Measurements
Numbers that went up, and whether that meant anything.
Two things here count themselves without being asked. GitHub keeps a running total of stars; a small program on my laptop writes down what the machine is doing every 322 seconds. Neither was collected to prove a point.
What makes them worth a page is the same in both cases: a cumulative number is flattering by construction, and the interesting question is almost always about the rate, the distribution, or the shape — none of which a total can show you.
GitHub stars
A star is a bookmark, not a vote
Twelve years of stars across every repository, and what happens when you look at the arrival rate instead of the running total. The total only ever goes up; the rate has been falling.
Read it →Telemetry · Live view
Four months of readings from my laptop
The last 36 hours at full resolution: what the machine is doing right now, how much it is moving over the network, and how warm it is while doing it.
Read it →Telemetry · Rhythm
What my laptop knows about my week
When the work actually happens, how sharply the weekend differs, and what a locked screen at 3am reveals about sleep. Stack up four months and a shape appears that no single day contains.
Read it →Telemetry · Heat and pressure
My laptop is a thermometer for my office
Throw away every reading where the machine was busy and what is left measures the room, not the processor. Plus what a fleet of AI agents costs in memory.
Read it →Telemetry · Network
What the internet costs, depending on where I sit
Latency as a distribution rather than a line over time, measured from home, hotels, libraries, two airports and a plane. Public Wi-Fi has a reputation it does not deserve.
Read it →