falcor84

Concrete alone isn't enough, you also need to have it be enclosed in a Faraday Cage.

> true but misleading

It seems to me that for many newspapers the bar is now significantly lower, at something like "not quite entirely untrue"

That was my feeling when I first heard about Lisp Machines. It's unfortunate that I never got to see or use one in person.

That's on me; I set the bar too low. When I asked "How", I was hoping for someone to literally write up an alternative and better way of phrasing this, which would still work as part of a corporate press release.

Well, to be fair, they didn't say "redesign" anywhere, and Webflow is the name of the company, so I don't find it too strange that the "evolution" of a company involves layoffs. Maybe I'm just too jaded to these "An update on ..." announcements, but this really didn't strike me as bad. At the very least, they announced the layoffs in the very first paragraph, rather than beating around the bush.

Why? How would you as a leader would announce this in a way that better reduces bad feelings?

And there's at least one more level of inception at the data center level, where they use AI to optimize power usage (particularly by predictively controlling cooling, and adaptively rescheduling tasks).

I actually had this happen to me the other day - I asked Claude Code to create an arbitrary example website for me, and it built one about lighthouses.

I would argue that it's more important than ever for effective engineering change management. With how easy it is have AI make changes to a system, I'm reminded of this tweet from DevOps Borat:

> To make error is human. To propagate error to all server in automatic way is #devops.

I'm not sure if YAML specifically is the best format, but we absolutely need something that offers a single readable source of truth for people to manage and reason about changes.

I wonder what the endgame here is, for our AIs to best reach that "I know kung-fu" capability. If we can evolve an optimized set of external skills, we can then distill them all into a model's weights, right? And if there is some benefit to being able to pick-and-match, would LoRA be the best approach? Or is there some real benefit to the free text approach?

I agree that AI sycophancy is an issue, but as is typical in these studies, it's missing a realistic human baseline. They say "instead of reaching out to other people", but they don't appear to have done that themselves, and instead compared the AI responses to curated datasets of relatively good advice. Seeing how many people around me (and my past self included) stay in bad situations for years, I'd want to know how AI does compared to actually talking to one's friends.

Who's trying to avoid talking about tobacco? TFA is titled "Scientists solve 200-year-old puzzle of how tobacco plants make nicotine", and my sense is that the thread is very much about which of the effects of smoking are explained by nicotine, and which are better explained by other factors of smoking tobacco.

The HN guidelines say "Please respond to the strongest plausible interpretation of what someone says". The way I understand it is that different people on the thread often have different ways of thinking about the topic, and we shouldn't dismiss something because it's not what "we" were talking about. In this case, it was obvious to you that the parent was talking about smoking tobacco, right? So you can either engage with it, or not, but there's no need to reject someone's comment for not adhering to what you decided is the topic.

Well, I did, and to save others the time, the most relevant resource I found appears to be the book "Smuggler Nation: How Illicit Trade Made America” (2013) by Peter Andreas

What? How is someone born in 1947 relevant to ip theft in the 19th century?

I would appreciate some reading pointers about this.

Seeing how it's hard to identify sarcasm on the internet, I'll just clarify that it of course is the same person - https://en.wikipedia.org/wiki/Marshall_Brain

And I'll also link to the HN thread following his death a couple of years ago - https://news.ycombinator.com/item?id=42228759

> As a human, I have many stupid and wrong ideas all day long - most of those don't bubble up to my conscious awareness

I'm sorry to nitpick, but isn't an unconscious idea an oxymoron?

It is most likely significantly driven by both

Well, it really depends on your definitions, but I'll probably put the biggest theft in history on European imperialism in the 14-19th centuries, seizing unfathomable amounts of land, resources and slave labor from other civilizations.

If anything, I would argue that the whole Intellectual Property bubble is The Emperor's New Clothes. It never made real sense to me to treat ideas as property, and I for one would absolutely prefer to live in a future society where it's possible to just copy a car.

> how would one enforce a copyright claim if the creator is amorphous and all-knowing?

I love it! There's a great seed here for a short story about God being sued by a peer of his for copying some of her physical constants and not putting a proper copyright notice about it in our universe.

I think you're right, the ease in which AI can do task that we previously considered unique to human creativity does force us to further rethink and acknowledge how creativity is in a large part about "remixing" prior works, although of course we've had discourse about this for at least as early as Richard Simon's 1678 "Critical History of the Old Testament", which identified it as being a remix of earlier sources [0].

[0] https://archive.org/details/hisyo00simo/page/n1/mode/2up

Ironically this phrase was said by Jafar in Disney's 2019 live action remake of Aladdin, but wasn't part of the original 1992 version. And I personally would argue that this corporate remake is a worse creative "theft" than what random people are doing with GenAI.

A software system can be fully interoperable without any part of it (other than the interface) being open. In my definition, an Open Shell software solution is one in which a company makes it clear that they're focusing on building and selling (or making money from hosting/supporting) only the "Core" functionality, while exposing all of the higher level functionality as open source plugins and fostering a community to make it as easy as possible for others to improve these and maintain them over time. And preferably, the company would also pledge never to offer its own closed source functionality outside its definition of the core.

I'll just put the Willy Wonka tunnel clip here - https://youtu.be/Xkg7dp1QY9k

I'm wondering why this was posted now, seeing how the latest actual code commit there is from 2 years ago, and the documentation section of the readme is literally a blank header:

> ## Documentation

> That reminds me of the short scifi/horror story "Valuable Humans in Transit" ...

I think you're probably referring to "Lena", from the same story collection - https://qntm.org/mmacevedo

Just extracting the awesome result (with minor simplification):

> To evaluate c = sqrt(a^2+b^2), assuming a is the larger number (if it is not, swap them):

> Compute c = max(a, 0.9a+0.5b)

> Done! That’s the result. This is an estimation and it does come with an error, but the error is at worst 3 %, and on average it is 1.5 %.