What Are You Doing? - low-effort personal time-tracker.

It is a tool to estimate shares of time you spend for various activities without invasive data collection or extensive journaling.

The idea is very simple. Program asks user from time to time (e.g. few times a day) what user is doing now. User is expected to provide short answer like "work/coding", "work/call" or "websurfing". It's fine to not provide response when you're away. Program waits response for a limited amount of time (30 seconds by default).

Once enough datapoints collected, it is possible to approximately calculate share of time spent to each activity: the share is proportional to number of occurences of some activity in the dataset. This approach is similar to sampling profiler, but applied to human user instead of a running program.

Potential strenghts of this approach:

• No invasive tracking. Application doesn't log all of your activity.
• Fewer but higher quality data points. It does not try to infer kind of your activity just from names of programs you use. It's not practical to distinguish between work and entertaiment activity just from active application or domain in the active browser window because some of them (especially instant messengers) may overlap. And vice versa, some similar activities might be done via different tools or communication channels. Instead it requests manual input from time to time to get some quality data point.
• Low effort. Entry requested just few times a day. No need to clock start and stop of specific activities.
• Gently guided operation. Application reminds about itself when needed. No need to keep in mind anything, no need to remind yourself to do some actions as for usual time tracking approaches.

Data is sampled with exponentially distributed random intervals. That makes probes quite fair: not skewed by periodical schedule and not anticipated by user so it doesn't influences user's behavior.

For now the only target platform is Linux, but probably it will also work on macOS.

Just put script somewhere and make sure either zenity or kdialog is available in the system.

See deploy directory for useful deployment templates.

Just run the script. Collected probes will be available in CSV file located at ~/.wayd/log.csv by default.

Following environment variables are recognized: