[j7ake]

I had a similar tracker set up using orgmode eMacs.

The most difficult part was that I needed a device to input my data, which was bad for a lot of my offline activities.

Ideally I would want a watch that can predict what I am doing and automatically guess. It would then fill in all the details and just ask me to review at the end of each day.

Training can be done both supervised and unsupervised, ie for supervised you explicitly tell watch “I am now working out”, and it would learn. Unsupervised is at the end of day, watch segments activity into discrete activities then asks you whether it segmented correctly, and assign label to segmented activity.

If it is Apple Watch, you can additionally leverage more information by figuring which apps I have open on my phone or computer.

[9_weslay]

Thanks. Agree that that would be great. But I feel that the complexity and the unpredictability of that additional step outweighs its benefits. What if a user doesn't have an Apple watch? I am sure there are workarounds but the fact that that is even a question a user has to ask already reflects some additional complexity.

[j7ake]

Yeah what I proposed is several phds worth of work, not thinking practically at all.