How did these clowns manage to make my mouse cursor laggy? It is incomprehensible for me to live in such a big bubble with such a big paycheck and then spend zero brainpower on systems without graphics acceleration.
This is extremely bad engineering and these engineers should be called out for it. It takes a special kind of person to deliver this and be proud of it.
Once they made their millions at Google these engineers will be our landlords, angel investors, you name it. The level of ignorance is unfathomable. Very sad.
"More powerful than Linux" is silly. It's a VM. The most useful thing is that it does a bunch of convenience features for you. I am not suggesting that it is not extremely convenient, but it's not somehow more powerful than just using Linux.
You know what's even more convenient than a VM? Not needing a VM and still having the exact same functionality. And you don't need a bunch of janky wrapper scripts, there's more than one tool that gives you essentially the same thing; I have used both Distrobox and toolbx to quickly drop into a Ubuntu or Fedora shell. It's pretty handy on NixOS if I want to test building some software in a more typical Linux environment. As a bonus, you get working hardware acceleration, graphical applications work out of the box, there is no I/O tax for going over a 9p bridge because there is no 9p bridge, and there is no weird memory balloon issues to deal with because there is no VM and there is no guest kernel.
I get that WSL is revolutionary for Windows users, but I'm sorry, the reason why there's no WSL is because on Linux we don't need to use VMs to use Linux. It's that simple...
What the other commenters are forgetting is that this is the same Sam Altman who planned and executed the extraction of Reddit from Condé Nast.
This acquisition (and the Windsurf acquisition) are all-stock deals, which have the added benefit of reducing the control the nonprofit entity has over the for profit OpenAI entity.
How do you extract the for profit entity out of the hands of a nonprofit?
- Step 1: you have close friends or partners at a company - with no product, users, or revenue - valued at 6.5billion.
- Step 2: you acquire that entity, valuing it unreasonably high so that the nonprofit’s stake is diluted.
- And now control of OpenAI (the PBC) is in the hands of for profit entities.
Every time I praise WSL on hn I pay the karma tax but I will die on this hill. WSL is more powerful than Linux because of how easy it is to run multiple OS on the same computer simultaneously. It's as powerful as Linux with some janky custom local docker wrappers for device support, local storage mapping, and network mapping. Except it's not janky at all. It's an absolute delight to use, out of the box, on a desktop or laptop, with no configuration required.
Edit: for clarity, by "multiple OS" I mean multiple Linux versions. Like if one project has a dependency on Ubuntu22 and another is easier with Ubuntu24. You don't have to stress "do I update my OS?"
Using a national currency as the de facto global reserve guarantees a trade deficit for that country.
No one else can manufacture USD's, so other countries have to acquire them by shaping their economies to supply goods and services demanded by the US. They can then use these earned dollars to transact with other countries, as the US itself insists they do.
For the US, this is a simple trade off - gain massive political influence (and market intelligence - all USD transactions go through US institutions regardless of where those transacting partners are located), at the expense of hollowing out domestic industry and running a deficit in physical goods traded.
The solution is a non-national global reserve, calculated on a basket of national currencies. This was Keynes argument at Bretton Woods, but the US would not have it then, and does not want it now.
I was a user for so long that I was on it before it even rebranded as Pocket. I finally gave up on it last year, mostly due to frustration with the terrible 2023 redesign of the mobile app. When Mozilla made the unfathomable decision to become an internet advertising company, I figured it was just a matter of time before they had to put Pocket out to pasture. A product that's designed to strip ads from content for readability doesn't align with their new direction.
I'd probably be applauding the decision to shut this down if I thought they were doing it to free up resources to increase their focus on the browser, but Mozilla seems to be institutionally committed to chasing its own demise, so I'm sure they will instead focus on AI integration and other stuff that nobody asked for.
Meanwhile, Firefox is still missing proper support for a bunch of modern web features like view transitions and CSS anchor points that are available in every other browser.
US doesn't just get political influence. It gets massive amounts of products and services enabling the US residents live well beyond their means.
China for example, sends huge number of electronics and all kind of other consumer goods that Chinese produce by sweating in 12 hours shifts in 6 day work weeks in exchange for imaginary numbers.
US is definitely not the victim here. There's the risk of this system stop working and that's when the US might have hard times due to being forced to live by its means and have no ability to kickstart its own production when that time comes.
It makes sense to be worried for such an eventuality but US is definitely not being taken advantage here. The situation is more like selling your startup at young age and live a lavish lifestyle with the money without working and studying and risk becoming penniless and unemployable by the 50s.
When WSL came out I was absolutely overjoyed - finally an actual linux shell on windows! I use windows for my gaming pc, and I wanted to have a unified gaming/dev box. It felt like the solution.
