Journal Entry For 2026-05-09

• I have started to put my phone on the charger in the kitchen when I'm working, instead of having it on my person. I think it helps being more in the zone. But who knows?
• Lobster.rs is cool. I sometimes would like to be a member, but I think it's mostly out of Fomo.
• I have a new project: It's called "from scratch" and lives on a new git hoster called tangled.org
  • tangled is interesting, because behind it is a Finnish company and it uses the AT Protocol (which is what powers Bluesky)
  • the project itself is described as: "an experiment to see if I could create a website from scratch by not relying (at least all too much) on already created frameworks, libraries, templates, AI coding tools, and so forth. Hence "from scratch"."
  • It has an extensive Write Like Ron Jeffries-style journal section in the README
  • In order to avoid using AI tools for this, I installed Sublime Text after not using it for at least 5 years. Still a pretty solid editor!
  • Progress is slow because there is actually a surprising number of things I do not know how to do from scratch or if I kind of know, then it takes me a while to write it down correctly. On the other hand: I learn a lot and it's fun. So there's that.
  • (2026-05-09) I haven't had yet any time to devote to this since first starting out last weekend. Classic.
• Since I refer to Ed Zitron relatively often for the economic case against current AI companies, I think I should link this article by Kelsey Piper, attempting the tightrope act of criticizing Zitron by trying to injecting some more realism and perspective without acting as an AI booster, which I appreciate, as I enjoy a grounded take over an overly emotional "ought"-based take: AI's biggest critic has lost the plot
  • That said, the counter arguments the article makes are not always that great either. I don't have the time to dissect them here, but use doesn't necessarily equal usefulness, for example. And usefulness (which is undoubtedly not 0 but also not unlimited) doesn't necessarily somehow safe "AI" from the enormous costs it incurs. Even if LLMs have actual real economic value as coding tools, that doesn't mean that value is high enough to make up for the enormous costs of datacenter build out, training and running these things.^[1] And I think burning money (the AI companies that is) on LLM subscriptions is bad business in general.
• Since I have shifted my Music listening away from the awful UI black hole on top of a questionable discovery algorithm that is Youtube Music and towards Jellyfin, I struggled with discovering new Music. I wrote about the German Plattentests.de but that itself wasn't enough. I tried Bandcamp's editorial but I am not that adventurous a listener, I guess. And using Last.fm after years, although cool, is only a supplement. So I have now started a Qobuz subscription. I will stop paying for the Gym soon-ish as I have still not used it this year. I went with their studio plan as the sublime plan didn't make immediate financial sense, as I am not an audiophile. Although some Albums I checked were still ~2€ cheaper (the Arctic Monkeys Discography, although not all albums) when buying "Hi-Res Audio" from Qobuz using a sublime plan compared to just going with CD quality. I'd have to buy around 2 albums a month to make up the difference (€199,99 for sublime vs. €162,00; 37,99 difference).^[2] It might happen, as I backfill my library as we speak, but I'll start like this to see if I like the product first and how much albums I'm actually buying if I have a non-atrocious streaming service at my disposal.
• (2026-05-09) in an interesting turn of events my partner also wanted to get rid of youtube music and so we now have a studio duo subscription.

• ← Previous
  DailyDogo 1619 🐶
• Next →
  DailyDogo 1620 🐶