ActivityWatch / ActivityWatch/activitywatch

Research Edition: classification signal varies drastically by platform/browser — document and narrow the gap

Open
#1,440 0 comments 0 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
18.9k
Forks
1k
Avg merge
1d 4h
Merged PRs (30d)
28

Description

Study-design concern surfaced during v0.14.0b5-research dogfooding:

Browser categorization quality depends on what URL/title signal each platform+browser exposes, and the variance is large:

| Setup | Signal | Quality |
|---|---|---|
| macOS + Chrome/Safari | full URL via Apple Events | strongest |
| macOS + Firefox | URL via AX tree (gecko walk) | strong |
| Firefox + [Add URL to Window Title](https://addons.mozilla.org/en-US/firefox/addon/add-url-to-window-title/) extension | domain embedded in title | good, cross-platform |
| Firefox / others, no extension | title keywords only | weak |
| Windows (any browser) | title only — no URL inspection path in the watcher today | weakest |

Implications: (1) macOS-Chrome participants produce systematically better-categorized data than Windows participants — a per-platform bias in the study dataset worth documenting in the study methodology; (2) the participant guide could recommend the URL-in-title extension for Firefox users as a cheap signal booster; (3) longer-term, Windows Chromium exposes the address bar via UIAutomation — a candidate enhancement for aw-watcher-window to narrow the gap.

Filed as follow-up work, not a release blocker.

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.