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

This like saying generic systems are bad because you and a hacker both can make sane assumptions about it, thus even if more performant/usable it's also more vulnerable hence shouldn't be used.

I don't understand this.

I have seen bad takes but this one takes the cake. Brilliant start to 2026...


I think you are proposing something that's orthogonal to the OP's point.

They mentioned the training data is much higher for an LLM, LLM's recall not being uniform was never in question.

No one expects compression to be without loss when you scale below knowledge entropy that exists in your training set.

I am not saying LLMs do simple compression but just pointing a mathematical certainity.

(And I think you don't need to be an expert in creating LLMs to understand them, albeit I think a lot of people here have experience with it aswell so I find the additional emphasis on it moot).


The way I understood OP’s point is that because LLMs have been trained on the entirety of humanity’s knowledge (exemplified by the internet), then surely they know as much as the entirety of humanity. A cursory use of an LLM shows this is obviously not true, but I am also raising the point that LLMs are only summoning a limited subset of that knowledge at a time when answering any given prompt, bringing them closer to a human polymath than an omniscient entity, and larger LLMs only seem to improve on the “depth” of that polymath knowledge rather than the breadth of it.

Again just my impression from exposure to many LLMs at various states of training (my last sentence was not an appeal to expertise)


Gpt 5.2 pro is well beyond that iirc


Whoa! I had no idea. $21/$168. That's 75x / 400x (1e1.875/1e2.6). https://platform.openai.com/docs/pricing


It's probably spark dgx competition. So around 3-5k would be ideally it.

Any higher and its not really a disruption


Or contrary to that they should not have the ability to delete user data... And make a threat about it.


Because they are not?


Sample size of 1, but the most popular article on my blog is about to practically build a modular monolith. It was inspired by a Google paper published 2 years ago: Towards Modern Development of Cloud Applications[0] which touched on the shift from microservices to monoliths that can be deployed modularly.

Very much worth the read as it is easy to skim.

[0] https://dl.acm.org/doi/pdf/10.1145/3593856.3595909


That RFC and Polonius, which Rust folks have been working on for the last 5-6 years is proof that there has been much effort made in related directions.

Rust being sub par for so long just shows how much people won't want to fund these problems and how hard they are to solve during program compile.

I ofc like Zig quite a bit but I find Rust to suit my tastes better. Zig feels too much like C with extra steps. And the lack of good tooling and stability around Zig hurts large scale adoption.

But I think in 10 years Zig will be the de facto better-ish C.

And Rust will be the low level language for any large project where safety is amongst the top 3 priorities.


The problems Rust are trying to solve are both novel and difficult so it isn't particularly surprising that it's taking time. The team has also landed great improvements, like NLL. I'm optimistic about the direction of this, even if it takes time.

Zig feels much younger than Rust so we'll see how it develops, but it's certainly interesting. In particular, comptime and explicit allocators are two ideas I hope Rust borrow more from Zig.

> And Rust will be the low level language for any large project where safety is amongst the top 3 priorities.

Personally I don't really see what'd be left for Zig, because in most software of consequence safety is already a top 3 priority.


Now let's consider a different form of govt/shared charges.

Taxes, why are individual and corporate taxes and structures different?

Both are doing work and generating income?

Why do corporations get to deductions and do so much tax magic that individuals don't?

why don't we charge them both with the same laws, and structures?

Cause corporates generate jobs? Isn't that unfair to the people born individuals...

I don't think anyone is saying there should be stabilization of electricity prices. But costs for grid improvements for industrial, data center or AI usage should be on the said companies.

They are using that resource to generate a profit.

While people in residential homes are just living their lives, you are comparing cost of essential commodities to production inputs...

We make taxes on poorer people less as well for the same reason.

But yes if someone is saying electricity prices should be stabilized for early consumers that's definitely unfair. But I didn't see or read that here.


Apologies, I misread some of this yesterday.

The easy way to justify why corporations get to deduct their spending, is it encourages corporations to spend. Something that a for profit company would not necessarily do otherwise. With the obvious note that spend sent to people is taxed as income for that person.

Now, I agree that this gets super odd when people also make an odd "corporations are people" argument.

As for my assertion that this is effectively arguing for price stabilization. The entire thing hinges on the complaint that costs have been going up. Which, of course the price of a good that has increasing demand is going up. I'm not sure how to read that other than an appeal to price stabilization for early consumers.


>Why do corporations get to deductions and do so much tax magic that individuals don't?

Individuals can do deductions if they're a sole proprietorship. The requirement in general is that the expense is used for a business purpose. Buying a laptop for personal consumption isn't tax deductible, but buying it for your consulting business is.


Isn't that proving my point you change laws based on what you do with your money and where you get it from.

If so data centers should also be separately booked for their electricity consumption.


I am not sure have CEOs gone insane over AI? I wouldn't even agree to Sex Ed with AI for kids this is getting insane.

Can we not stick to coding stuff I know you folks aren't making profits, but please try to think about the consequences dammit.

Edit: I don't like AI code but atleast it can't harm anyone if we have decent guardrails.


> I don't like AI code but atleast it can't harm anyone if we have decent guardrails.

Are you sure about that?


Are there any differences from git, or tbh is there a project I can maybe contribute to on Radicle and learn from experience?

Can pitch in with helping out Windows support


It's "just" git, but you push to a special remote which will synchronize your repo on a p2p network.

There's also a CLI for issues and pr's, which also get's stored in your git repo.


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

Search: