waterTanuki

This is the cost of living in a federated republic. Power is diffuse across the states so they all have different opinions on how to go about things like vehicle registrations and insurance. The only alternative is for states to relinquish the power over vehicle registrations/plates/insurance requirements to the federal government which could enforce a universal standard, but I'm sure given current events you would see why that would be a terrible idea.

I think as programmers we naturally want a single, universal standard for everything, but there are genuine advantages to having the current system as well, even with all of its flaws.

> The tulip became a luxury item and many varieties were introduced. The varieties were classified and the most sought-after, prized tulips were the streaked tulips, especially yellow or white streaks on a red or purple background. These flame-like tulips were highly sought after. Interestingly, the streaks or “flames” of the tulip petals were caused by a virus. The virus is the tulip breaking virus, or tulip mosaic virus.

Source: https://www.canr.msu.edu/news/tulip_mania_the_history_of_the...

What's old is new.

Let's bring the subject back into focus: This is an example of an LLM being used to search a massive database of scattered text and files. Are you saying LLM's are not suited for searching through text?

You can disagree but the fact is people will take the path of least resistance. I'm not saying we shoudldn't strive for 100% accuracy but we also need to be realistic about what an LLM is. I'd rather we not pretend there's some magical combination of weights out there that will make it completely perfect.

Early on in my career as a software engineer, I was told by my managers that I had to sell my ideas as an elevator pitch to get anywhere. If I couldn't explain why a feature or fix would improve the business, it would go into the wastebin.

Colour me surprised then when I see, since circa 2024, an absolute deluge of AI products coming from these companies, each with their own naming scheme, marketing page, and accompanying hype post, and despite giving them MORE than the 30 second elevator pitch interval, I am left not understanding what it is they want to sell me. I have claude, I have codex, I get my work done just fine with each. I don't need another product for this. Any software these companies try pushing always ends up being something I could've made with their own AI models in a day and then never end up using. What's the end game here? What's the goal? What's the moat? I'm not sold on a single product outside of codex and claude code.

> I agree in general, but how do you resolve the problem that people will take the output of the chatbot as authoritative even when it could possibly have errors or miss important nuance?

The bar for something like this should not be 100% accuracy. It should be 100% accountability and transparency, followed by being at least as accurate as google search.

Similar to autonomous driving: it doesn't need to be perfect, just less likely than humans are at causing an accident.

Alot of mock services have come out of the woodwork now due to genAI, but I don't think these are the solution from decoupling your services from cloud providers, because they have to keep up with schema updates for every single service which isn't practical even with genAI.

You should be writing everything in a sans-IO pattern: define your core, the interfaces, and your own types, (even if it seems redundant) and never, ever import a single dependency from a cloud provider in that core. Make two implementations of the interfaces: one in-memory that can be spun up quickly for testing and one that uses the actual cloud provider APIs. Now instead of hoping fakecloud or localstack update their mocks to be inline with AWS, all you have to do is update your dependencies and the implementation code and the core can be left alone for a long time.

This is also why I avoid lambda and other "serverless" compute like the plague because it's designed from the ground up to lock you in.

When did Windows come into the picture? I was just talking about Ubuntu.

I find it hard to believe any enterprise willing to drop over $40k on a workstation PC would choose `Pop!_OS 24.04 LTS with the COSMIC Desktop Environment` over Ubuntu.

> [01:42] A pilot called for flaps to be set to 20 degrees, and the other pilot cautioned they were too fast. The other pilot made additional comments regarding the aircraft’s excessive speed throughout the remainder of the recording. There was not a consistent verbal response to the comments about speed

The interaction here is eerily similar to the pilot interactions preceding the 1997 Korean Air Flight 801 (B747) disaster in Guam [1]. It's not clear based on the NTSB update if the Captain was ignoring the First Officer or the other way around.

[1]: https://en.wikisource.org/wiki/Korean_Air_Flight_801_-_Aircr...

Turnaround at a gate and a life-or-death go-around judgement are two entirely different operations governed by entirely different sections of the FAA.

Also this was Amazon, not Southwest.

> - the proof generated by our model was very different from theirs and also goes far beyond the published literature

I'm hearing two completely conflicting stories. Buckmaster is claiming the approach used by OpenAI is so strikingly similar to the one he used, that mere coincidence is astronomically small. Yet OpenAI is claiming that the methods used are entirely different.

Anyone care to provide primary evidence proving one way or the other?

Ok, so what is the exact number of preventable deaths per year to build a silicon god you would be ok with?

> What's the most dangerous thing that's happened with an LLM so far?

I don't know, maybe a mass shooting?

https://www.npr.org/2026/09/02/nx-s1-5953021/openai-tumbler-...

Oh, and let's just forget the uncountable early deaths from the environmental disaster of the Datacenter buildout. It's not as sexy and doesn't make headlines, so those deaths don't really count or matter do they?

Hi, fellow American here.

Me and quite a few others are getting real tired of the constant name-dropping of mega-corporations that do not care about your safety, health, or well-being. This does not resonate with anyone except a small circle of wealthy tech elitists.

If they can make a profit from those things, great, but they will also not hesitate to make a profit from your suffering as well. Ask the lung doctors of the people living next to xAI's data centers if their patients are better off than before.

