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

Personally, I think it's more that text diffusion is not the ideal driver of an agentic work loop than that text diffusion is a total dead end. I am still hoping to see how it does on authoring and editing with further scaling and optimization. I think the push for AGI has put a bit too much focus on the idea of one general model doing everything.

Honestly, it's an addressable market versus development cost question... How many clients will you lose if you support OIDC but not SAML? Does the delta justify carrying a SAML implementation? If so, do it. But the post is still correct that SAML is a fractal of bad design either way. And it's good to say this openly, and to run this calculus each time you are considering a new SAML implementation.

No algorithm requires tail-call elimination in a general-purpose language with imperative mutability. It's just another way to express iteration.

Sure, but mutual recursion might require `goto` for example. Or an explicit state machine.

I can see how it might require an explicit state machine (keep a mutable state number and switch over the inlined bodies of what could be functions), but I'm not seeing how it could require `goto`.

Are there more-complex relationships that might require it?


I meant it more in the sense that “you need one or the other” rather than “some cases require one and some other cases require the other”.

There has stopped being a formal procedural consequence for OpenAI leaders to declaring AGI, there is a clear (small) business benefit to doing so, and the capabilities of all the frontier models are impressive. So why not declare AGI? It's not like anyone can prove it's not...

Don't be surprised to see other (or even the same) people declaring AGI again and again, as it becomes the best time to do so for different parties.


It seems to support USB4 with what they call "stock firmware". They haven't clarified what that means, but my assumption has been that it's a binary blob made by the manufacturer of their microcontroller that doesn't support all of their fancy debugging features (thus actually turning my into an Aliexpress special eGPU dock with a high quality buck converter).


Seems like they are closer to scratch than reasoning... Generating some scratch to draw from helps make it easier to compute the real answer.


That's my personal theory too. The model is stuffing its own context with vaguely related tokens, which helps the attention heads retrieve the right tokens.


Yup. You basically just need something for probability to push off of


It's also interesting because in humans the existence of "Aha!" moments that are not preceded by or are only loosely related to a chain of thought is taken as the proof of the fundamental mystery and irreproducibility of human intelligence. Now the same argument is made to deny that LLMs actually think. Go figure.


It feels apparent to me that LLMs don't do what is colloquially thought of as thinking.

What is less apparent is that humans do.


> What is less apparent is that humans do.

This seems indeed one obvious hole in the argument of the paper. There is no indication whatsoever that human thinking process is more reliable than LLMs intermediate tokens. Which doesn't make our thinking useless, as messy as it might be. We reorder and explain it after the fact.


I assume theyre searching the local gradient to see if theres a better descent before proceeding.


LLMs dont do gradient descent to generate tokens.

They are trained by gradient descent, but inference doesnt involve it.


I don't think there's anything like that going on. They just word vomit into a secondary area, and then there is an internal prompt that says "clean this up and summarize for the user".


Less "internal prompt" and more "they are trained to summarize after a </think> token"


The training methods try not to apply any particular rules to the contents of the thinking text. That's called "optimization pressure on CoT" and is thought to reduce safety by inducing the model to lie (or stop clearly printing its intentions) in the thinking text.


IPv8 is just a reactionary response to IPv6 not being "IPv4 with larger addresses" like people who know IPv4 well would prefer.

There would be no reason to adopt it if IPv6 were adopted.


Imo the insistence on 'no NAT' has killed IPv6, as it made talking to legacy IPv4 nodes very awkward. If it could interop well with v4, then the transition wouldn't be all or nothing.

And it turns out the simplicity of 'no NAT' still buys you nothing. It's very common for modern systems to run multiple network interfaces with connections having to be juggled between them. In that case, connection and session management falls to a higher level protocol, often in the application.


Hrm, but the insistence is more "NAT should not be a practical requirement in a pure IPv6 network", not "No NAT ever for any reason".

NAT-based compatibility strategies are widely deployed in the real world (most cell carriers and Google's IPv6-mostly corporate network use NAT to allow communication with the IPv4 world from IPv6 only devices).


NAT is accepted for transition purposes; for example NAT64 makes it trivial to connect from v6-only clients to v4 servers, and is used by some large ISPs (e.g. T-Mobile in the US) to avoid running v4 inside their access network.


Yet IPv6 hasn’t been adopted. For decades.

Because it wasn’t implemented transparently at the OS. I can’t plug in to my ip6 only network and reach an ip4 address because end user OSes don’t implement clat transparently, even today in 2026, let alone 20 years ago.


Yet IPv6 hasn’t been adopted.

Well over half of Google’s traffic is IPv6 in the USA, and it’s recently become over 50% of all Google’s traffic worldwide at weekends.

https://www.google.com/intl/en/ipv6/statistics.html#tab=per-... https://www.google.com/intl/en/ipv6/statistics.html


> Yet IPv6 hasn’t been adopted.

news to literally anyone on a mobile internet connection (which is most internet connections!)


Yes you can able to see IPv6 at your browser, but actualy there is always middlewhere sofware level (NLB, ALB, NAT) to reach real backend server, which creates bottle neck. For large scale enterprises this creates issue, even for cloudproviders, I have encountered bottle neck at some Major cloudprovider which causes snowball impact to whole cluster retry mechanisms and 4 hours outage for us and some other customers :D if the 2 enterpise has machine to machine IPv6 connectivity, this is will not happend.


I’m on 4g

My v6 endpoint is unreachable. My v4 one is fine.


I learned recently that France IPv6 adoption is also very high.


> Yet IPv6 hasn’t been adopted. For decades.

I'd argue it has, but that's not the point. IPv4 is pretty much mandatory for most users of the internet. Yes, individual users probably neither know nor core whether they're on IPv4 or IPv6, but it's so difficult to run IPv6-only that you just wouldn't bother.

Trying to teach someone about how to use ULA vs. GUA only leads to headaches. In fact, I'd go so far as to say the reason why the internet became so ubiquitous was because of NAT -- static internal addressing is pretty much fundamental to an easy-to-use experience. Having multiple addresses on a host only makes things far more confusing, so people don't bother, because IPv4 just works.


That's not a fix. A fix would be undoing the change and making it opt-in.


I'm incubating my experience over on Linux


Consumers would buy bread with sawdust and then, upon learning about it, expect the sawdust to come out and the price to stay the same


So you agree that consumers mostly see the sticker price and don't think much about the quality of the product, therefore there must be protections ensuring that the product is at least fit for purpose.


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

Search: