"Computing is a pop culture with little regard for the past" this literally just clicked so many pieces in my brain all at once I had to close my eyes and sit with it.

"Social media content factory" "[Automated] Ebay flipping"

What a vile, vile thing.

SOUL

Cried seeing this. So much loss.

Really good use of the toupee fallacy here.

Sorry, do we have actual clear implementation details for Jev? I keep seeing these "recreations" or "Do Jev at home" but do we have access to their architecture? I haven't even used the product, I just find it strange.

What they did was immediately obvious and trivial to replicate

They make it clear that their edge is in their synthetic data. Train your own jevs miss the point. And if you have that much data, you didnt need jev or these replacements in the first place.

HN's desire to pretend the data pipeline doesn't exist or isnt meaningful is silly.

I'd argue that a tiny fine-tuned Jev on my dataset that I can run for practically free is far more valuable if the idea is to make this a dependency in real work

People have been making their own calibrated classifiers and decision makers for a very long time. Mine outperform jev. But that's not the point of jev.

What is...?

Being ZERO-SHOT. No fine tuning or re-training needed. Why are people so dense over this model?

i guess it's not needing to finetune for each specific use case

Then the probabilities are not calibrated. Jev is basically the LLM version of the "is-even" library.

That is indeed reasonable. But you still need a bunch of trusted data to validate the accuracy of Jev.

Yes. The difference is you can immediately benchmark and iterate via prompting without having to retrain. It's such a time saver.

Is it though? I have found it much easier to fit/diagnose/improve simple classifiers on embeddings/hash vectorized features/etc than iterate on prompts, especially true when using the more modern LLM's (like gpt 5.6 Luna) where you can't even set the temperature to get any sense of determinism.

I do note though, that is infinitely easier to 'deploy' a Jev/LLM based solution than a data/model pipeline.

The comparison i'm making is against a training of encoder/small decoder model so I think we agree on most things. I dont think jev replaces the benefit of embeddings nor think they are mutually exclusive. All part of a handy utility belt.

The website, to me, looks like Claude's project with sprinklings of the author's intent, and not the other way around.

Yes, I agree. It only having one working link is the killer to me as well, because no one checked it besides the pretty front page. There was an article here yesterday about recognizable signs of AI-designed UIs and it was accurate in terms of what most of these sites now look like.

https://news.ycombinator.com/item?id=49867038

Related and not related. Fans of https://andymatuschak.org/ might want to know that he is now working at https://x.com/pico, which apparently has the same tagline as this article "Attention is all you have". But really truly don't click that link that site is a hellhole.

(Edit: Wording)

Not to shill but if you're struggling with apps on your (i)phone, I can't recommend https://www.foqos.app/ enough. My weekly screentime (as measured by my iphone, not true screen time) is <1hr. NFC tag is optional you can literally buy a candy bar and tape the barcode to an index card and that will work. The physical layer is massive. Now my problem is just limited to desktop which is easier to control.

+1 on physical layer being the game changer. I've been using getbrick.com with amazing results.

talysman.app does the same thing as Brick but for desktop. So you can code/etc with a physical layer of protection

Definitely a game changer, maybe even more than Brick because the desktop is where I do all my work