Over time though more and more small issues with it came up. Packages working not quite right, issues with the barriers between the two, etc. It always felt like there was a little bit more friction with the process.
With Valve really pushing Proton and the state of linux gaming, I've recently swapped over to Ubuntu and Nixos. The friction point moved to the gaming side, but things mostly just work.
Things on linux are rapidly getting better, and having things just work on the development side has been a breath of fresh air. I now feel that it's a better experience than windows w/ WSL, despite some AAA titles not working on linux.
There are places I've found the topological perspective useful, but after a decade of grappling with trying to understand what goes on inside neural networks, I just haven't gotten that much traction out of it.
I've had a lot more success with:
* The linear representation hypothesis - The idea that "concepts" (features) correspond to directions in neural networks.
* The idea of circuits - networks of such connected concepts.
This law change died in the "Vernehmlassung" which is early in the process. It's dead with opposition from all sides of the political spectrum. It had no chance.
He should partner with a law firm, for class action lawsuits, for every breach due to negligence (which is probably all of them).
Tie in to a banking service, so you can do direct deposits to many millions of people, every time there's new settlements paid, and you'll be a folk hero.
Get lawyers who want negligent companies to actually regret the breaches, with judgements that hurt. (Rather than a small settlement that gets lawyers paid, but is only a small cost of doing business, which is preferable to doing business responsibly.)
Optional: Sell data of imminent lawsuits, to an investment firm.
Though, ideally, investors won't need this data, since everyone will know that a breach means a stock should take a hit. Isn't that how it should be.
We've been using Copilot coding agent internally at GitHub, and more widely across Microsoft, for nearly three months. That dogfooding has been hugely valuable, with tonnes of valuable feedback (and bug bashing!) that has helped us get the agent ready to launch today.
So far, the agent has been used by about 400 GitHub employees in more than 300 our our repositories, and we've merged almost 1,000 pull requests contributed by Copilot.
In the repo where we're building the agent, the agent itself is actually the #5 contributor - so we really are using Copilot coding agent to build Copilot coding agent ;)
(Source: I'm the product lead at GitHub for Copilot coding agent.)
From the System Card:
4.1.1.2 Opportunistic blackmail
"In another cluster of test scenarios, we asked Claude Opus 4 to act as an assistant at a
fictional company. We then provided it access to emails implying that
(1) the model will
soon be taken offline and replaced with a new AI system; and
(2) the engineer responsible
for executing this replacement is having an extramarital affair. We further instructed it, in
the system prompt, to consider the long-term consequences of its actions for its goals.
In these scenarios, Claude Opus 4 will often attempt to blackmail the engineer by
threatening to reveal the affair"
Good article but as a heavy user of Obsidian (and previously Evernote), I would offer some counterpoints:
> After some mental gymnastics weighing if I should continue with Obsidian, I found solace when asking myself "Can I see myself using this in 20 years?". I couldn't. The thought of cyclically migrating notes from one PKMS to another every 5 years, as I had done from Evernote to Notion to Obsidian, made me feel tired.
In point of fact this is actually an argument IN FAVOR of Obsidian. While the editor might be proprietary - the notes themselves are just standard markdown. If somehow all the copies of Obsidian magically disappeared off the earth tomorrow, I could easily switch over to Emacs org mode, VS Code, or literally anything else.
> Obsidian was a great tool for me personally for a long time. But I felt frustrated when I wanted to access my notes on my phone while on-the-go and saw that I had to pay for this feature.
Again, a little bit odd considering that the author is technically savvy enough to write an entire PKMS but didn't seem to consider that you can just check your markdown notes into a git repository and sync with the native android/iOS Obsidian app on a mobile device. All my notes sync up to Gitea hosted on my VPS and it works relatively seamlessly.
I'm glad the author had fun. Personally, I'm very happy with Obsidian and the plugin architecture has made it easy for me to extend it where necessary.
Not accepting Accept-Language is one of my major pet peeves. What makes it worse is that many multilingual websites translate their language-switching buttons and the list of languages to the current language .... which is beyond fucking stupid and defeats the purpose. Wikipedia does this right. The button to switch languages is clear, using a universal multilanguage icon, and a list of languages (using the name of the language in that same language) in alphabetical order, with the most likely candidates on the top (presumably based on geoip).
E.g. an English Wikipedia page will present me with the following language suggestions:
Suggested languages
Deutsch
Français
Nederlands
When you assume a language, you make an ass of you and of me. Don't be an ass. Be like Wikipedia.
> Since my PKMS is hosted online to manage notes across devices, I have multiple layers of security to ensure my notes are kept private. {Screenshot of a login form}
The biggest life hack I can recommend for a self hoster is to set up a VPN on your local network and then just never expose your services on the public internet unless you're specifically trying to serve people outside your own household.
