Hans-Christoph Steiner @eighthave@librem.one

@surendrajat @IzzyOnDroid we welcome help on #FDroid! Probably the easiest place to start in speeding up the process would be to profile the `fdroid update` and work on speeding that up. Otherwise, there are issues in our issue tracker tagged with "help-wanted": https://gitlab.com/fdroid/fdroidserver/-/issues/?sort=updated_desc&state=opened&label_name%5B%5D=help-wanted&first_page_size=100

"If you can’t be sure whether you’re creating a self-aware #AI, this is alarming not just because of the moral implications of the “self-aware” part, but because being unsure means you have no idea what you are doing and that is dangerous and you should stop."
https://time.com/6266923/ai-eliezer-yudkowsky-open-letter-not-enough/

@vitriolix hehe like 2012?

@vitriolix the flatpak part of it was easy, remembering how to kickstart that crazy old build system was the hard part.

@eharlitzkern @histodons @histodon wow that is an excellent observation. As an Austrian and American, I've often felt how different things like smiling are used between the two. Those images in the blog post are definitely full on #American-style toothy grins. I feel like it protects Austria a bit because it makes #AI generated media feel more foreign, similar to computer generated voices. German text-to-speech never sounds Austrian, from what I've heard. I'll bet the Swiss feel the same.

@gamambel "Das muss doch machbar sein, mit sinnvollen Dingen Geld zu verdienen... Was ich mache, das können ganz, ganz viele tun"

@matthew_d_green uh yeah, and then the repressive government tweaks that same chatbot to start out with dissident material to hook you, then switches to convincing you that the repressive government are the one true way. It looks pretty clear to me that the repressive, resourced governments will dominate that arms race. It is much harder to break into human networks.

I love the idea of probing undocumented #APIs to find interesting data patterns. Everything that a website or app shows a user was probably generated using some #API calls, and it seems likely there are many that can be probed. #Hackers dig in!
https://themarkup.org/levelup/2023/03/03/journalists-you-should-be-looking-for-undocumented-apis-heres-how-to-start

So I completed a minor revival of Pd-extended, the #PureData fork with batteries included. I've published it on #flathub, it also runs on #RaspberryPi https://flathub.org/apps/details/info.puredata.Pd-extended

I've been teaching it a bit, and I was missing the UI as compared to the other forks. Hopefully this is useful to someone else.

I run some #Tor relays, and I've noticed recently, like in the past year or two, that quite a few websites and services now block direct connections from the IP that the tor relays are on. These are not exit relays, they are just relaying Tor packets to other Tor relays, so the website and services should not see the IPs. My guess is there are lists of IP addresses of Tor relays and services are using that list to block traffic. That is alarming since anyone on the net with a relay is penalized

@micahflee Actually, I forgot to mention, GitHub does an OK job of it, given that public repos are fully readable without logging in, and it works pretty well over #Tor. But they also assume most users are logged in all the time. They could change their #UX so that it only uses logins when they are actually needed. I fear too many companies like #Microsoft are fully hooked on tracking users to even consider this.

@micahflee With my #security hat on, I totally agree. With my #UX hat on, I'm sad that modern computing forces users to do arcane procedures. I think part of the problem here is due to how the internet has been structured: the services want all users to login so they can be tracked and sold to the highest bidder. #Jitsi is a great example of a better approach: work without logins, only use auth when needed, and allow users to handle auth on the fly as well as long-lived.