• 1 Post
  • 66 Comments
Joined 1 month ago
cake
Cake day: September 24th, 2025

help-circle






  • I run my own matrix server, too. I’ve never had any issues with updates. Honestly, I just pull the new docker image without thinking twice about it. It is risky, of course, but I have daily automated backups, so I’m not too concerned. Personally, I felt like setting up a matrix server for the first time was the biggest pain in the butt. It’s extremely convoluted with very poorly written documentation that is often outdated or incomplete. After I got through that, it was smooth-sailing from there. Setting up mautrix bridges one you got Synapse to work is actually really easy, though.





  • A very rewarding part of getting into 3D printing is that after the initial purchase, I started repairing things left and right like a standing mixer for my grandma or a thread adapter for my partner. I didn’t picture just how many things could be saved and how many purchases would be avoided. I’m always glad to be able to print some niche thing for under a euro worth of material that would otherwise cost 30€ or so and only be available on Amazon. I do realize the irony in having to make a non-insignificant purchase to acquire such a device, but it did open up a whole new world of making things on my own.









  • I love the correction system we have at my university. All the exams are pseudonymized with a sticker you receive during the exam and scanned after completion. About 10 to 30 people are involved in correcting the exams for one course. We don’t know who the exams belong to as we only see the scanned version on our tablet or computer. Each task is corrected by a different set of people. We can select to see only a single task or subtask to streamline the process of correction, too. Furthermore, all the tasks are checked twice independently. Once done, the system can assign the exams back to the students. I love how it’s fair and “anonymous” by design.


  • It will only get worse from here. The internet is getting flooded with slop, diluting the actual human knowledge. If you tried to train it again now, I’m sure the results would be worse due to less quality content in relative terms. This is the thermal death of the internet as the entropy plateaus.