I'm incapable of vibe-coding. I keep trying to do it and end up elbows deep in the code-base after one or two commits. I have no idea how anyone can stand it.

When you're nervous about using randomly-generated nsecs to sign events during automated tests because you think, "But, what if someone selects that key-pair, later, and sees the test event?" and you realize that you're a math whiz, but your brain still can't comprehend really large numbers. I mean, it can. But not _really_. #math #devs

Woke up. Have the day off. Feeling cute. Vibe-coding a sloppy mashup of #Copyparty and #NKBIP-04, for the lulz. https://github.com/Silberengel/copyparty/tree/nostr-implementation https://next-alexandria.gitcitadel.eu/publication/naddr/naddr1qvzqqqrcvgpzplfq3m5v3u5r0q9f255fdeyz8nyac6lagssx8zy4wugxjs8ajf7pqyfhwue69uhkcmmrv9kxsmmnwsargwpk8yqqsmntvf5hqtfsxs7cfsef

Introductory video to Alexandria. You have to download it, to watch it, as it's .mkv format. View quoted note →

I keep having to turn down work, including for Nostr projects, because I'm swamped. I understand that this is a chronic worker-shortage problem and vibe-coding is meant to solve it. I get it. We also don't have enough developers on the team, which is why I had to switch roles. I just think vibe-coding will create a lot of new problems and professionals are going to have to code and architect and test and DevOps defensively, to avoid getting crushed under the sheer ubiquity of this mysterious code. Also: I suspect that some of the most popular Nostr libraries are vibe-coded because we're slowly replacing them with our own hand-rolled stuff and the quality of that functionality immediately jumps. The problem Nostr is now facing, is that the vibe-coded apps are using vibe-coded app-building tools built with vibe-coded libraries...

Ignore vibe coders, when they tell you what their software does. They don't actually know what it does. They also have no clue how computers work and networks function. They couldn't design efficient and effective algorithms, if their lives depended on it. They say they tested it. They're good testers. No, I am a good tester. And part of being a good tester is... Knowing precisely what the software is meant to do. Knowing how computers work. Knowing how networks function. Knowing how to design efficient and effective algorithms. The jokes on you: Good testers know what the software does, better than the people who coded it. Also, be honest, the AI writes the tests, too, doesn't it? They've "reviewed" it. Except that it's also code. So, no, they didn't. This is like purchasing food from people who can't tell you what the ingredients are or how it was produced. It's usually going to turn out fine.

No, it's not a waste of time to use normal programming and clever architecture to make big data directly accessible. I know that you can ask ChatGPT to print out what might be chapter 14 from Oliver Twist, but there is value in simply asking relays to show you chapter 14 directly, instead. Not everything has to be AI.