Write mission/vision statement

I have a lot of ideas about the future of ActivityWatch that I haven’t written down, and some of these are highly important since they concern the direction I want things to go in.

Communicating this is important for people who wouldn’t otherwise understand the use of the software. While we have some writings in the docs they were hastily written, don’t really get to the point, and an update is long overdue.

My ideas so far concern:

• Use by professionals
  • Consultants, teams, independent developers
  • Could lead to a source of revenue and financial sustainability of the project
• Use by individuals who are interested in self-improvement
  • Helps with community building and attracts contributors
• Use of data in privacy-first applications (like Thankful)
  • Enables new applications, helps with getting contributors.
• Use in research
  • Help society and individuals understand how we use our devices, how these usage patterns change over time, and how they affect other aspects of our lives.
  • One of the areas I’m specifically interested in is the war for our attention that is going on. If we’d have started this research sooner, we could have avoided this mess. See this video by CGPGrey for some context.
  • There might be opportunities to get grants for this type of research. (From where?)
  • We should add a way to cite the project, as done with GNU Parallel and scikit-learn.
  • Try to classify if a set of activities should be considered exploration or exploitation, see this forum post.
  • Try to predict depression from recent activity, see this forum post.

Finding synergies in these areas are important. I think ActivityWatch should be able to help will all of the above, but not all of it should be dealt with in the core software. We should also start thinking about ways to push these different use cases forward, and that includes reaching out to people who might be interested in one or more of these uses.

This project began some time after I first wrote my blogpost in 2014 titled “What would you do with your data?” and the question in the title has gotten much attention from me over the years. I now have some answers to what I’d do, but I’ve only done a small fraction of the things, and I can only do a subset of the things I’d want to do with the data I currently have. Notably, data of how others spend their time is lacking, so any type of comparative analysis is not yet possible.

Experimenting with the data caused me to create the aw-research repo where I can experiment with potential future analysis features of ActivityWatch. Other projects in the same vein includes chatalysis to analyze chat conversations and QSlang to serve as a timestamped diary where, among normal diary entries, there are also entries for supplements and drugs which are interesting subjects for analysis.

I’m open to comments and suggestions.

[ErikBjare]

Another thing I’ve been interested in, but that doesn’t really fit into the existing categories, is the implementation of auto-tagging and thereafter integration into a personal memex-like system or a personal knowledge base. All that would be needed would be a mapping (Event -> Tag) and (Tag -> Memex ID)

Doing this would incidentally do a lot of necessary work to build KnowTree (subject classification and automatic learning log).

Thanks @mekarpeles for pointing me in that direction.

[pcuci]

so any type of comparative analysis is not yet possible

Silvio Micali’s work on zero-knowledge proofs should hopefully invalidate this statement 😃 https://www.youtube.com/watch?v=zNdhgOk4-fE

It looks, at least theoretically, possible to generate secret common knowledge to query for shared insight without needing to know individuals’ private data. We might be able to see how similar one’s data is with another’s without divulging what exactly the similarity is, and without the algorithm having access to the underlying private data.

All that would be needed would be a mapping (Event -> Tag) and (Tag -> Memex ID)

This reminds me of https://github.com/Zettlr/Zettlr/issues/1690 and sister projects like https://github.com/laurent22/joplin/issues/3723 and https://github.com/foambubble/foam/issues/88

In terms of vision, why not scope down towards “the best holistic passive personal activity suite”?

It would start with desktop and mobile personal use at first, then the IoT revolution will add new smart devices’ data stream to the mix, as well as in-body devices.

I think the unit of privacy to design for (since privacy is a collective construct in addition to an individual responsibility) should be “the home” and all the devices one “owns” on their property, to help safeguard the privacy of the “family unit”. Microscale vs. individual scale.

How about integrating with something like https://github.com/home-assistant/core where “the humans” would essentially be another data stream alongside our private smart agents?

It’s not just my behavior I need to quantify, tweak, and understand. I also need to know what things in my immediate physical environment (as well as my digital environment) are cueing me towards certain habits.

Perhaps the vision is to take a serious stab at the oracle problem.