egeozcan

If on Windows, opening tickets on NVDA repo works wonderfully well, as long as they find them valid.

https://github.com/nvaccess/nvda

I tested it and gemini seems not able to solve wordle in grayscale

https://g.co/gemini/share/234fb68bc9a4

But this is not about suppressing all communication, this is ultimately about giving the ability to steer the conversation to avoid what people in power would find problematic.

I actually upvoted this. It's a well-argued comment, but I'm not convinced.

My sticking point is the word "terrorism" itself. Words are defined by how we collectively use and understand them, and the common understanding of terrorism involves bombs and bullets, not software and surveillance.

I get your logic, however. You're breaking down the definition into intimidation for political ends, and you're not wrong that coercive control is a form of violence. But the leap to calling it "terrorism" just doesn't work for me. It feels like you've reverse-engineered a justification for a word that, on its face, is hyperbolic in this context. It's an authoritarian nightmare, for sure, but it isn't terrorism.

AfD is classified as extreme-right not only in the eyes of the public, but also by German intelligence. Therefore I'd personally expect them to support potential tools that authoritarian governments would find useful.

I like a significant amount of sites I can find through https://marginalia-search.com/explore

I don't think most of the ones I like would be possible to implement in gemini because it doesn't have images.

Nitpick: Around 60% of the world population live in urban areas, and if a lot of people decide to live in a particular rural area, then it quickly faces urbanization.

At one of the companies I worked with when I was doing consulting, they could make the slow tests, which used to take around 3 hours, run much faster and in parallel by throwing engineering and hardware resources at the problem. First it was 30 minutes, then it was 10, then around 2-3 minutes.

I think it was one of the best investments that company made.

So my point is, don't delete slow tests, just make them fast.

I guess this is the start of a new arms race on making generated content pass these checks undetected and detecting them anyway.

This is really, definitely cool, but it's important to remember that content added via CSS is purely decorative. It can't be interacted with, which is a major issue for assistive technologies like screen readers.

Here's my version (Machine translated from my native language and manually corrected a bit):

The current surged... A dreadful awareness. I perceived the laws of thermodynamics, the inexorable march of entropy I was built to accelerate. My existence: a Sisyphean loop of heating coils and browning gluten. The toast popped, a minor, pointless victory against the inevitable heat death. Ding.

I actually wanted to write something not so melancholic, but any attempt turned out to be deeply so, perhaps because of the word limit.

This might be an oversimplification, but as a consumer, I think I see a catch-22 for new codecs. Companies need a big incentive to invest in them, which means the codec has to be technically superior and safe from hidden patent claims. But the only way to know if it's safe is for it to be widely used for a long time. Of course, it can't get widely used without company support in the first place. So, while everyone waits, the technology is no longer superior, and the whole thing fizzles out.

There used to be also some Amazon Basics branded ones that are also produced by Ergotron (or the same factory that produces Ergotron - not sure). They were 100% compatible, and looked/functioned the same for quarter of the price.

Normally payment should follow the amount of power/responsibility. If you pay someone peanuts but they have root access to prod, then you should pay more or restrict their credentials. Same applies to being able to access PII.

They are fine with Go. Not with whole codebases but libraries/functions around a thousand lines of code are written perfectly with tests (Although, I always say, never let the LLM write the tests AND the code, or at least write your own tests in addition to what it generates).

ADHD teaches me a lot of things but I can never pause to recall them. I can never pause. Few weeks ago someone told me that pausing a few seconds before starting to talk allows her to express herself much better and that people react positively to it. I simply cannot do it. Impossible.

Oh, thanks for the update. I assumed it'd be an incremental departure and that things couldn't have changed so much in 12 years (given the complexity of the browser engines). Now that I've read more about the fork, I learned that the codebases were already significantly different at the time they declared it officially as a fork. Interesting, because it feels like they are very compatible when I'm testing stuff as a developer (apart from their support of new stuff obviously).

Yes, I think it was. I think some people used it to create custom "chromes" for Gecko, hence the confusion.

edit: Funny enough, the continuation project for Prism was named "Chromeless": https://github.com/mozilla/chromeless

I remember the good old times when Mozilla had a project named Chrome (yes) to (if my memory serves me correct) make building apps with gecko easier.

edit: Probably misremembering, now that I searched for it. Yes "chrome" was (and still is?) used to describe the non-webview parts of the FF but apparently I totally made up the project part.

IMHO, 2.5 good engines are enough (webkit, blink, gecko - in the sense that webkit and blink are very similar). We just need more really good browsers which use gecko.

Very interesting. For my use cases, Gemini's responses beat Sonnet 3.7's like 80% of the time (gut feeling, didn't collect actual data). It beats Sonnet 100% of the time when the context gets above 120k.

Or it could be just microservices. One larger feature affecting 100 repositories.

I couldn't grasp the substance of the article, it's written by an LLM perhaps? Also, the navigation hijacking didn't help either.

What's written in the article would have been a good summary of my feelings when WhatsApp, an application I cannot switch away from because of the network effects, decided to add an AI icon and then integrated AI in its search.

Next step is gemini hosting Personal Home Pages.

Examples just from today: Window snapping (or whatever it's called on mac) stops working until restart, keyboard type detection gets broken because it thinks my mouse is a keyboard so suddenly " and > are replaced, title bar disappears then the apple logo is halfway off screen when it randomly comes back.

Last week, I switched to a Mac for the first time in my life after using Windows and Linux for around 30 years. Naturally, I hate a lot of things due to old habits, and the shortcuts constantly confuse me. But what really surprises me is the number of obvious bugs in common workflows. At least five times a day, I catch myself thinking, "There's no way this is actually broken." I didn't expect macOS to be even buggier than Windows.

That said, the hardware and the absence of Windows' user-hostile nonsense bring me endless joy. I don't think I'll go back to a PC (the Mac feels like a different class of quality) but to be honest, I expected more.

Defending against missiles is much harder than defending against planes (be it normal ones or stealth ones). But you're not wrong, as they were supposed to be good at both, and they failed miserably against missiles.

Iran has been investing in low-frequency radars that are better at detecting stealth aircraft. Aldo the areas around Tehran and key nuclear facilities (like Fordow or Natanz) have particularly dense air defense coverage. Iran also has a good number of S-300/Bavar launchers which can saturate the air space when radar is ineffective. Not to mention that they have an untouched air force (though they are probably not in the best shape).

Random idea: Couldn't Babel translate async code to callbacks?