kivle

So, their defense in court will be "The AI did it"?

Proton VPN had their own controversial episode earlier this summer:

https://news.ycombinator.com/item?id=48446899

When I was a PS3 guy I remember the prices in Playstation Store typically being $10 or more higher than what you could get physical copies for, and things never went on sale. Having one storefront for all game sales seems like an absolutely terrible deal for consumers.

I'm in Norway, and I never consented in that dialog box, so my Facebook account is in limbo. The dialog still appears when I go to Facebook, years later, so I don't think anything has changed?

It must have improved considerably since I tried the "3.5-flash-preview" a couple of months ago if all these claims in the presentations are true. Because it couldn't even make changes in a 200 line Python script without doing major mistakes (like messing up argument order when calling functions) when I tried it.

If you taxed the rich and corporations like Europe does you could easily have social security. The rich in the US love making Europe the boogeyman that stole your social security, while they laugh all the way to the bank.

Unless China makes a deal. They definitely want their trade with Europe. Europe is currently a much more reliable trading partner to them than the US.

It doesn't take much for somebody to seem like a more reliable party to make a deal with than Trump. Trump has already made direct threats of military action against Europe. Making a deal together with China might be the better option.

Also, Trump asking the EU and other allies to contribute seems much more to be about shifting blame and cost. Participating in the quagmire he created, with such an unreliable partner seems like a terrible idea.

And in the olden times, people got nightmares from reading books, or by hearing a horror story around a campfire. Banning everything that is scary or can cause nightmares or trauma would be a very difficult effort, and deciding a boundary of what is too traumatic and what is not would be very arbitrary.

Or will they ban you from using something like https://github.com/filosottile/age to encrypt and armor text encode things you send inside of the non-encrypted chat?

echo "Am I doing something illegal, France?" | age -e -r age1ql3z7hjy54pw3hyww5ayyfg7zqgvc7w3j2elw8zmrj2kg5sfn9aqmcac8p -a -

  -----BEGIN AGE ENCRYPTED FILE-----
  YWdlLWVuY3J5cHRpb24ub3JnL3YxCi0+IFgyNTUxOSBjTVQ5VTdMaTlnRkEyT1BY
  MHZPc0lncHFvbS9FMTlDa2FkK3JQZy9sQnprClRFN3lNQUtnNzJWK0RxQVlYNE1q
  NCtlNFJTUWpwZExJSDMvSGlRL2VHc1EKLS0tIC95bEErRU9NNERJRVVuYlMwUFg4
  WUx1R0IyTHd1d2dxQTdqU0NJWlF0MXMKL1x9fz+ZVObYrn3bY/IdVBsd4KYxn78P
  aWePVjaRUityGTkndNSy6gg1meVky22iv4rxd9MZ4XYnsGJDfRUmkVZhQcCxag==
  -----END AGE ENCRYPTED FILE-----

Well, you can still run unsigned software (by clicking through to a bit of a hidden option in the popup dialog), and they also even remove that through "reputation" if enough people approve said binary (exact bitwise binary, so every new version released will go through the same issue).

Sourceforge died in a very different way though. Bundling spyware/crapware in install packages for open source software was a serious breach of trust, and was pretty much the direct reason for mass migration to Github. Github is failing on the technical side, but they at least mostly have their brand value intact. I think it will still take quite a lot for a mass migration of the same scale to happen.

I think it has more to do with digital verification for social media in a hope of killing bot accounts that are interfering in the public debate. Some of the biggest social media influencer accounts turns out to be Chinese/Russian bots trying to fuel hate/division our democracies, and with LLMs it is only getting worse. Some form of digital ID to verify social media account identities is probably the only hope left of having a real public debate.

I agree.. Maybe parts of the cache contents are business secrets.. But then store a server side encrypted version on the users disk so that it can be resumed without wasting 900k tokens?

With the help of coding agents it's easier than ever. Just get Claude/Codex to create Helm Charts / Docker Compose files for you. Struggle with some command line juggling to fix some obscure error? An agent can mostly help you in no-time.

I will consider this.. Opus 4.7 has effectively made Claude Max 5x into Max 2.5x.. 5 hour caps went from being a minor annoyance to being painful.

Edit: And honestly, I can't say I see much improvement for a "doubling in price".

