Remembering

| emacs

A friend and I were thinking of papers to submit to the Loyola Schools
Review. I joked, “What could I write about PlannerMode?” It’s an
organizer that I happen to really like, but which I don’t think I can
get other teachers to adopt (Emacs’ll scare the heck out of them!).

I thought about why I liked PlannerMode. Planner as a TODO manager
isn’t particularly special. Although I can assign tasks to categories
and so see a breakdown of what projects are taking up my time,
Evolution and Microsoft Outlook provide more powerful task support. In
other task managers, you can e-mail tasks, assign multiple categories
and fill in all sorts of metadata. You can even synchronize your tasks
with devices like a phone or PDA. So why use Planner?

I realized that integration into my way of life and automatic context
clues are what really make planner tasks worth it for me. I don’t have
to switch to another application to create a task. I can just hit a
keyboard shortcut. Planner uses a minibuffer to get the task
description. My windows are not rearranged in any way, and I can look
at the data that’s relevant to a task. Not only that, tasks
automatically pick up context clues, like whom I’m talking to on IRC
or the file I’m editing at the moment. This cuts down on the explicit
context I need to include and makes it easier for me to bring up the
task again.

As a scheduler, PlannerMode is also not particularly distinguished.
Sure, it can display my ~/diary, but for that matter so can M-x diary.
Evolution and Outlook can give me a more graphical view of my time,
sync with my PDA, and coordinate my schedule with other people. Those
applications support detailed schedule entries with powerful cyclic
options. On the other hand, PlannerMode gives me a personal, plain
text view and (at least the way I use it) requires me to edit a
separate file to add new appointments. However, it does have one
advantage – my schedule is always loaded. I used to use Outlook on
Windows, but having my schedule in a separate application meant that I
actually looked at it very rarely, as I had turned off reminders
because they got annoying.

PlannerMode‘s notes, however, are what really convinced me. I can hit
a keyboard shortcut from anywhere and type my notes into a buffer
which automatically keeps context information. After typing the note,
I can then categorize it. I think that the critical thing here is that
interruptions—fleeting thoughts—don’t break my flow. I can just pop
up a remember buffer, stow that thought away somewhere, and go back to
it whenever I want. In contrast, creating a note in Outlook means
switching out of my application, making a couple of keystrokes, typing
the note in, and then switching back. The context switches make it
hard to keep track of where I am and what I’m supposed to remember.
Not only that, I need to enter context by hand. Even though I can
color my notes and reorganize them in Outlook, I find the context
switch too expensive. I used to keep notes in other knowledge
management tools as well. Some applications allowed me to
drag-and-drop links into the current note, and that was cool. But
that required a manual action, and those applications didn’t feel
integrated into my way of working.

I guess that’s why I like PlannerMode. Unlike other organizers which
don’t know anything about the applications I use, PlannerMode tries
its best to integrate into the way I work, and it’s easy to extend.
Fortunately I do almost all my work in Emacs, so I can think of my
organizer as integrated into my e-mail client, Internet Relay Chat
client, web browser, file editor and even games. It automatically
picks up context clues from these applications and allows me to easily
jump back to relevant files. It doesn’t distract me. It allows me to
key in data and then it gets out of my way.

The processing that happens in the background (publish to RSS and PHP
for me) is a bonus, and publishing my task list and notes online has
greatly helped me. It gives other people a way to see what I’m working
on and what I’ve planned for the future. Occasionally people write in
with additional resources and helpful tips.

This made me wonder how to integrate it with more applications and
make it available to more people. Yes, even the vi users and the
non-Linux/Un*x geeks. Here are some of the thoughts that went through
my head.

– What if everyone could have the equivalent of M-x remember even if

they’re not using Emacs?

– What if we could capture context from any application?

– What if remembered thoughts could have multiple categories and

could be reorganized later?

– What if we could use statistical methods to suggest categories for

text and find related items?

– What if whenever you interacted with the system, related notes would

show up? When you open a file, browse a page or look at a contact, a
sidebar could allow you to quickly jump to related topics.

– What if you could visualize your remembered thoughts to see how

they cluster around certain topics?

– What if you could perform powerful search operations on

remembered thoughts?

– What if it could pick up even more context clues, tying together

different applications? For example, if you do your e-mail in Eudora
but keep your main contact information in Outlook, wouldn’t it be
nice if remembering from an e-mail also linked in the business card
from Outlook?

– What if we could select some of these remembered thoughts and

publish them in different formats, automatically or manually?

– What if we could share our stream of remembered thoughts with other

people?

– What if computers could generate remembered thoughts as well, and

we could select which ones we’d like to include in our streams?

The more I think about it, the more I feel that this is worth studying
further. If we develop this idea, other people can benefit from it –
not just those who incidentally use it in the course of using
PlannerMode! John Wiegley was onto something good here, and it’d be a
shame to keep it only within a niche community. I’d like to make it
easy for my mom to use this. =)

For my MS degree, I want to develop a framework for remembering
thoughts from different applications. This means separating the
front-end, the back-end, and the context gathering functions. Right
now, everything is within Emacs. I’d like to split it up so that
people using windowing environments like GNOME or KDE can get a nice
dialog box if they want and people using other editors on the console
can use their favorite editor to add notes. I’d like to split the
backend up so that people can remember to SQL databases, plain text
files, mail—anything and everything. I’d like to split the context
gathering up so that application developers can easily hook into the
system.

For my PhD, I want to look into managing all of this information.
How do I find relevant documents now that I’ll be capturing
vastly more information? How can I make browsing my personal data
store easier? How does it change my work style? Can I think of
interacting with my computer—working on e-mail, writing documents,
etc.—as interacting with a stream of remembered thoughts? Can it be
my primary interface? If we shared our remembered thoughts with other
people, how would that affect the way we work?

Related projects and concepts:

– Yale’s Lifestreams project (concept of interaction with

time-ordered data in streams)
– MIT’s Remembrance Agent (implicit queries for relevant files)
– MIT’s Haystack
– Zoe
– Personal information managers
– Knowledge managers
– Information retrieval
– Intertwingularity
– Web logging

../../notebook/plans/2004.03.25

You can comment with Disqus or you can e-mail me at sacha@sachachua.com.