Model using index cards

Agile software development processes seem to have a love of index cards. It might relate back to techniques such as CRC cards [ref] first described in 1986 by Beck and Cunningham, two of the fathers of Agile development. But, I'll guess Beck and Cunningham had index cards laying around for other reasons. User Centered Design approaches have been using card modeling techniques for decades now. So why should you? And what exactly do we mean by card modeling?

Writing ideas on cards shows you're listening

Suppose you and I are having a conversation about something important, I write notes as we're talking and you're giving me instructions. You can't see my notes, since my notebook faces me, so you don't know I misheard you unless you read my notes or I start using active listening to parrot back to you what I heard. If we don't confirm our mutual understanding fairly quickly after discussing a topic, the opportunity may be lost. I flip pages in my notebook; you start talking about a different topic. Any misunderstandings fade into the past.

Writing notes on cards and placing them in plain site allows you and I to both be clear about what was heard. It's like the active listing thing without me having to repeat what you say back to you. You can read it, and correct it if I am wrong. You can easily move back to something we talked about a long time ago and refer to or correct it or recall additional important details.

Arranging ideas written on cards rearranges and changes ideas

If part of our notes were a to-do list or a list of goals, I might ask you which are most important so I can number them in the order I'm to perform them. You'd imagine the list, and tell me the first couple important things and I'd label them as such. If as we're discussing the list you change your mind or something else comes up, I will have to erase or cross out the numbers and indicate the new correct order. If the list was long - over ten items, you might want to get back to me.

Writing ideas on index cards, one per card, makes it easy to prioritize lists. If I ask you to prioritize the items, it's easy to shuffle the cards into order. New items are easily added in their proper place. Even large lists are relatively easy to cut through and rearrange.

Given that we can both see and manipulate information on index cards placed between us, our interaction changes from merely a conversation to collaboration. You quickly see what I'm hearing, and together we deepen our understanding of whatever we're talking about.

Using our eyes and hands in addition to our ears seems to engage a bit more of our brain. Natural ways of communicating begin to emerge. If two ideas are closely related I might naturally place cards with those ideas written on them close together. If one thing happens before another, I might locate it to the left of the other - if I'm from a western country anyway. If one idea is more important than another or subsumes another, I might arrange its card above the other. These important but time consuming to describe relationships are trivial to indicate with simple card arrangements and gestures.

Guidelines for card modeling

If you buy all this, start by keeping a deck of index cards and a fat felt tip pen at hand during your next conversation about software you're working on. Actually, don't stop there… try it with your spouse or significant other the next time you talk about finances or planning a family vacation. I converted my wife when helping her sort through our daughters' Christmas lists. But, before you do here's a few simple guidelines to keep in mind.

• Write a single idea per card. Many of the opportunities we have with card modeling come from exploring the relationships ideas have with each other. We need to manipulate ideas independently. Writing more than one on a card connects the ideas and takes the opportunity with it.
• Write large and clear. The people you're communicating with need to be able to read what you understood when you wrote it. You need to be able to read and refer back to it later in the conversation, days, or weeks later.
• Rip up cards. If an idea is bad, incorrect, a card is poorly written, or cluttered with too much scribbled out and rewritten text, destroy it. Bad ideas written on cards in conversations are dangerous to have hanging around. Don't you wish you could get bad ideas out of your head this easily?
• Write out loud. As you write, say what you're writing. Collaborators will detect when what you write doesn't match what you say and call you on it. Another strange thing happens: later when you go back to read the card, you'll be able to picture and hear the person who wrote it and what they said at the time. Brains are pretty interesting recording devices.
• Everyone writes. Make sure stacks of cards and pens are available on the table to allow others to write. No need to slow the conversation down to the rate that one person can write. Most people can listen to discussion while writing. Allowing others to write turns on other parts of their brains, their tactile parts, and helps them relate to and understand the information being discussed. One thing to be aware of, lots of people writing at the same time is much the same as lots of people talking at the same time. After a while no one is listening or reading and collaboration stops.
• Everyone touches cards. Place the cards close to others you're collaborating with so they can easily reach out and touch them, change their order, or tear them up dramatically for effect, making sure the flying shreds don't put anyone's eye out. If you're facing someone across a table, place the cards on the table so they read right side up for that person. Or, even better, consider moving to a corner position around the table so you're sitting to the side of your collaborator - that way the card can read right side up for both of you.
• Arrange cards spatially. Don't let cards pile up in meaningless arrangements. Try to place them in an arrangement that makes sense. Even if they're simply placed in the order they were discussed, that's better than a pile. Once you start arranging you'll find a naturally tendency to let the importance and similarity of ideas influence your arrangement. When this happens you get information almost for free from the arrangement chosen.
• Model in color. Use different colors of cards to represent different kinds of ideas. Adding color to your model adds another channel of communications and another layer of information. If you find an idea written on one color of card should have been another, rip it up and rewrite it. If you're listening to someone and writing down ideas as they speak them, change colors as their ideas change type. They'll catch on.
• Plan to preserve the card arrangement. You can do this by covering a table top with poster paper and taping the cards down when you're done. Alternatively use double-stick tape on the back of cards and stick them directly to paper on the table or mounted to a wall.

