Talk about some of the problems we're trying to solve.

What Chandler is and isn't

• Chander is a PIM: Personal Information Manager
• Chandler is a platform for an integrated General Information Manager: Documents, Photos, Music
• Chandler is not a feature list or a set of tools for getting things done (ie. toolbar)
• Chandler is a way of getting things done
• Chandler is a way to manage your life information
• More of an approach to life, less of a diet pill

Vocabulary

• Item: a discrete piece of information
• Collection: a grouping of pieces of information
• Attributes: a characteristic of a piece of information
• Kinds: a type of information or a grouping of attributes

• Kind is to Goat as Attribute is to Horns.

Current PIM software structure produces unnatural workflows for managing information

• What computers are bad at

• Doublethink: The ability to hold opposing ideas in your head the same time. This is salty yet sweet. This is good and bad. I love you and I hate you. I am a compassionate conservative. I am fiscally conservative and socially liberal. I am goat and I am man. Satyr example
• goat_human_satyr.gif:
  

Software doesn't need to think like humans. We just want software to allow users to think like humans. Which brings us to the question. Why is fuzziness and doublethink important to PIMs and GIMs.

Designing for a new generation of users who are more interested in what software can do for them, change their lives. Less interested in software as technology for technology's sake. Tools, widgets, gadgets. As software goes mainstream and isn't so much a form of technology as simply a product like Campbell's soup or Honda Civics.

• Reason #1 Sometimes things in life are hybrids. More often than not, things in life are hybrids. And more often than not, it's the hybrid, fuzzy, doublethink things that bite us in the butt, that need to be managed the most. (ie. Emails that are really tasks. Tasks that need to go on the calendar. Events that need to be emailed out as invitations.)
• venn_diagrams.gif:
  
• Why is why we modeled our Kinds of items to be able to hybrid-ize

• Reason #2 As David Allen might say: Why manage 2, 3, or 4 items when you can manage 1. Hybridizing let's you do that.
• How many times do you copy things down? re-enter the same information? just to get it from one software-defined format into another? Email into a task. Note into an email? Voicemail onto your calendar? How many times has this prevented you from getting that information out of your head into your PIM? How much does that render your PIM useless as a trustworthy system that reflects the reality of your life? How many of you would never even consider to think of your PIM that way? What if you could change or morph the format of the information directly instead?

• Design maxim #2: Divorce the thing, in this case information, the user cares about from the software instantiation of the thing (ie. the email, the event, the Word document)
• Traditional applications don't allow you to do this kind of thing because the technology constrains how much you can manipulate an item of information. (ie. editing an email, changing the creation date of a task). Yet, why should we be forcing users to interact with the technology? Users should interact with the stuff they care about. The software is simply there to facilitate that interaction. It shouldn't get in the way of it. So if an user wants to change the date received attribute on an email, why stop them?

• Reason #3 Sometimes you don't know what Kind of item you want until you start making it. Other times, what you want changes over time.
• Design maxim #1 Let people make decisions when they're ready to make them
• any_order.gif:
  

• Note
• Task
• Event
• Invitation

* Fuzziness: Circle-square example...

• • Computers have trouble with -NESS and -ISH. Roundness, squareness, blue-ish, red-ish. It causes them to malfunction. Humans however, manage pretty well with vagueness, to a point.
  • We can understand the general concept of a circle-square and recognize circle-squares when we seem them. Computers on the other hand, need a specific definition of a circle-square in explicit terms. They require a mathematical articulation of circle-square. Which frankly, humans find very hard to do. So until some human figures out a general math rule to define circle-square, computers are unable to recognize circle-squares and will require a human to explicitly label each instantiation of a circle-ish-square-ish thing.
  • circle_squares.gif:
    

• • Humans understand gestalt the big picture, we find it hard to calculate the details. Some humans are like computers. They grow up to be engineers ;o). Some people can see that there are x-number of dots of each color, but can't see that the dots make up a picture of bathers on the beach. So the next time you get frustrated at your computer...
• seurat.gif:
  
• seurat_reimagined.gif:
  

• Productive members of society. Challenge is to design software so that it complements humans and facilitates human abilities, not trying to automate human functions. That's an oxymorons. Because humans are specific not automatons. So why should we try to automate human thought processes. Software has traditionally been too focused on automation. Instead it should leverage what computers are good at to provide a framework to help people be more human.

• Reverse Heisenberg principle: The bigger the picture, the fuzzier the details. As a result, a lot of our gestalt ideas are pretty subjective. For example the notions of good v. evil as applied to pornography and SPAM. Hard to articulate what they are precisely because they differ from person to person. Moral relativism. As a result, how can computers judge what's good or evil? Instead, we need to focus on non-subjective attributes that help provide people with at least a coarse-grain big picture so that they can go in and
  • SPAM: All caps, Center-aligned, Too many font colors, HTML
  • The problem is that humans aren't perfectly rational. They are rational-ish. Or rational some of the time. Irrational some of the time. So as your SPAM filter tries to learn from your good v. evil decisions, it will inevitably get confused as you contradict yourself. (ie. When your Mom forwards you SPAM that she thinks is funny.)

