Hierarchies, Tagging and Personal Information Management

Org-mode has a dizzying array of features for organizing an infinite number of work flows. For a long time I resisted learning org-mode because it seemed too complex. Anecdotally the problem most people seem to have when starting with org-mode is org supports so many different ways of organizing information without any expectations or guidance on how to apply org's tools to your own personal information landscape. Org-mode is built around hierarchical outlines and so it supports organizing information in tree structures quite well. Tree structures can be quite limiting- especially when there are leaf nodes that you 'feel' should belong in multiple branches of a tree.

I have been trying to think about this feeling, and I think what it boils down to is providing access to the information we record. We write things down because at some point the number of things we need to manage overwhelms our cognitive system. We lack the ability to keep track of everything in our mind. By writing them down the cognitive task shifts towards keeping track of where we have put the information rather than keeping track of the information itself. We maintain complex mental models of pointers to the information we need. What happens when our pointer based models (reading lists, TODO lists, calendars, etc) overwhelm our cognitive system?

This is the question that catalogers have been trying to answer since Alexandria. Catalogers are primarily concerned about formal models of access to information. To this end in 1998 the International Federation of Library Associations and Institutions (IFLA) came out with recommendations called the Functional Requirements for Bibliographic Records (FRBR). FRBR is essentially an entity-relationship model for managing works, expressions, manifestations and items (e.g., books, periodicals, dvd's etc), it is intended to replace a 'flat' record-based information model that is currently in use. FRBR is a huge specification and obviously makes no sense in the context of my personal notes and research. However the idea of functional requirements for my own information management is appealing. In the next couple of blog posts I'm interested in trying to develop my own informal functional requirements for (my) personal information management (FRPIM?). I'll start with an outline:

Facts about Org-Mode

  • Org-mode supports hierarchies
  • Org-mode supports tagging
    • Tagging are more flexible, tags can represent anything
    • Tagging needs tighter controls to ensure it provides effective access
      • tag authority (e.g., authority records)
      • thematic tag groups (projects, subjects, action types (e.g., read, write, email) ?)
  • Org-mode supports networks (sort of)
    • linking between headlines
    • What is the difference between a network and a bunch of linked files?

What do I do now?

  • empirical evidence?
  • Why is this working
  • Why is it not working?

What should I do?

  • What aspects of my personal information should the org-mode hierarchy model?
  • What tag themes should exist?
  • How can I control the tag fixed vocabulary?

How to transition?

  • I have ~50,000 lines of notes, how do I transition these into the new system?

What about process based PIM?

It seems like really good information systems should be predictive. They should provide the information that you need 'Just-in-time.' The most obvious reason this might not be a good idea is Clippy. Is predictive information management just a bad idea? Or is it an issue of control over how predictive information management interfaces with a work flow? If you implemented a system like this how would you empirically know it was working?