They already effectively halved it with the introduction of Opus 4.7 and the new tokenizer that basically gives you about half as much usage for the same price.. Convenient to price based on tokens, and leave what a token is a moving target..

Hate that about Claude Code. I have been adding permissions for it to do everything that makes sense to add when it comes to editing files, but way too often it will generate 20-30 line bash snippets using sed to do the edits instead, and then the whole permission system breaks down. It means I have to babysit it all the time to make sure no random permission prompts pop up.

RIAA already tried to take down the Github repo for youtube-dl (basically the original yt-dlp was forked from) back in October 2020. But outcry from among others EFF got it reinstated just one month later. Google is probably on the fence about this because they saw how it went last time. The slow killing of adblockers in Chrome seems to be something they are getting away with, so maybe that will make them bolder once things have moved along far enough that there's no way back.

I also get tons of helpful product recommendations and tips in my TV shows, every 20 minutes.

A bit like "suggested apps" in the start menu. It's "suggestions" and certainly not paid ads.

I had the first two iPhone models, but then moved to Android. So I've been an Android user for ~15 years. This will probably be the drop that makes me go back and try an iPhone again. If all phone OSes are going to be walled gardens, might as well go for the best one.

Android has always been lagging on usability/performance/polish, but I stuck with it for the openness and because it generally was first to tryi new things. I remember how people at work laughed at me when I got a Samsung Galaxy Note ("It's so big it looks like you have an iPad in your pocket"), yet a few years later every phone was that size. And now Android is leading with foldables. I love my OnePlus Open, but OnePlus seems to be pulling out from the Western market so further support is looking "iffy", so might as well get an iPhone.

I've noticed recently that my Windows 11 start menu isn't a 50% gray empty blob anymore. I always disable "recommended apps" (eg. ad supported app suggestions), so 50% of the space in my start menu was unused, yet it was still there of course. Kind of refreshing to have a somewhat normal looking start menu again with something as revolutionary as a list of apps right when you open it. Let's see how long it takes before they mess it up again.

The fund owns about 1.26% of Microsoft (data seems to be for 2025), which according to Gemini is about $37.5b in today's value. Stock value of Microsoft changed about 5.23% over the last year, which comes down to about $1.96b, so you're not far off...

https://www.nbim.no/no/investeringene/investeringsoversikt/#...

I tried the Anthropic models because gemini-pro had already been rate limited with a 5 day wait. I got some actual usage out of the Google model, but laughably little compared to what I got with ChatGPT Plus. This is definitely not an imagined thing from my side, you just have to look at the Antigravity forums:

https://discuss.ai.google.dev/new

Limits are so low that I cancelled after about two weeks on my initial $0 trial. I tried making a change to a tiny code base with Claude Sonnet (which they offer in Antigravity). It couldn't even finish the change before my weekly limit was used up, reset in 7 days.

He is probably talking about this repo: https://github.com/navikt/DSF

Description translated:

This system was one of the oldest IT systems in NAV, and ran in production for 51 years, from when the National Insurance Scheme was introduced in 1967. In January 2018, Presys was put into production, which together with Pesys became the successor to DSF. At that point, DSF was also shut down. The system is written in PL/I.

It's like the Apollo 11 code, but for social services.

There used to be a lot more of that, but a system was put in place where you have to identify yourself with electronic ID to access the information, and the information is logged so the other party can see it.

Nowadays I think mostly journalists use it to pull up information about politicians and other people that are in the public spotlight. There are of course the yearly "richest people in Norway" lists in various categories.

We're so open, we even leak our government source code _ourselves_ https://github.com/navikt

I switched from ChatGPT Plus to Gemini Pro instead of Claude, since I'm a hobbyist and appreciate having more than just text chat and coding assist with my subscription (image gen, video gen, etc are all nice to have).

At first I found the Gemini Code Assist to be absolutely terrible, bordering on unusable. It would mess up parameter order for function calls in simple 200 line Python. But then I found out about the "model router" which is a layer on top which dynamically routes requests between the flash and pro model. Disabling it and always using the pro model did wonders for my results.

There are however some pretty aggressive rate limits that reset every 24 hours. For me it's okay though. As a hobbyist I only use it about 2-3 hours per day at most anyway.