(Mid-April brain dump)
It’s Friday evening and I have a lot of things on my mind. Let’s see what shakes out.
http://editable.website is a cool idea. It’s a simple blog that you can click straight into to make edits (if it’s your own site). From that site, I met the creator, a guy named Michael. He’s working on something called HNA.
His idea for HNA is that blogs can talk to each other. So maybe you read a blog post and think it’s cool, so you reply. But instead of using standard blog software, your reply is authenticated through your website. So you use personal websites as a way of confirming identity and trust.
So we’ve been chatting about that on a Discord server run by this other guy who wrote a post about weird webpages, and I had reached out to him to share notes.
When Twitter spun off their Bluesky idea, I got involved on their Discord server. I also wrote a long comic about some interesting issues they’ll come across in terms of moderation. Now it’s been several years and Bluesky is nearing a public release. I reached out to some Bluesky contacts and jumped on. It’s been interesting.
I’ve been introduced to some Bluesky folks and I had a mentoring call where I worked with a designer involved in a different project idea. It reminded me how much I love teaching people design.
Work, work, work
I got to dive really deep into data this week, which I love. I’ve been in a great loop of learning new things from data, designing a better solution, coding it up, shipping, tracking the results, and repeating. It’s been cool.
Speaking of research, I’ve been starting research into my big Master of Design project this year. I dusted off Obsidian and I think it will be a helpful research helper. More importantly than tools and processes, I just love being stuck into a new research challenge. It’s been fun.
I recently learned what “vector databases” are and they still feel like magic to me. Apparently you put a bunch of stuff into them, they break everything into tiny little pieces, and then you can load it back up pretty easily. This works really well paired with chatGPT. So that’s where Vault AI comes in.
Imagine uploading the entire works of Shakespeare, then searching for the word “love.” In a traditional search, you’d get x results, then you’d sift through them. With Vault AI (which uses Pinecone, a vector database), you can say “what did Shakespeare think about people in love?” and it will give you an actual answer based on an analysis of the data.
This is a really big deal, so naturally I uploaded everything I’ve ever written. The results were a bit underwhelming on my own content, but it could be a game-changer for a company’s knowledge base.
My aperture is too wide
I love learning new things. But I’m learning too many new things right now. I need to sleep more and better. I need to have fewer headaches. I need to work less hard. I need to cut everything back to catch my breath a bit more.
Less input, more output.