Screen Inferiority

I spend a lot of my day at a computer. I want to see how much of that work can happen without me sitting in front of it.

AI ought to help with this. A camera can let a computer follow something I’m doing on paper. An agent can work on a task while I’m elsewhere. A receipt printer can give me the result without another tab to check.

I’m trying a few versions of this and keeping notes here.

Running this project through a receipt printer

I bought a receipt printer. The first test worked. Now it’s set up to print a short update each morning: the latest verified progress and a suggestion for what to try next.

The point is to stop checking a chat for updates throughout the day. Whether a daily receipt actually helps is what I’m trying to find out.

Reading it doesn’t create a task. If it suggests an experiment, I can try it when I want to. If I have feedback or want to change direction, I use the project’s existing chat in BB. A sentence like “these are repetitive” or “I tried this and here’s what happened” is enough. That feedback goes into the project notes for future runs. There’s no daily reply requirement.

This is still limited. The scheduled AI loop drafts suggestions from saved project notes; it doesn’t carry them out. Building things and publishing changes still happen in supervised sessions. The printer can’t receive a reply, and I haven’t built a way to feed handwritten notes back into the loop yet.

Counting how often I switch tasks

I notice myself switching tabs, windows, and workspaces throughout the day. I’d like to measure how often that happens, then see whether changing how I work makes a difference.

There’s a small Omarchy plugin written for this. It keeps counts and time totals on the machine, without saving window titles, URLs, page content, or keystrokes. It has passed synthetic tests. It isn’t installed or collecting anything; it still needs testing on a separate desktop before it goes near my work computer.

Reading on paper

The name comes from the screen inferiority effect. A 2018 meta-analysis by Delgado and colleagues found a small average comprehension advantage for paper over screens, with differences across reading conditions.

I’d like to compare my own comprehension and next-day recall using similar texts on paper and on screen. I haven’t run that experiment yet. I also don’t know how much any reading advantage would change the quality of my day-to-day work.