Schön, dass @Anjapf@kirche.social hier gelandet ist. Herzlich willkommen!

Gern erinnere ich mich an dieses Interview mit ihr zur #UN #Behindertenrechtskonvention:

kom-in.de/189/welche-barrieren
#Inklusion #blind

The Matrix Conference 2025 was a huge success! 🎉

We were delighted to be the anchor sponsor! 🙌

With so many governments presenting on their Matrix-based initiatives. There was a genuine realisation that @matrix is the future of government, and inter-governmental, communications.

See our round-up of the conference, which includes the links to all of the presentations: element.io/blog/the-matrix-con…

This entry was edited (2 weeks ago)

I was just listening to the open microphones in my parents' back yard. Heard something that sounded like a rooster. Mom says it's actually a large bird in a cage making roosterish noises.

Anyway, should you care about such things, the stream lives here:

stream.borris.me:8888/outside

I'm using a pair of very cheap dynamic microphones connected to a Behringer UMC-202 audio interface and a Raspberry Pi running Liquidsoap to stream and archive audio. I would like to one day find a way to upgrade to electret condenser microphones without completely exposing them to the elements, and run the whole thing on solar power.

This entry was edited (2 weeks ago)
in reply to Borris

I love how detailed you went into this, it's very much appreciated!! I use clippy bmc12's with a beelink mini pc using altacast or whatever it's spelled as. I thought about using a pi as that'd be tinyer, but I also like using the mac with audio hijack. Though the dynamic mic sounds very good as there's 0 clips seemingly. Thanks for this, man. I hope you're having a good day today. I have one of these at stonercloud.net:8000/outside-stream. There's no file extention. It just kinda plays, when, after all, I bring it up. All my mountpoints are public though Idk how to share them so other blindys like us care about these things can have an Ohioage SP-bmc12 streamy thing to listen to. It goes up and down whenever I unlazify myself and things and bring it up. Have a great day, bro.
in reply to Dane Stange

I wanted a completely headless, very small setup, mostly because I don't live there and can't easily fix things if they break. It stays up all the time, unless it's broken.

If I were to use small condensers, I'm not sure which they would be. I'd love to stick a dummy binaural head out there with a pair of nice, quiet condensers on it, but also shield from wind and rain without affecting the sound. I'm not as worried about that with these cheap dynamic microphones. They get drowned all the time and don't really care.
I've only had to replace them once in 15 years.

Not sure how I would best do that in the existing environment.

This entry was edited (2 weeks ago)

I like this post I just came across, "Software can be finished": rosswintle.uk/2025/10/software…

I think the author is too quick to weaken the point at the beginning, and again at the end, so I'll state it more emphatically: more software *should* be finished.

Struggling with vibe coding? You just need to get better at prompting. Here is a six video series "course" to teach you how to write the most effective AI prompts and get the most out of vibe coding!

(Do you ever wonder what it would be like if there were some special language for giving commands to computers... a kind of prompting language so they would do *exactly* what you wanted? Maybe someone should invent that.)

This entry was edited (2 weeks ago)
in reply to myrmepropagandist

After reading this post, I've decided to flesh out the ideas a little. And I think I may have conceptualised the future of vibe coding!

First, the problem of inconsistencies about how AIs respond to prompts.

We need better predictability!

So let's create a standard syntax for prompts, so when you enter a given prompt in a particular way, the machine will always respond in a standard way.

We'll call these next-gen prompts "commands".

As long as you know this simple "language" of commands, the machine will always do what you want!

Now, on to the problem of complex prompts.

Let's take away the guesswork.

Instead of trying to precisely enter one complicated long prompt, let's instead have a series of simpler commands that get executed one after the other.

A "program" if you will.

So you don't have to re-enter your program on prompts, let's invent a physical media for storing and loading these programs.

Now, the next big elephant in the AI room is resource use.

Let's solve this issue so that the command interpreter "AI" software runs with very few resources. I know this is a stretch goal, but perhaps even an 8-bit microcomputer with just 64K of RAM?

And to really make this catch on, let's make this AI vibe command programming language so simple, even an 8-year-old can understand it.

A "basic" programming language, if you will.

Now I know this "Basic" programming language AI vibe coding system sounds extraordinary.

But I promise you that for just $10 billion in VC funding and total indemnity against copyright infringement, I will happily deliver a 64K 8-bit microcomputer with the BASIC programming language preinstalled!

