Updates for October 2026
Here's what I got up to in October:
My sewing classes wound down and I'm almost done with my chore coat and pants, but still need to assemble my quilt's blocks into a top.
We weathered a rough patch with our daughter's sleep and mood that very unscientifically seems to have resolved with iron supplements?
Fall is starting to show and we visited a more rural pumpkin patch that's not an hour away from our house.
I picked up The Rivendell A. Homer Hilsen bicycle in Walnut Creek and started using it for commutes and preschool drop-offs.
On Halloween, we watched Bluey's Big Play in San Jose and a quick trick-or-treat circuit of our neighborhood.
I kept at the notes, including some rare bicycle photos that I hope to expand:
I had to document Rivendell Bicycle Works after following the company for almost two decades now.
The Rivendell A. Homer Hilsen bicycle includes what I fitted on the bike and some photos.
I updated Bicycle gear companies a little bit with more companies I've purchased from.
I started reading Howl's Moving Castle but got stuck around when Sophie starts cleaning. I've only read a few articles this month, though:
"We should be more ambitious with the software we build"
This admonition reads much like how LLMs have spurred the development of straightforward, usable tools: an obvious good that only was identified as lacking when it impacts development velocity and the cost of measuring that impact is "cheap" (in effort). We should always have been more ambitious and it's just frustrating/disheartening that it took the advent of LLMs to realize it.
Software is slow because of bad design, not slow languages
Most real programs are I/O-bound
Sean has been around the corporate block but this is not the best take on performance I've seen. Maybe his "real programs" are servers that delegate to databases? Most real programs I encounter are bring-up and time-to-market focused and cut corners because they're in a rush. Which leads to all manner of performance sins.
Modern vision models can read information in the world in milliseconds, much faster than we can; it'd take me at least a few minutes to go through all the books on these shelves, or I can just ask.
This is a really neat idea, but I'm worried about the latency breaking you out of the flow with a tool like this. Regardless, it seems like the most promising of the various "ambient computing" initiatives out there.
All of which is related to the fact that one of the lolsob-mantras of courses I teach is "Now you know the true meaning of data". In the graphs and maps below, I’ve done the best I can, where appropriate, to uniquely identify individual dogs in the dataset (even though ultimately we cannot).
I love data visualizations and I love dogs. And as someone who helps manage a CRM database for a non-profit that's been around for three decades, I've seen some really messed up data too.
2 tools for asking San Mateo for safer streets
None of the staff's recommended speed limit increases are required by state law, and each segment's page in the tool lists the legal justification that the city (council, staff, city attorney) can keep the current limit–which is worth raising when the study session is scheduled in the coming months, as raising speed limits has a ratcheting effect that is difficult to reverse.
What's wild to me is that most of the decisions made by local governments are done on vibes, without asking for hard data or evidence from the planners and staff that support them.
I've started going out of my way to wear an analog watch, a Timex Weekender with a black dial in a silver case (a "panda" scheme). I also wear an Apple Watch to work, which makes sure I don't miss any notifications, but there's something reassuring and pleasant about a watch with a physical dial. I'm also getting a chronograph to try out as a bike computer replacement.
The walls in our house have been pretty bare since we moved in, filled with mostly hand-me-down photographs that my parents bought in the 80s and 90s. I wanted to buy art that we liked but kept finding clearly LLM-generated pieces online. Eventually, I found a few that seem largely authentic and bought some wood gallery frames from Blick Art Supplies. I'm excited to buy more pieces that resonate with us now that we've found a formula that works!
In a fit of delirium one afternoon after a handful of WebEx call dropouts, I rushed to replace my old Mikrotik router with a Mikrotik hAP be Lite. It's powered by USB and supports the latest WiFi standard, which are the only things that really mattered to me.
On Halloween morning, we went to the San Jose Center for Performing arts to watch Bluey's Big Play with our daughter. She loved seeing the characters in real life as human-sized puppets that actors move around the stage, sometimes two actors to one puppet for complicated motions. The show's voice actors were recorded for the script and the puppets had little mouths to move in time with the words. Oddly, my daughter kept asking at each scene change or rearrangement "is this the end?" so we had to keep reassuring her that it was not over just yet.
My daughter outgrew her little balance bike so we got a bigger one after she started showing interest in going out on walks with it. She's getting the hang of it still and only doing short launches to balance, but her progress is really promising! She's also not that frustrated when she falls off, which is different from most other things she tries to get the hang of.
I got a Ploopy A+ trackball as a kit but didn't realize that the optical sensor needed to be soldered by hand because it fits on the other side of the board. I had to bust out the soldering iron again for some narrow-pitch through hole soldering. It mostly went fine, but I did get a couple cold joints and the first pin I soldered was a bit overheated. Unfortunately, I don't think trackballs are for me; I want something that helps me move my whole arm to manipulate the pointer.