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


This website, thru one link and another lead me to (The rabit hole of ) "mechanical calculators". A mechanical marvel for me.

I know it’s tiring to talk about “hallucination”, but truly, models still do hallucinate

They constantly say they did a thing they didn’t, say they know how to solve something when they don’t, etc. Regardless of guard rails or tests - AI forces a constant vigilance of a new kind.

Not just “what might have gone wrong” but also “what do I think is working but isn’t actually”.

And we’re not even talking about how it chooses substandard solutions, is happy to muddy code/architectures, add spaghetti on top of spaghetti etc.

Agentic coding often feels like an army of unexperienced developers who are also incredibly eager to please.


"still" isn ghecorrect word. They always be having hallucinations

"Still" means "it always had hallucinations, and it still does, despite people thinking that it doesn't anymore". People think we've moved past that. We haven't.

Hallucinations are baked into the design, you can't make a LLM that doesn't have them.


Those were super fascinating and inspiring to read! Thank you


I didn't know about HERMES.md ... (??) - found information here for others who are curious https://github.com/anthropics/claude-code/issues/53262


This github thread is incredible, thanks for sharing. This link should be its own HN topic.



That is insane, if you billed me an extra $200 for a bug in your system I'd flat out cancel my subscription. If you're not going to credit that back to me, you don't deserve anymore of my money. I'm a Claude first guy, but if you're going to bill me incorrectly, that's on you, own it, fix it.


They did credit it back to him. There's a comment in the linked issue.


Where? Just searched the entire thread for both the word "refund" and the word "credit" and I'm seeing nothing about credit being issued.

Also what's with @sasha-id talking to himself? Looks weird as all get out.


Looks like he copy pasted responses he got from their support agents.



Where? All I see is Boris saying "we are unable to issue compensation for degraded service or technical errors that result in incorrect billing routing".


Keep this in mind next time you hear someone talking about "removing the human in the loop".

Anthropic apparently won't take responsibility for issues their own systems handling billing cause. You think they'll take responsibility in your system when a bug in their models can be demonstrated as the cause?


> Anthropic apparently won't take responsibility for issues their own systems handling billing cause.

I think with every org, especially the big ones, trying to dodge responsibility (setting the intent of "customer support" to be annoying them enough for them to buzz off), the only recourse people have is to give them enough bad press where they wake up and do the refund, it's less than a rounding error for them.

I think Anthropic is hardly unique in that position and being able to chat with a human with any sort of power to actually make things right is becoming more and more rare. If any human eyes saw that, the correct thing to do would probably be passing the message up the chain like "Hey, this will have really bad optics if we don't do the right thing. Can you take like 5 minutes and hit the refund button while I draft up a nice message about it?"


Bad press is meaningless where it matters most these days. The kind of people who are most responsive to threats of bad press are the kind of people who don't need to be threatened with bad press to do the right thing.

I really wish it carried any weight. It just doesn't. If someone at the organization just says "never admit fault, always attack", it's very likely they'll get away with it.


> You think they'll take responsibility in your system when a bug in their models can be demonstrated as the cause?

Flag on the play: AI doesn’t replace responsibility for your commits.

It doesn’t matter what promises a service makes, what you say is valid code is still on you.

Act accordingly.


The issue is less what's in your commit and more if you're using these models as a foundation for some other service.

I know this is a rather hackneyed example, but if a customer service agent model were to call a customer a racial slur, that's not the software surrounding the agent, it's the agent's model.



take a look at the ufactory xarms - https://www.ufactory.us/xarm

we have a bunch at work, happy to chat about them (email in profile), they are about 10k


Are you interested in the graphics (how to render 3d, opengl/directx, shaders etc), in the “game engine” aspect (all of the various parts of a game coming together), game design, networking, or something else? All of it?

As people used to say on stack overflow - what have you tried so far?

Also - this might be a good starting point: https://www.gameenginebook.com/


- i have started brushing my c++ and learning 3d math for now

- i am interested to know for starters what does the 10000 ft landscape look like

- what are the major steps involved?

- what have I tried? I spent a lot of time looking into what kind of libraries exist. Rust has something called bevy which I would not touch given it is new and I dont want to work with double unknowns

- C++ has a library called raylib that seems to have a lot of functions for basic stuff.

- There seems to be a library called jolt which claims to handle physics well

- how is rendering / graphics pipeline stuff handled in c++. what libraries do we have for loading models, working with shaders, handling networking for gaming etc?