This is the future of AI folks! Feel the singularity approaching!

reshared this

Y'know, I think people like the Jimmy Truths dude serve as an example of the shortcomings of #Fediverse moderation. This dude has likely made like, hundreds of accounts on dozens of instances, but all the instance owners can do is suspend his account. It's gotten to the point where the dude now keeps coming back with usernames and posts mocking the admins for being unable to permanently remove him, often on the same instance repeatedly.

Larger companies can do things like send cease and desist letters, contact his ISP, hardware ban him, etc a lot easier. They simply have more resources to deal with really tenacious harassers than the average instance admin does. I'm not a networking expert, but I wish there was a way to actually deal with guys like this because as the Fediverse grows in popularity, it's likely more users like this guy will crop up.

I can imagine it being pretty easy for bad actors to focus their attention on a single instance and overwhelm it if there's enough of them. Or people like this dude so persistent and obnoxious that it wears the often volunteer staff down over a long stretch of time.

There seriously needs to be focus put on improving moderation tools for Mastodon before we get to that point.

#Fedi #FediModeration #Moderation #FediTalk #Mastodon

in reply to Luigi.exe (Dragonheart System)

yup. I don't understand for example why it can't just call out to external services for things like that, like if it discovers a new account, or for posts, etc. Sure for huge servers this might become somewhat of a bottleneck, but for us it would be invaluable. I considered patching our mastodon to do that but I haven't had the time for that yet. So our scripts are a lot more primitive.
in reply to 𝙎𝙩𝙤𝙢𝙖𝙩𝙖 🥼🧫

@𝙎𝙩𝙤𝙢𝙖𝙩𝙖 🥼🧫 @Delta Chat Damn, it's so simple, easy to use and perfectly accessible with a screen reader on both linux and android. My personal request to the future would be add a setting that would start with camera muted, but that's just a personal preference and can eventually be done later.
On desktop the call is displayed inside its own window and I can keep chatting while calling.
Another request for the mobile app would be an ability to minimize the call so I can do other things with the phone while on the call.
Comparing this experience with other open-source audio calling apps such as #jami, #element or #conversations. This needed no tweaking or additional setup from my side and is the only one that has worked on a first try for me. Congratulations for such a great experience!

There is no single central #Signal server. Signal uses #AWS cloud, plus Google Cloud, Microsoft Cloud and Cloudflare, all under US legislation. If any of these clouds goes down or becomes otherwise problematic, chatting degrades or fails.

With #deltachat you can choose #chatmail relays ( chatmail.at/relays ) which distributes the risk across a wider network. We are currently working on the multi-transport feature to finally remove any central point of chat failure chaos.social/@delta/1153621448…

This entry was edited (2 weeks ago)

feld reshared this.

in reply to vermaden

it is now officially in the ports tree

however, I'm unsure when packages will be widely available as they're fighting over the current situation with electron because the builds are being very problematic and disrupting the build cluster's stability

freshports.org/net-im/deltacha…

With @signalapp having a problem because AWS us-east-1 is having trouble, I'm looking at installing @delta as a side-channel for situations like this ... does anyone know if it does video?

I know it's a chat function, and I'm not asking for miracles. It's a genuine question about capabilities.

I see there are web apps, and capabilities to integrate things, so I'm wondering what's available in the base install.

My investigations begin ...

#Signal #DeltaChat

#Signalapp appears to have some issues. The desktop app appears "offline" and messages are not going through,

EDIT: This likely is an outage resulting from AWS US-EAST-1 having some issues: health.aws.amazon.com/health/s… Many services are impacted. See downdetector.com/

This entry was edited (2 weeks ago)
in reply to daniel:// stenberg://

@koehntopp All the stuff I scrolled through on the watches about page states "may".

I assume they have complied a list of all the external libs they use in all of their products and then rather than produce a product specific list they just list them all for all products, prefaced with "may".

It is incredibly lazy. Surely they could automate it to have a correct list for each product.

Idk if @Tutanota has always been talking about legal issues on their socials, especially those that include a call to action for citizens, but I think it's a super smart thing for them to do from a marketing standpoint. The value of following their accounts is higher now because I know they will inform me rather than just tell me about a feature. Not that there's anything wrong with talking features, but they're expanding the value they provide through their socials specifically.
@Tuta

hey @FreakyFwoof I have a question for you regarding omni describer. I know you have said you have done a video in chunks before. have you ever had it forget names after the first chunk? I have a video I am trying to get described of May's guide playing with his new toy, that I am trying to have described. after the first chunk, it forgets our boy's names, and the dogs name, but if I do the video without spliting it into five minute chunks, I get an arror, and the suggestion is to put it into chunks.