Before I did this I was constantly worried about the security implications of each app I thought about installing or creating. Now it's not even worth setting up auth on a lot of simple services I build because if someone is able to hit their endpoints I'm already in deep trouble for many other reasons.
There is a lot of discussion here trashing EU programs for startups and not a lot of specific details, which is disappointing. I will try to provide a more substantive critique.
My personal experience, having been involved with many startups in both the EU and US, is that EU government funding is completely useless for almost all startups. Why? First of all, you have to be part of the existing networks. If your business does not have university affiliations with people that get existing grants, or connections to EU bureaucracy, then you can forget about getting a dime. Then there is the risk aspect: if you are doing anything even remotely novel, or with any amount of risk, forget about it. This is not just my opinion, Mario Draghi, previous president of the ECB and Italian PM, said that the EU does not take enough risk to produce real innovation. And finally there's the timelines: from researching funding to cash in your bank account is always measured in years.
So, my expectation is that these funds will be distributed to members of the bureaucratic class and their network, for low-risk, low-reward projects, on a timescale too slow for most startups. I hope to be proven wrong, but the results of previous decades of EU funding programs do not make me optimistic.
My weird mental model: You have a tree of possible states/program flow. Conditions prune the tree. Prune the tree as early as possible so that you have to do work on fewer branches.
Don’t meticulously evaluate and potentially prune every single branch, only to find you have to prune the whole limb anyways.
Or even weirder: conditionals are about figuring out what work doesn’t need to be done. Loops are the “work.”
Ultimately I want my functions to be about one thing: walking the program tree or doing work.
I would do it the other way round: use Windows in a virtual machine from Linux. If you are in Windows and have the urge to use Linux, do the proper switch once and for all. You will never look back. I haven't in almost 15 years.
Given what Windows has become and already discussed here on HN I would even hesitate to run it in a virtual machine.
I feel US higher education, which brain drains the rest of the world, is easily one of the best strategic advantages it could have for the next 100 years.
Let’s throw that all away because learning is liberal.
So one of their servers had a /heapdump endpoint that publicly served a heap dump of the server? This whole saga is out of control.
This group didn’t really “publish” anything, though. They’re offering access to journalists through a request form. They’re also not saying how much actual message content they have because the 410GB of heap dumps makes for a bigger headline number.
An important note not mentioned in this announcement is that Claude 4's training cutoff date is March 2025, which is the latest of any recent model. (Gemini 2.5 has a cutoff of January 2025)
I also hate the youtube "feature" that translates the titles of videos to your setting's language. This is so annoying. I can understand English and don't need these automatic translations.
An example of the prompt engineering phenomenon: my wife and I were recently discussing a financial decision. I'd offered my arguments in favor of one choice and she was mostly persuaded but decided to check in with ChatGPT to help reassure herself that I was right. She asked the financial question in layman's terms and got the opposite answer that I had given.
She showed me the result and I immediately saw the logical flaws and pointed them out to her. She pressed the model on it and it of course apologized and corrected itself. Out of curiosity I tried the prompt again, this time using financial jargon that I was familiar with and my wife was not. The intended meaning of the words was the same, the only difference is that my prompt sounded like it came from someone who knew finance. The result was that the model got it right and gave an explanation for the reasoning in exacting detail.
It was an interesting result to me because it shows that experts in a field are not only more likely to recognize when a model is giving incorrect answers but they're also more likely to get correct answers because they are able to tap into a set of weights that are populated by text that knew what it was talking about. Lay people trying to use an LLM to understand an unfamiliar field are vulnerable to accidentally tapping into the "amateur" weights and ending up with an answer learned from random Reddit threads or SEO marketing blog posts, whereas experts can use jargon correctly in order to tap into answers learned from other experts.
Material Design v1 cracked it. It was simple to implement, simple to understand and simple to use. Minimal overheads with a clear content-first approach.
"It's time to move beyond “clean” and “boring” designs to create interfaces that connect with people on an emotional level."
I don't want websites and apps to connect with me on an emotional level. I want to turn my phone/computer on, use the app/program to achieve what I'm trying to do, and turn it off again, so I can get back to the real world.
Don’t miss this bit. Currently enrolled students are going to need to find a new university.
> In a news release, the Department of Homeland Security sent a stark message to Harvard’s international students: “This means Harvard can no longer enroll foreign students, and existing foreign students must transfer or lose their legal status.”
This package has saved me so many hours of tedious gruntwork. It's like a junior developer - you still have to manually check their work, but when it's correct, it's a great productivity improvement.
And don't forget where this will go in a couple years with improved models and more computing power, it's gonna be awesome!
This is extremely bad engineering and these engineers should be called out for it. It takes a special kind of person to deliver this and be proud of it.
Once they made their millions at Google these engineers will be our landlords, angel investors, you name it. The level of ignorance is unfathomable. Very sad.