> The market for people who want to install Linux on a Mac laptop is very tiny.

It's only going to grow every year as apple stops releasing OS updates for perfectly functioning hardware.

People didn't start using iPhones because they were just better than existing alternatives. Apple did a lot of marketing to get the name out and people excited. Then they stuck around because the product was actually good.

We're seeing a mass exodus of companies and people from the Windows ecosystem. If this gets more people using/familiar with Linux I can only see this as an absolute win.

It's clearly marketed for remote communities with little to no connecting infrastructure. Think Alaskan wilderness or Hawaiian island hopping. This isn't for getting from LA to Denver for skiing.

> For the many people who resist the AGI label possibly ever being achieved, I'd be curious to hear takes on what would make you think Astra is yet to be AGI, and what would still need to be achieved for this to effectively be AGI from this point forward.

Stick to the original definition of AGI of an AI model being able to self-improve independently with 0 human intervention and become an "everything" solver. Ever since money got involved in this, the goal posts have shifted considerably. If OpenAI truly had an AGI on their hands they would then be able to crack encryption, destroy world markets, and funnel all resources back into their new for-profit organization. Since their mission is now share price, until I see any evidence of an infinitely growing stock I will reserve my congratulations.

The mere suggestion that this admin is doing this or anything else for the wellbeing of its citizens is a complete departure from reality.

It's very clear what the motivations are, and just because they don't seem harmful in isolation doesn't mean it's going to have no consequences for everyone else.

You're conflating the opportunity cost of having a child with the actual logistics of raising a child.

Cities are by definition economical logistical hubs. It's where the jobs and opportunities are. Most people aren't saying "gee I better pack up and move to suburbia before having a child" it's more like "having a child could cost me my high-paying job so I better not". And that issue is exacerbated in the U.S. with the worst parental-leave policy among developed nations. Regardless though, the trend will still exist in countries with generous parental-leave policies because having a child is an opportunity cost in an urban environment. That doesn't mean the environment itself isn't conductive to raising the child. Completely different stories.

Cities being dependent on workers being present at commercial office buildings appears to be a uniquely American problem.

Most of my American friends I know turn to the suburbs as a default for "family" style living once they have kids. Imagine their bemusement when they go to a place like Amsterdam or Tokyo and see plenty of families with small children walking around. Heck, you don't even have to go far to see how a city can be a great place for a family, just look at Boston.

Cities, when done correctly, are a better place to raise a family than the suburbs. And yet, when done incorrectly, are terrible.

If you're referring to Masayoshi Son he's Korean by ethnicity, Japanese by nationality.

> append-only-vec

> This is a pretty simple type, which is a vector that you can push into, but cannot modify the elements of. The data structure never moves an element once allocated, so you can push to the vec even while holding references to elements that have already been pushed.

This is the left-pad of rust. Reconsider your life decisions if you need to pull in an external dependency replicable with 30 lines of code:

    use std::vec::Vec;

    #[derive(Debug)]
    pub struct AppendOnlyVec<T> {
        inner: Vec<T>,
    }

    impl<T> AppendOnlyVec<T> {
        pub fn new() -> Self {
            Self {
                inner: vec![],
            }
        }

        pub fn push(&mut self, element: T) {
            self.inner.push(element);
        }

        pub fn clear(&mut self) {
            self.inner.clear();
        }
    }

> I think we should be taking a more “batteries included” approach to language and library design. The entire reason we’re in this mess is because we’ve decided it’s ok or maybe even preferable if stdlibs are rail thin, rendering base languages near-unusable.

Rust is one of the better languages when it comes to batteries included in it's stdlib. Even then, it's impossible to have a 100% coverage batteries included language because nearly everyone on planet earth has some different/unique use case for their code that doesn't fit a stdlib.

New person in sales joined my company a few months ago and on the very first introduction they handed me a vibe coded app and asked me to review it for them and put it into production. Haven't spoken to them since. They haven't proposed anything else since then either to engineering. You may think it harsh but they didn't even ask my name, what my specialization was, etc.

Engineering to these folks is a one-and-done endeavour. You just prompt and fire it off and ride off into the sunset, leaving the mess for someone else to clean up. They never see the work that goes on after the launch party. The midnight debugging sessions, the database migrations, the DNS failures, debates over which auth provider to use, the list goes on. Until the "citizen developer" is willing to engage in the less "sexy" part of software engineering, the world will move on without them as well.

how can you make such a bold claim so early with 0 evidence? A successful (and much more realistic outcome) could be to make a font that is financially infeasible for LLMs to parse but easy for humans.

They are in the business of justifying blame from the executives to the frontline. Executives keep or increase their bonuses, layoffs ensue. If you are the decision maker of a company, and you can pay a team a few hundred thousand a year using company money, netting a 10 million increase in your own salary, its a deal too good to pass up.

It's harder to leak multiple S3 keys than a single postgres key.

Anyone can re-create the hosting part of GitHub. This is not the difficult nor the expensive part.

It's the actions. The compute to run them isn't cheap at scale nor is the amount of auditing you need to do for security and compliance if you want enterprise/government customers.