Keep these guidelines in mind and you can invent a limitless number of modeling schemes to suit your purposes. But watch out for a few things that tend to go wrong.

Card Modeling Gotchas:

• Illegible cards cause the information to be ignored or misunderstood. Think of illegible cards as mumbled conversation. Don't do that either.
• Participants are afraid to touch the cards. If a meeting facilitator is the only one handling cards then be concerned your participants aren't really engaged in the conversation. Be concerned that they don't really believe what's being written. Be concerned that they have information they could be sharing by changing arrangement or text on cards, but aren't. Try to make the environment safe to do so by inviting them to write a card or tear one up. I often pretend to be busy with something else and ask them to do so, or invite someone who suggests a change of card positions to do it themselves. Once they start, they don't stop.
• Too many cards like too many ideas and too many words are hard to work with. Consider narrowing the scope of the conversation or raising the abstraction level to talk less about details and more about general concepts. Use different colors of cards to add more channels of communication. Identify focal or key ideas in the model, and annotate the model appropriately.

Card models arrange themselves into several common shapes

When you begin to arrange ideas on index cards you'll find those ideas begin to take on a few common shapes based on the bigger idea communicated in the model.

Arrangements by affinity

Given a number of ad hoc ideas, its common to put similar ideas closer together, and dissimilar ideas farther apart. The result is a model composed of many clumps or clusters. This is a simple affinity diagram.

Some modelers arrange those clumps into neat columns.

Others arrange them into what look like disorganized piles. But, when a group is allowed to arrange cards based on how they feel they should be, often those piles aren't as disorganized as they look.

• Clusters close to other clusters often have affinity with each other.
• Individual cards or ideas may straddle two clusters effectively making them a bridging member of both clusters.
• clusters often, but not always, arrange themselves top to bottom by importance.
• clusters often, but not always, arrange themselves left to right chronologically.
• often a central or most important idea finds it's way into the center of the model, and other ideas cluster radially around it - sort of like the circle surrounding L'arc de Triomphe is Paris. Look for a central monument or focal idea in your model.

Chronological arrangements

Often, especially when modeling work we arrange ideas chronologically.

When building chronological card arrangements time moves left to right, and ideas are arranged according their occurrence over time. It's common to place ideas occurring at the same time along the same vertical alignment.

Decompositions

Decompositions show how ideas decompose into their constituent parts. A decompositions looks a bit like a visual hierarchy where an idea on a card, through discussion, is decomposed down to the parts that make up that idea.

When completed, a decomposition model can look a bit like a well arranged affinity diagram. But, where affinity diagrams are usually built bottom up from detail level ideas, decompositions are usually built top town by decomposing big ideas.

Ad hoc charts

At times it's helpful to distill information by arranging along quantitative axis. Choose a data element for an x and/or y axis and arrange cards or stickies to see how the information would look in a chart.

Mix model types

In all likelihood, models you build will mix different types of modeling approaches. For example it's common for chronological models to contain clusters or decompositions. Look for ways to combine modeling approaches to communicate what you want.

Annotate models

You can't communicate everything you need simply through card arrangements. Fix a model down on paper, and use pen to annotate it with additional notes and information.

• Circle clusters of cards and label them.
• Draw lines from cluster to cluster and label lines with information on how those clusters relate.
• Give the model a title.
• Note the scope of concern for the model.
• Write down the names of the people who created the model.