IBIS

Thinking Tools5.2 Selecting the Right Decision ToolsAlso known as Issue-Based Information System, Dialogue mapping

Developed by Werner Kunz and Horst Rittel (Issues as Elements of Information Systems, 1970), later carried into software and meetings by Jeff Conklin (gIBIS, with Michael Begeman, 1988; Dialogue Mapping, 2006).

Maps a deliberation as questions, candidate answers, and the arguments for and against each, so the question is named first and the reasoning outlasts the meeting.

A procurement decision recorded as a question, three answers, and the arguments on each

The question comes first; every answer proposed hangs from it, and every argument for or against hangs from its answer. A contested argument opens its own question. Select each box to read it, and mark the two arguments that carried the decision. Illustrative.

0 of 2 marked1 of 2 marked2 of 2 marked

Select a box to read it, then mark the two arguments that carried the decision.Should we centralize procurement? The question the meeting had to answer, written before anyone argued for an answer. The group chose the hybrid: central contracts for the common items, with orders still placed by each plant.Centralize all buying. Proposed by finance. Outweighed in the end: its main argument is captured by the hybrid, and its main objection is not answered.Volume discounts near 8%. Finance's estimate from pooling the six plants' spend on the same items.Plants lose same-day local buys. Plant managers say a central desk cannot buy a pump seal by noon. Contested: how much buying is like that?How much spend is urgent and local? A question opened by a contested argument, so the dispute gets settled with data.About 6% of spend, by invoice data. A small share, and it matters a great deal when a line is down. Any answer has to keep it local.Keep buying at each plant. Proposed by the plant managers. Outweighed: it gives up the discount to protect a need the hybrid also protects.Plants know their suppliers. Long relationships get faster service on short notice.The same items at 30 prices. The invoice review found thirty different prices paid for one standard bearing.Central contracts, local orders. Proposed in the meeting. Chosen: it keeps the discount on common items and leaves urgent buying with the plants.Keeps most of the discount. The common items are about 70% of spend, enough for most of the volume discount.Plants keep urgent buys. Answers the objection to full centralization directly.Two buying routes to run. Real, and smaller than either alternative's cost. Recorded so a later review can check it.

An example

A manufacturer with six plants had argued about centralizing procurement at three quarterly meetings. Finance pressed for a single buying desk; the plant managers said a central desk would leave a line down while it waited for a part. Each meeting ended where it began, and nobody could say afterward what had been agreed.

At the fourth, the operations director mapped the discussion live on a screen. She wrote the question first, then each answer anyone proposed, and hung every argument for and against under its answer. When the plant managers' objection was disputed, she opened it as its own question, how much spend is urgent and local, and the group agreed to settle it with invoice data. The data showed about 6 percent. A third answer came up in the room: central contracts for common items, with plants still placing their own orders. It kept most of the discount and the urgent buying, and the map, saved with the minutes, showed a new CFO a year later exactly why.

Why it works

Meetings on contested problems go in circles because people argue answers before agreeing on the question, and because the reasoning lives only in memory. Writing the question first makes everyone respond to the same thing. Hanging each argument under the answer it bears on shows at a glance which answers have support, which objections stand unanswered, and where the group is talking past itself.

Opening a sub-question wherever an argument is disputed moves the fight off personalities and onto something that can be settled, often with data. And the finished map is a record: it lets someone who was not in the room see what was considered, what was rejected, and why, which is what an organization needs when it revisits a decision.

The thinking behind it

Horst Rittel Horst Rittel, a design theorist, created IBIS, the Issue-Based Information System, with Werner Kunz, in Issues as Elements of Information Systems (Working Paper 131, Institute of Urban and Regional Development, University of California, Berkeley, 1970). It records a deliberation as issues phrased as questions, the positions that answer them, and the arguments for and against each, so a planning group can see what has been argued and what is still open. Rittel also named wicked problems, in a 1973 paper with Melvin Webber.

Jeff Conklin Jeff Conklin, a design rationale researcher, built gIBIS with Michael L. Begeman, a graphical hypertext version of IBIS described in gIBIS: A Hypertext Tool for Exploratory Policy Discussion (1988), for recording the questions, positions, and arguments behind a design. He later turned IBIS into a way of facilitating meetings by mapping the discussion live, set out in Dialogue Mapping: Building Shared Understanding of Wicked Problems (Wiley, 2006).

How to run it

  1. Write the question first. Phrase what must be decided as a question everyone can see, before any answer is argued.
  2. Record every answer proposed. Write each position under the question as it comes up, without judging it.
  3. Hang each argument on its answer. Mark it for or against, in the speaker's words, under the position it bears on.
  4. Open a sub-question when an argument is disputed. Turn the dispute into a question of its own, and agree how it will be answered.
  5. Map the meeting live. Have one person draw the map on a shared screen while the group talks, and let people correct it.
  6. Strike what is outweighed. Mark the positions whose arguments have been answered or overtaken, and say why.
  7. Save the map with the decision. File it with the minutes so the reasoning can be audited later.

Where it goes wrong

The question is an answer in disguise. "How do we centralize procurement?" has already decided the issue. Rewrite until people who disagree can accept the wording.

The mapper takes sides. A facilitator who paraphrases one camp's points better than the other's loses the room. Write arguments in the speaker's words and ask them to confirm.

The map becomes the meeting. Some groups polish the diagram and never decide. Set the time for deciding in advance, and record the open questions that remain when it comes.

When to use something else

Where it sits in SOLVE eX

Go to the source

  • Issues as Elements of Information Systems, Working Paper 131, Institute of Urban and Regional Development, Berkeley, 1970
  • gIBIS: A Hypertext Tool for Exploratory Policy Discussion, ACM Transactions on Office Information Systems, 1988
  • Dialogue Mapping: Building Shared Understanding of Wicked Problems, Wiley, 2006

Back to the catalog