• So how does that translate to the PIM domain? Well it has to do with the main design challenge. How do we structure a person's data so that it's most useful to them? the way it's currently structured has proven itself to be a major impedence to productivity. By Kind, In v. Drafts v. Out v. Sent v. Folders ordered by Date.

• The dream of AI is that magically computers will be able to automatically guess things that even humans are incapable of articulating to themselves. So computers favor structuring apps so that they depend on a minimum of human effort and a maximum of automation. The problem is that most of the time the types of things that are appropriate for automation are not very useful for humans.

• These are all objective attributes that the computer is comfortable figuring out. Indisputable that an email is an email and it was 3rd from last thing you sent out today. However, they are the wrong attributes for yielding a big picture. They miss the point in the same way the color dot count misses the point of the Seurat painting.

• The problem is that often the most useful attributes for big picture are also the most subjective. Priority. But priority is really hard. It's too explicit and users have a hard time prioritizing in a vacuum. You can only really prioritize in relation to other things. Is this important? Important relative to what? In what context?

• So let's bring the user into the picture. Let them sort their stuff and then let's present the data wrt the way they sorted their stuff. So we decided to go even more coarse-grain. Fuzzy, but allows you to see the big picture. Triage status in our case. Let's allow users to define Triage status. And then let's present them the data primarily in terms of Triage status.

• Design maxim #2 Separate the technology from the user's data so as to empower users with maximum control over the content and structure of their data and how that data is presented.

A different way to approach this problem is from the process of decision making

• Current PIM software suffers from ill-fitting legacy design
• Legacy of file system metaphor originally intended to deal with rock-sized documents. Misapplied to manage dust particle sized information items.
• Legacy of 1-dimensional organizational affordances of the physical world. Things inside of things. Explaining the file system to my mom. There's an old Chinese proverb. Once there was a mountain. Inside the mountain was a temple. Inside the temple was a monk. And the monk said, once there was a mountain. Inside the mountain was a temple. Inside the temple was a monk. And the monk said,...
• Basically you organize things by putting things inside of things. But soon, you have so many things, and they're all tucked away inside of things which are inside of other things. That you don't know where anything is.
• Not only that, it assumes that this organizing is the best way to manage information. But a file cabinet metaphor feels like overkill for filing granule-sized emails and tasks. After all, we don't file our sticky notes. Only big, important things get filed. As our president might say: It's like going squirrel-huntin' with a bazooka.
• bazooka.gif:
  

• Chandler hopes to provide a better structure for more natural and appropriate information management workflows.
• David Allen: Collect, Process, Organize, Review, Do v. Categorize, Organize, Collect, Flag
• First you need a central collection point. Immediately, current PIMs don't let you do that. Newly created items and newly received items are immediately categorized by Kind and then further filed or organized into folders. In v. Draft v. Out v. Sent. Everything from there on in is all about trying to override these categorization and organizational barriers. (ie. Centralizing flagged items into flag folders with rules. Copying information from the email client into the task app.)
• Then you need a single workflow for triaging items in the central collection point to decide if they're Done, to be dealt with Later, or need to be worked on Now.
• Then you might categorize by Kind.
• And then lastly, you might single out a few important items to actually organize into collections or topics.
• This turns current software structures on their head, which immediately categorize and organize and then leave users to struggle over collecting and processing in really awkward ways.
  • Date sorted collections. No way to see flagged items and new emails.
  • No OOTB way to see flagged items across folders.
  • No OOTB way to see items related to a single topic across kinds.

• aka Design maxim #1
  1. Triage
  2. Orgcess
  3. Organize

• Organize_hierarchy.gif:
  

• Organize_Levels_of_Use.gif:
  

• Organize_XYZ.gif:
  

• Triage_Workflow.gif:
  

• Triage_Workflow_Linear_Time.gif:
  

---

Why is organizing so hard?

• Taxonomies are hard to construct, especially a priori. People are asked to know the future. Construct a structure without knowing the content of that structure. How many times have you...Created a structure and filed into it only to discover that it's not really the right structure but then you've already filed so much stuff, it's too hard to change.

• Project
  • People
    • Status

• Status
  • Project
    • People

• What about General Topic? Timeframe?

• Once you have a taxonomy: where does an item live?

Physical v. Virtual...1-dimensional v. n-dimensional

• dolls.gif:
  

Cool pictures of n-dimensional space Design is like a rubiks cube or a snowshaker...

Whats hard to do in your file cabinet is easy to do on a computer. Slice and dice your information in different ways. Turn the structure on it's head. A flexible taxonomy that changes depending on what you need and want.

Introduce sidebar

Then present some of the solutions we've come up with.

• sample_sidebar.gif:
  

• Entourage_CustomViews.gif:
  

mary carillo furry art

• n-dimensions.gif:
  

• mary_carillo.gif: