0x20cowboy

Or visual source safe

I agree with the spirit of this, but I’d go one more step with the example:

export function clamp(value: number, min: number, max: number): number { return }

That is just adding an extra jump and entry on the callstack where you could just have done:

Math.min(Math.max(value, min), max);

Where you need it.

(The example used was probably just for illustration though)

> LLMs are PhD-level reasoners in math and science, yet they fail at children's puzzles. How is this possible?

Because they are not.

Pattern matching questions on a contrived test is not the same thing as understanding or reasoning.

It’s the same reason why most of the people who pass your leetcode tests don’t actually know how to build anything real. They are taught to the test not taught to reality.

If the author is around here, I think there is an error in the Equations section. In the Convex Circle equation I think it has an extra set of parens. sqrt(1-(1-x)^2) I believe as it is it’s just linear.

My guess is because some people (including myself) are turning off auto correct.

It's crazy, I've been trying to hire an architect to build a house for me, but I don't trust them (they're all liars) so I asked them to draw up blueprints in front of me and describe everything they are doing in great detail. I need to understand what they are doing. This building is important and I don't want a bad hire. They should be able to do this in 10-15 minutes. If they knew what they were doing, this request wouldn't be an issue.

Then I was trying to get someone to do my taxes. So I've been asking every applicant to do my taxes from last year so I can see if they really know what they are doing. I mean sure, they've done taxes for years, but these are my taxes. I've even tried giving them math puzzles around asset depreciation, but people just keep hanging up the phone.

And then, I wanted to add a wing added to my house and I've been trying to get these entitled contractors to come build a shed in my backyard so I can see if they really know how to use nail guns. I've heard people are just big'ol lairs lately, and I need to see how they work. I mean, sure these guys have built many houses in the past, but we have high standards here and only hire A players.

I've only been getting horrible candidates! No one is worth hiring! There is a huge shortage of qualified people.

If only there was a way to fix this.

> I won’t hire anyone that can’t pass a live coding round.

Excellent - please be sure to mention this in the job description so I can know to never apply to where you work.

If nobody has a job, I am unsure who will be left to pay the subscription fee.

“Hardening”[1] apologies.

[1]: https://developer.apple.com/documentation/xcode/configuring-...

I have a smaller tracker app that has similar problems on mac.

When you use the keyboard for navigation , holding down a key brings up the accent dialog thing. It also starts doing that “bllllllllllpppppp” noise like you’re doing something wrong.

If you try to add sandbox and harding the SDL timer seems to go nuts, loops too fast and crashes often.

Also, a lot of old tracker code uses 32 bit tricks that arm64 just doesn’t like and there doesn’t seem to be a nice way to force a 32 bit mode.

If you dont build with sandbox or harding it behaves better, but then of course you can’t validate or distribute it.

LLM are a compressed version of their training dataset with a text based interactive search function.

> It's really, really, really good at confidently jumping to hasty conclusions and confirmation bias.

Kind of like entry level software engineers.

I am kidding, I believe the market has more to do with tax changes than AI. I just couldn't pass up the joke.

I hate webp. Not for any legitimate technical reason like, but I often just want to download an image from the web for an image board or drop it in a diagram or ppt or for a joke and nothing works with that format. Nothing. Google image search is useless because of it.

Cmd+shift+4 is now the only way to grab an image out of a browser. Which is annoying.

It has made my life needlessly more complicated. I wish it would go away.

Maybe if browsers auto converted when you dragged ann image out of the browser window I wouldn’t care, but when I see webp… I hate.

Always the bridesmaid, never the bride.

I did a stint in Devops and I found every models to be like this for all of the infra-as-code languages. Anything yaml based was especially bad.

Even Amazon’s own offering completely made things up about Amazon’s own formats.

I’d be curious as to why that is. It seems like there would be enough training data, and for Amazon in particular it seems like they could make a validation tool the model could use.

I’ve written a pure C Mac desktop app in 2024 that called the objective-c bindings directly. It is using a modified version of this library https://github.com/ColleagueRiley/RGFW

While I don’t think that library does iOS specifically, you can have a look at the code to see how you can call objective-c from c. Spoiler: doing so kind of sucks, and also the library code isn’t the most elegant thing I’ve seen.

This browser extension doesn’t have ambient music yet, but various noise types: https://addons.mozilla.org/en-US/firefox/addon/noice-noise/ (there is a Chromium version too). There might be an internet radio thing you could run from VLC. There might be something on the horizon with that plugin.

100%. I am not a fan of AI generating code in a professional context (for a number of reasons), but for side projects and as a learning tool I’ve become less yells-at-clouds about it.

Using AI tools in a professional code base, currently, seems a bit dangerous to me. However I have changed my mind on using it for vibe coding.

I used to type in program source code from magazines and had no idea what I was doing until something broke then I had to fix it. If I am honest, that was how I learned how to code.

AI will either teach that kind of thing to the new generation, or coding will become irrelevant. Either way, I think that’s good.

But I still don’t want my bank or airplane guidance software using it.

I love doing all aspects of building software. However, I’ve noticed when I am feeling lazy I’ll just copy pasta a stack trace into an LLM and just trust what is says is wrong. I won’t even read the stack trace.

I only tend to do that when I am tired or annoyed, but when I do it I can feel myself getting dumber. And it’s a weirdly satisfying feeling.

I just need a chair that doubles as a toilet and I’ll be all set.

Back in the day, the internet was an escape from the TV.

It allowed access to information, alternative views (real ones not insane made up ones), no ads, and an escape from having to hear a dominant narrative.

Now the internet has become the new TV. A lot of younger people I see are shunning it.

That makes my old hacker heart smile.

I am old and I can easily code / design all day. 10-16 hours if it’s something I am in to. It’s dealing with people / social issues that exhaust me.

This is just how junior and intermediate devs behave. It’s like a goth phase or something.

It goes along with being into BJJ, chess, vim, keto, linters, and “the dominance hierarchy”.

It’s annoying, but most everyone went through it. If you didn’t know better, how could they?

I’ve been looking for, (and I’d even pay for) an os9 platinum theme that worked in 2025 (wayland or newer GTK).

It matches my experience as well.

I find the models very useful to chat about library documentation or high level algorithm concepts, but I find the code it generates to be… I don’t know how else to say it… really bad and often out of context.

I know developers who blindly follow the hype and use them to generate production code. That scares the poop emoji out of me, and the code reads like an asset flipped 3D game.

Unpopular opinion: it would be easier to replace the C suite people with reasoning models than it would be to replace engineering.

VCs would probably like that as well.

Maybe pitch that to CEOs as a cost saving meausre.

The faster leetcode interviews can be completely broken to the point they are abandoned the better.

I agree, but limited isn’t the same as opinionated. In these types of usage, people seem to think they are opinionated because they are right.

I think it’s misunderstanding the angst of knowledgeable people’s frustration with the world, by trying to emulate the frustration without having the actual knowledge or experience.

I find most often people who are opinionated have very little knowledge on the subject. They are opinionated in an effort to try to keep discussions in a realm they feel they have some sort of expertise, but by definition their limited view makes them look silly to those with deeper understanding.

I apologise for being off topic, but I really want “an opinionated X” to stop having some kind of positive connotation to junior devs.

> I'm considering that I should set up some kind of code system with my family for this. As in, if I ever end up in a situation where I need help, I'll use a particular code phrase.

You absolutely should. And include something for videos (like FaceTime calls). Especially if these members of your family are boomers+