cryptoz

The pizza-bitcoin story is about $1B, not a measly $100,000, FWIW. Assuming you're referring to the 10k BTC for 2 pizzas story, anyway.

In 1999 when I was about 12, I memorized 120 digits of pi with little issue. I had some extra motivation: once I got to 20-30 digits, a classmate bet me I couldn't get past 100, and he'd give me $1 for every digit I memorized past 100. Never saw my $20 :(. Pretty sure it was a well-executed nerd-snipe.

Having not practiced at all since about 2001 or so, I can still get to about 30 pretty easily. I didn't use any techniques specifically, I just kept practicing as far as I remember.

I treasure my memories of being on subways and busses and getting lost in thought while staring out the windows, yeah. I mean, people have done this for basically all of human history (okay walking or something, not on a train, but, yeah). It's not a bad thing to have some time with just your senses existing in the world.

What you've described is just a new benchmark, though. It'll be called CarParkBench, various embodied LLMs will then be run against that benchmark, and some will score better than others.

I do see where you're going, but that's already what's happening: we have so many different benchmarks because there's no real single way to test for general intelligence.

Also, it takes a human probably at least a decade of world experience, growth, learning, etc, to pass your benchmark. I'm quite confident that it will be very soon that an embodied LLM will pass your new benchmark, much sooner than a human would take if born today.

I'm not familiar with MyselfGPT (and didn't see an obvious result in a quick search), so, I guess it probably cannot use MyselfGPT. If you have a link, I'll check it out and then we'll see!

> Should be "You're thinking".

This should be "This should be 'You're thinking'." don't you think? Why bother correcting someone's grammar with a sentence fragment? You're just trading one mistake for another. I'm hoping someone finds a grammar error in my post, because continuing this would be hilarious.

The linked page implies there is no linux support, I wonder why. It's there in the docs if you hunt for it.

FWIW, I used to love this phrase but over recent years have come to understand it is quite damaging. We live in a society where evil frequently hides behind a ‘stupid’ label, and people bring this quote up to defend or soften actions that are indeed done out of specific malicious intent.

Still can't make a Gmail label named 'Buzz'.

Yes, of course, it seems that type of ID was specifically rejected in the story above. Only drivers allowed to sign up. Non-drivers rejected because they don’t drive.

I'm reminded of the Alameda Weehawken burrito tunnel:

https://idlewords.com/2007/04/the_alameda_weehawken_burrito_...

Really cool stuff and definitely going to try it; I’m also finding it wild that Google put effort into adding ums and erms into their text to speech model a while back. AI puts it in, AI helps take it out.

It's not even a throwback. That U2 album still shows up if I accidentally open Apple Music. I haaaaate it. I disliked U2 before any of this, but now I have absolutely sworn to never ever listen to any of their music.

Google did the same thing with Transformers 2 I think. It still shows up as Purchased for me even though I absolutely did not purchase that. Good way to ensure I never ever watch any Transformers movie!

See also a story about an implementation from Max Levchin: https://max.levch.in/post/724289457144070144/shamir-secret-s...

Some wild irony is they once forcefully removed purchased copies of 1984 from Kindles while people were reading it.

“If McDonald’s offered three free Big Macs for a DNA sample, there would be lines around the block.” - Bruce

I have no idea about the eye thing taking off. But I think your comment is very HN and a bit out-of-touch with regular people. What "you're seeing" is a bubble and not representative of the general population. The eye thing is a slow frog boil and it will be commonplace before you can blink.

This is the approach I take with code edits to existing files at Code+=AI; I wrote a blog post with a simple example of AST modification to illustrate: https://codeplusequalsai.com/static/blog/prompting_llms_to_m...

Agent instruction files are code, though. And none of this is really workstation-specific, it is codebase-specific. Should each developer keep a nearly identical copy of CLAUDE.md? The instructions really aren't for a developer, they are for an LLM agent. In most cases (I'd imagine, anyway) the agentic instruction files must be in source control for them to even provide much value.

Video of the booster landing: https://xcancel.com/JeffBezos/status/2045874068763632017

AST-based code edits from LLMs: https://codeplusequalsai.com

It's an LLM-webapp-builder, sure, but different from the rest! I have the LLM write python code when it needs to modify an HTML file for example (it'll use beautifulsoup; then I run the code: it parses the source into a data structure, modifies the data structure, and then outputs the resulting html).

