Tag
#research-methods
4 notes tagged research-methods, newest first.
Note 005Three times out of three, it refused a free changeI marked my own CSS system's specification with provenance — which claims are measured, which are judgement — and tested whether it helps an agent read it. The information helped and the medieval notation for carrying it earned nothing. The result I did not predict is that the unmarked specification made a careful reader block a change that was always free.Note 004Nothing built, and the design lost a serverI set out to bridge Signal into a Cloudflare pipeline and spent the session reading documentation instead. Capture and collaboration turned out to be two jobs, and one channel was buying both — which is where the phone number, the JVM and the second always-on host came from. Nothing was built, so the finding is documentation and arithmetic.Note 003Six adjectives got seven of the nine checksI measured my own notes and turned the counts into a voice document. Against a controlled test it beat the adjective list already in the repo on two checks out of nine, and both were habits a style guide would have talked me out of. The other seven the adjectives already had.Note 001Three kinds of bug, and the control I didn't runFinishing a real task with a feature surfaced problems I doubt a walkthrough would have reached. I can't prove that, because I never ran the walkthrough. What I can report is which problems needed which conditions before they would appear at all.