You can still get old Mac minis for less than that, which have more memory and can run Debian. Probably best performance per dollar hardware available on the used market
You seem to fall into the same set of criticisms as everyone who’s bearish about ai. It’s somehow so powerful that we can’t handle the ramifications. Meanwhile, it’s a waste of money and doesn’t do anything. You have to pick a criticism and stick with it. Otherwise, it’s just angst-driven noise.
All these bills about age verification have nothing to do with protecting kids. This is just an easier pill for folks that aren’t privacy minded to follow. In the end, all your online activity and offline activity (flock cameras) will be tracked, because it gives our politicians and national security apparatus the type of power they crave.
This like many other attempts at this type of thing don’t understand the in distribution vs out of distribution aspect of ml models. Using something the model has the most training on will have the best outcomes. A new handcrafted programming language will perform significantly worse than a poorly thought out programming language that is widely used. Everything is going to revert to the mean as llms continue to progress.
Not clear to me how cursor can remain relevant in the era of agent coding. The main things I care about now are navigation and quick editing. Vanilla VS Code or vim with some extensions does the job for me now.
You’re absolutely right. The labor market is still quite strong. All the doom and gloom from places like HN is coming from the many layoff announcements and fear of AI.
At the end of the day, he built something people want. That’s what really matters. OpenAI and Anthropic could not build it because of the security issues you point out. But people are using it and there is a need for it. Good on him for recognizing this and giving people what they want. We’re all adults and the users will be responsible for whatever issues they run into because of the lack of security around this project.
I think in the end he’ll mostly be remembered for his support of trump and his abhorrent political views. He had a great comic strip that reasonated with people. He also wrote some interesting books. It’s a good reminder that your accomplishments can easily be wiped out by bad choices.
I feel like cli agents are the main benefit of going back to Linux. It’s such a joy to have all the solutions to customizations and fixes I want completely automated, using an agent that can control anything I permit and understands my OS completely.
What a bizarre comment. It’s not clear if you are upset about you not being able to dig deep into these subjects or what your deal is. The post is about things this guy thought were interesting this year. Why the hostility?
I really wish this guy would write a blog post about LLMs. He had a great section about his lack of use of them at the end of this post. Like 90% of my code is now written by LLMs, but this guy is incredibly smart and is working on very challenging problems that require novel solutions. Would love to know more why the tech isn’t good enough and which approaches he’s tried.
Sex is different from work. And you don’t have unilateral control over who anyone else has sex with. But, you do have say over how the government deals with the AI revolution.
I don’t think they serve the same purpose. Most of the instructions I have for an agent won’t apply to a human. It’s mostly around the the requirements to bootstrap the project vs what I’d ask for a human to accept their pull request.
I like this. I was using README.md, like many folks, which wasn’t bad. The main issue is the agent is typically updating it as well while building out the prototype.
I’m not sure if 10x engineers exist, but I do know 0.1x engineers exist. Being on a team with them makes a typical engineer seem like they’re driving 10x the expected impact.
Back in 2016 when I was at google, I started on a team that was all golang. I was working on my first project, building out a new service and got many readability approvals from Ian. One time I got an an approval with some follow up requests, which I somehow didn’t notice and landed the change. He got back to me asking me to follow up. I didn’t realize he was one of the core Golang developers! He was super gracious, even though he didn’t need to be and I’ll always remember that. It’s really something that he invested so much time into seeing how the language was actually used and identifying core problems. Very admirable.
There is an ongoing genocide in Gaza and genocidal language is commonplace in Zionist discourse. If there are cases of hate speech on the pro Palestinian side, they pale in comparison to speech from the other side.
Regardless we shouldn’t be rounding up and imprisoning folks if they disagree with your politics. This is what is getting lost in this specific case.
It was actually pro-facism. This isn’t the first time a movie was based on a book, where the movie was a dark comedy, by simply representing the argument in the book. Similar thing happened with Dr Strangelove, which was based on some crummy Cold War book.
It’s been operating safely in each market they’re in. The AI keeps getting better. They have no competition (please don’t bother mentioning Tesla vapor ware). Path to high growth seems pretty sure at this point.
This is a silly charade. To actually cut the debt, it requires congress. Not just saving a couple millions dollars and posting on twitter about it. We’ll need to cut Medicare, Medicaid, social security, the military, etc. All the popular stuff that was never easy and will never be easy to cut. We’re not spending 2 trillion per year on condoms.
Seems like a strange thing to be building around. Lots of money has gotten into VR and it has been around for a while now. It has never gotten out of the geek niche and likely won’t. It’s the only way I play games nowadays and wish it would gain wider traction. But, I’m very pessimistic about normies buying these headsets and watching these VR videos.
This paper’s method might look slick on a first pass—some new architecture tweak or loss function that nudges benchmark metrics upward. But as an ML engineer, I’m more interested in whether this scales cleanly in practice. Are we looking at training times that balloon due to yet another complex attention variant? Any details on how it handles real-world noise or distribution shifts beyond toy datasets? The authors mention improved performance on a few benchmarks, but I’d like to see some results on how easily the approach slots into existing pipelines or whether it requires a bespoke training setup that no one’s going to touch six months from now. Ultimately, the big question is: does this push the needle enough that I’d integrate it into my next production model, or is this another incremental paper that’ll never leave the lab?
“It’s refreshing to see a founder’s honest postmortem that doesn’t just blame ‘bad timing’ or ‘market conditions.’ Instead, he’s ripping off the Band-Aid and laying bare the pain points—no hype, no quiet pivot into mediocrity. It’s the kind of retrospective that should be required reading before your next ‘growth hack’ session: get the fundamentals right or prepare to pen your own ‘closing up shop’ blog post.”