Great device, wrong problem: Two months with the Ultrahuman Air
It's definitely a killer -- both the device itself and its companion app are really well executed. But it's not replacing my Apple Watch.
In the audacious pursuit of things that spark joy.
It's definitely a killer -- both the device itself and its companion app are really well executed. But it's not replacing my Apple Watch.
It's hard to find privacy-respecting (read: not Disqus) commenting systems out there. A couple of good ones recommended by Bear are Cusdis and Komments -- but I'm not a huge fan of either of them: Cusdis styling is very limited. You can only set it to dark or light mode, with no control over the specific HTML elements and styling. It's fine but I prefer something that looks a little neater.…
Turns out, we don't always have to go 100% -- it's often enough to just go.
A rumination on why faster coding models are better for me vs. smarter, slower ones.
I tried out my credit and debit cards with a couple of merchants, and the experience was generally great!
Agent mode but without any tools
I have 64 gigs of RAM. I thought that was plenty, but recently I've noticed Libre Hardware Monitor reporting that I was consistently at 90% RAM usage. How? Opening Task Manager reveals that VmmemWSL.exe was consuming 30GB of RAM -- but when I ran htop inside Ubuntu running on the Windows Subsystem for Linux (WSL), it was only reporting 6GB of use. I also run Docker, but it's reporting only 2GB of…
I’ve been trying to slowly ease into using LLMs for coding help again lately (after quitting cold turkey ), but something always feels off -- like we’re not quite on the same wavelength. Call it vibe coding or vibe engineering, but I think I’ve finally pinned down two big reasons why their approach to code feels so awkward. LLMs don’t copy-paste (or cut and paste) code. For instance, when you ask…
I recently shipped code in Typescript and Bun, and the experience was so much more joyful.
I recently wrote about how LLMs can make you lose grip of your codebase by eroding the mental model you built by writing code yourself. A buddy of mine shared this post with me, and I think it's the most clear and succinct description of this phenomenon -- and Jason Gorman, the author, calls it "comprehension debt". It’s pretty much guaranteed that there will be many times when we have to edit the…