It's also a marketplace where you can publish your llm-powered webapp, and earn $ on the token margins (I charge 2x token rates) when people use your site.

Related, I'm still upset at the lies told by landlords regarding phone number privacy in buzz-in intercoms. I've been told multiple times at multiple apartment buildings, "don't worry, while the system will call your phone when someone taps your entry code, your phone number won't be revealed". And then you sign the lease, get a delivery from Instacart in your new place, and find that your 'private' number is blasted out loud, heard a whole city block away, in a loud-ass DTMF tone sequence.

BS.

AST-based code modifications from LLMs: https://codeplusequalsai.com/

I'm interested in the idea that LLMs writing raw code and doing line-or-diff replacements will not be the future, but that having the LLMs modify the structure of the code may end up being the best.

Also, I think that building LLM-powered webapps should earn the dev per token call; so I've built a margin into token costs where the end user is charged 2x the provider's token costs, and then I get 20% of the remaining and the dev gets 80%.

The barometer data is for sure noisy, and must be cleaned and quality controlled. But that is possible to do, has been for 10 years now (there are published papers and demo apps that can do it). For one, rate of change of atmospheric pressure is pretty much the same inside as out, your main challenge for the raw value to be correct is user elevation. That can be corrected in quality control as well.

Plenty of work has been done on this front, and it can be demonstrated that you can assimilate the smartphone pressures into weather models and get some good results. It is hard, of course, and I’m not sure personally how much better the forecasts get. But it’s absolutely possible.

> Why do you collect any data??

There are like, billions of internet-connected barometers in the world that are not used in weather models. I don’t know if Acme has any of that in mind, but there is plenty of good reason for a weather app to collect data from phones. I know @counters may disagree with me, but I believe there are opportunities to improve short term forecast accuracy using data collected from phones.

Also, pretty much every day, all the apps and all the sites will tell me the incorrect current conditions at my location, much less the forecast. It’s 2026 damnit. Why doesn’t my phone know what the weather is outside right now?

I haven’t got the app yet, but I plan on it (gotta upgrade iOS first I think). Acme seems to have a lot of ideas I agree with, so, definitely following this.

One more thing. Weather apps have not been “solved”. Not even close. They all suck, there’s billions in untapped opportunity, and a stale existing market of bad solutions. People die all the time from severe weather. There is so much more work to be done in forecast accuracy and communication.

They sold their last weather app to Apple for like, tens of millions or something. These aren’t some random Apple employees.

Also, it seems a common misunderstanding about some weather apps: yes, most of them just package free data and steal your privacy, but some are really much more than a “weather app”. Some are attempts at building next-generation weather forecast models, which if successful are of course worth billions.

I’ve spent a lot of time building innovative weather apps, most of my career actually. And it’s always shocking to me when people say I’m wasting time or wasting my life or look at me like, “really? You’re dedicating your life to weather apps?!”

No dawg, I’m trying to improve short term forecasts to save life and property from severe events at scale!

I’m not sure what the Acme end goal is, but surely this isn’t just a “weather app”.

Super cool idea, but I don't seem to be able to type in the text box. Tried Chrome and Firefox, same issue. Something taking focus from the text box as soon as I input anything?

Even more than that, I've seen a lot of people confuse 4 and 4o, probably because 4o sounds like a shorthand for 4.0 which would be the same thing as 4.

"No Prompting"? Right after they show that the user initiates everything with a prompt? Confusing.

Also what is twin.so if it's not workos.com? The login with google says it will redirect me to workos.com, which I went to and seems like a totally different product.

The title is clickbait though, he admits near the end it is not in fact a perfect replication. I could feel this of course, long before even starting to watch it. Still, upsetting because otherwise it’s an entertaining video.

Been writing software for like 20 years now and I love it. I am also a fan of AI-assisted coding, but I only just started using Cursor. Gosh I do not like it at all for a simple reason: since I didn't write the code, in order to understand it I have to read it. But gaining understanding that way takes longer than writing it myself does.

When you write the code, you understand it. When you read the code produced by an agent, you may eventually feel like you understand it, but it's not at the same deep level as if your own brain created it.

I'll keep using new tools, I'll keep writing my own code too. Just venting my frustrations with agentic coding because it's only going to get worse.