Image by Paul Brennan from Pixabay
On the Fediverse, the microblogs (Mastodon, its forks, Misskey, its forks, etc.) lack a group functionality. Since tagged posts generally only federate to servers who already receive posts from the originating server, a lot can be missed. Groups offer a bypass. If I follow ActuallyAutistic, it doesn't matter whether anyone on my server follows your server or not. When you @ ActuallyAutistic, I'll see your post.
While they weren't the only group host, a.gup.pe was the main one that people were using to host groups. This had issues because the software that a.gup.pe ran on didn't play well with non-Mastodon microblogs. Sometimes it would work with non-microblog platforms on the Fediverse but it was kind of spotty.
a.gup.pe
, the flagship instance for Guppe, went down recently because the policy for .pe
domains allows others to buy the domains out from under the owners during a window leading up to the domain expiration.
This serves as a primer about Group functionality. It also covers serious issues with our current options.
If you get far enough into it with the LLM crowd (the ones who insist on calling what they're doing “artificial intelligence” as if they're talking about Data or Hal), they'll tell you it doesn't matter if we burn the planet achieving “AGI.” Why? Because it will give us the solutions to the environmental disaster we're in right now. It'll give us the solutions to our political differences. It'll solve world hunger, pollution, homelessness, poverty, and all our other ills.
Let's say we give birth to this AGI. Let's say that it has human like or human exceeding intelligence. Let's say it agrees to work with us on what we want it to do and what it asks for in return is something we're willing to give.
What makes them think that the answers it give us won't start with “you should have done the things you already knew how to do instead of burning everything into the ground making me”?
Today, when logging into Bluesky, I got a notice that Bluesky were changing their ToS, Privacy Policy, Copyright Policy (“effective September 15th, 2025”) and their Community Guidelines (“take effect on October 15th, 2025”). You can read Bluesky's official statement about this update on their blog entry Updated Terms and Policies published on August 14, 2025.
Unfortunately, the way these changes were handled are a huge problem.
Image by Gerd Altmann from Pixabay
If you use a streaming service to listen to music, you might already be familiar with having lyrics play on the screen as the music goes.
If not, you may not be aware that there's a way to do that locally with your library of music. I'm using LRCGet to download the LRC files. I can use the resulting LRC files either on my computer through LRCGet or on my music player. Your mileage may vary on how many music players support it. This isn't an ad for the concept or the software.
Just like some songs aren't “synchronized” when you're streaming (so you just get the lyrics without the automatic scrolling), some aren't synched in LRC's library either.
I've been going through and creating the synchronization for a few songs in my library every day or so. Thanks to that, I have a very specific version of So Long, Marianne stuck in my head from a few days ago.
This isn't the usual fare for this blog but I don't have anywhere else to put it. I hope this will prove useful to people who need to use a compatible assistive device.
Mastodon folks are often unaware of the different types of software federating over the Activity Pub protocol. It's not just Microblogs emulating what Twitter was. One of the areas is Discussion Forums like Reddit but federated. This includes Lemmy and Kbin. Because the focus isn't on microblogging, these platforms can offer feature sets that are beneficial for more nuanced discussion. What you may not know is that you can interact with discussions on federated discussion forums from your Mastodon account.
This isn't a comprehensive guide. It's not intended to be one. It's just some basics you need to get started with a bit of an explanation as to why you'd want to.
Photo by Eri Krull on Unsplash
This article isn't about the ethics of how models are trained or the ecological consequences of their use. DeepSeek shows that models can be generated on lesser hardware and that massive data centers with huge ecological consequencees aren't required to operate them. DeepSeek doesn't demonstrate the viability of ethically producing a model.
DeepSeek doesn't build a tool that actually does the kinds of things that real people would need it to do in order to be an effective tool.