievable if you want it. It's just not the default.

It seems the TL:DR is race conditions, rounding and other RTE specific factors.

Most of the HN bot accounts I see have a link-to-vibecoded-product in bio, and/or are trying to build up "organic" activity before a Show HN post for the same.

A less publicly-visible motive would be if they were building up accounts to use for paid-upvote schemes.

> Those who know have every incentive not to share

Why do you say that?

I wrote one that uses opencl: https://github.com/DavidBuchanan314/anubis_offload

Bravo, you even implemented the midstate speedup from Bitcoin, that's way more impressive.

It's not exactly rocket science heh, just baffling that the original anubis impl left an order-of-magnitude speedup on the table.

Maybe my imagination is just too accurate but this didn't tell me anything I didn't expect to hear.

> Here is a massive log file for some activity in the Data Export tar pit:

A bit of a privacy faux pas, no? Some visitors may be legitimate.

Like most things in the audiophile world, it's more about aesthetics than anything else. A big cable looks like it means business.

I think that's being a bit uncharitable to B&W specifically; they're one of the few headphone companies where the engineering does back up the price. The cable is the odd one out.

I don't have an informed opinion of B&W either way, but are you sure it's not an instance of Gell-Mann amnesia?

I did a <4KB one that supports DNS+TLS (albeit insecurely) https://github.com/DavidBuchanan314/kurl/

This is of course great, and very creative, but why not use an HTTP proxy if this TLS is still unprotected?

Did I misunderstand something?

Many servers support TLS only, so implementing the client side of TLS was a matter of compatibility, not security. If external dependencies are allowed then why not just use libcurl?