- the thing is i need to get a complete picture inside my head of how a 3d game is put together like how this https://visualrambling.space/neural-network/ gives you the picture you need to understand how a neural network works


A 3D game, like any interactive program, is a REPL over a state machine:

1. Setup initial game state

2. Render graphics in accordance to game state

3. Read input and update state in accordance with it

4. GOTO 2

The purpose of C++ is to prevent you from realizing this. Don't use it and don't talk to anyone who uses it. Same for Java, C#.


I agree with you on that metric being not great - I would have definitely swapped it for this:

"Claude Code GitHub Commits Over Time" https://newsletter.semianalysis.com/p/claude-code-is-the-inf...

Sure - also an imperfect metric. But less imperfect? And more indicative of... something? Not nothing?


@kevinsync's answer is 100% correct and it's been this way for the last ~~~20? years? at least - only it was "Photoshop files hold the (design) truth" before - now it's figma.

But yes, the "design to code" gap has always been where designers' intentions were butchered and/or where frontend developers would discover/have to deal with designs that didn't take into account that some strings need more space, or what to do when there are more or less elements in a component, how things should scroll in real life, how things should react to a variety of screen sizes, etc.

this short meme video is funny/not funny because it hits too close to home - https://www.youtube.com/shorts/r6JXc4zfWw4 - but yes, "designers don't code and developers don't design", roughly speaking

and then of course you meet some people who do both really well... but they are pretty rare. :-)


It's a real shame that people bought into this false dichotomy, because the base reality is that people who work in web dev that stubbornly pick either code or layout are more of a liability than an asset.

I don't believe that people who can design and code are as rare as folks seem to believe, either. What seems more likely is that there are a LOT of coders who are extremely fluent in CSS but aren't particularly gifted when it comes to making things look good.

It wasn't that long ago that designers understood that they couldn't just hand off a 2D comp of what they want to see. The job isn't done until the output can be integrated into the app. Nobody gets to launch cows over the wall and go for lunch.


> only it was "Photoshop files hold the (design) truth" before

You mean Fireworks. Photoshop was for graphic design. Web designers used Adobe Fireworks. Figma is a successor to Fireworks, not Photoshop.


Nah yeah Photoshop .PSD's were totally normal for website designs. I got extremely proficient at building functioning websites based on PSD files, going back as far as the days of using nested <table> structures with 1x1 transparent spacer.gif images :) I built hundreds of websites from .PSD files, and Fireworks was pretty much non-existent in my experience.


Yeah +1, at the start of my career (mid 2000s) all the web designs were done in Photoshop. I think this was true until 2010/11 at least, when I first encountered Axure being used for some design stuff. I’ve never used Fireworks and never saw it in the wild (UK, in case it was a regional thing)


Those days aren’t over. HN uses nested tables with 1x1 transparent spacer.gif images.

I like that HN’s design is timeless. It’s a shame people usually ignore it, or are critical of it.


The text is barely legible, and you can hardly ever click on the actual target you want (e.g. upvote vs downvote).

It's timeless in that it was never good to begin with, it is just actively malicious now for anyone using it from a touch interface.


> It's timeless in that it was never good to begin with

True but it shows that good design doesn't matter.

Also: Craigslist, Reddit, LinkedIn...the list goes on.


I was in web agencies since like 2002-2015, always got PSDs from either clients or internal designers


100%. I was always told to slice the PSD, fireworks never entered the conversation in the agency world I was a part of.


Sure, and also Illustrator sometimes, and Photoshop at other times. Some of the designers I know (very famous for their ui/web work) never touched vector components and just had a ton of layers in Photoshop and air/paintbrushed everything. Hence the meme...


I've even received designs in PowerPoint

Everyone used whatever they were familiar with regardless of the purpose of the application.


Many designers stuck with Photoshop sadly. Back when I did agency work it was absolutely normal to get PSDs of mockups.


I've seen the front end devs get PSD files as recently as 2021.


I think his point was made regardless of his mistake


I wouldn't even classify it as a mistake, just a difference in experience regardless of what Adobe's intentions were.

As someone who has done front-end development for both web and mobile devices for a very long time in the pre-Figma days I was handed a lot more designs that were mocked up in Photoshop than Fireworks.


It’s just one more example of people realizing that the code is the spec.


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

Search: