If AI makes you the 10x engineer, you may get the burnout 10 times earlier. You type less, but you work the same. Programmers are not paid as good typists, but because they can understand the problems they deal with. Typing time is the part when you do not need to think that hard. AI generates 200 lines in a few seconds, so you have to read and understand them almost instantly. If that fries your brain, you may consider making a few more breaks. It may be enough to become the 3x engineer.

Raycast seems to be a Github Feature or official Github Client, no? So I would say it is a self promotion, like when sites want you to install their app, not an ad like in "People who implemented sqlite also implemented oracle"

So much bullshit tech in the same product ...

For starters: Why does this need an append only blockchain? Once the vape is unlocked, there is no need to keep a record of the unlocking. It also seems like the thing needs wifi. And a camera. All of this while other vapes come without any of these features. Do they hope regulators will make other vapes illegal?

You can charge for access. It's just not open source then.

The GPL even allows you to take a fee that covers your expenses to provide the source to people who are entitled to get it (ie. people who got the binary from you). But that's like paying for postage stamps and not charging for access.

So if you want someone to pay, choose a license that requires that. Just don't call it open source.

Don't shove AI in everything.

The companies are pushing it, because they want money and data. If a small site like Slashdot would have to adopt age verification, they would implement the age-verification service provided by some of the companies, that the company is offering for free for some reason. The company then gets user data and the association of the user to Slashdot. They are just getting more user data, including verified IDs. Advertisers like when users are not only verified to be a human but also to be a single human, and best even have a verified real name in the data.

I just wanted to say, if you look at someone who bought Bitcoin, let's say in 2010, then they are still quite a bit in the profit zone.

They are also so dishonest. "No problem keep your initscripts, systemd just calls them" they said, then systemd said "Legacy initscripts won't be called anymore". Yeah, thanks for nothing.

The question is how long the supply chains are. Python is moderate, nodejs is horrible. With languages like C++ you usually have like 1-5 libraries with no dependencies themselves and you need to download and install them yourself, possibly doing manual upgrades instead of always updating to the latest version. That is a much more stable model than "recursively fetch my 20 dependencies and get 500 packages I've never heard of"

That's not completely true anymore, but in particular not that relevant if it can do a post-hoc explanation.

Why you're right: Generating something does not allow the LLM to explain it without making up a new explanation (with caveats)
Why you're wrong: Thinking models first generate a thinking trace and then answer based on this. The thinking trace is a good explanation for the answer that comes after
Why it doesn't matter: If a second run can explain the code (in a second run), it is not important what the reasons for its creation were beforehand.

And the caveat: A LLM has a lot more hidden state than what's represented in the tokens at the output layer. It is not clear if the hidden state of some LLM is not enough to have a post-hoc explanation (in the same session) that is based on the reasons why it generated the code the way it did. When generating text you only see the sampling of a probability distribution in the last layer, but the latents are much larger.

What if nobody implemented it? Do you think the politicians go around and shut down computers? If nobody supports this crap, then they need to make a new plan. It's open source, so they cannot get the programmer to do something. They can indeed make it illegal to run for the consumer, but if you suppose there is no alternative that implements this bullshit, do you think they will enforce the law?

Installer level disabling of the installation of systemd, please.

Systemd sucks with its philosophy "If we break things, then distributions are at fault not disabling them". Yeah, maybe a distribution may be able to fix the "A stop job is running (5h / unlimited)" issue to get the computer to shut down even when it runs systemd, but why isn't it limited out of the box? Classic sysv gives processes that were not stopped by their initscript further 5 seconds between SIGTERM and SIGKILL. That's it. Shutdown guaranteed.

Another issue is the preconfigured Google DNS. Who thought ever that would be a good idea? But ... distributions can disable it, when users don't like it!

I do not like Bitcoin (and the concept of proof of work), but saying it is history is bullshit if you look at the price. People can still make good money with it.

OpenAI stops producing videos, video company stops investing. Why should this be surprising?

Apple probably never promised that it would be anonymous to Apple, only that average joe won't get the information.

We already had this news like 2 or 3 weeks ago.