Anthropology - Part 2: Conducting the Meeting

Anthropology—Part 2: Conducting the Meeting

So you've decided what sessions to have, who to invite, and what to cover in each session. Now comes the actual conducting of these sessions.

Socratic Method

The best technique I've found is continual use of leading, open-ended questions. Unlike the typical Socratic dialog, which is meant to bring the questioner to a particular point, these questions should not presuppose the answers. You are there to find out.

Opening with a very general question about the topic will usually elicit one of two responses: either a glib definition or unbelieving silence. You could also open with a na ve question (even if you know better); for example, "So there really isn't much to this loan approval process, is there? You just look up the credit rating and make the determination." This will invariably result in the participants helping you understand that "it really isn't quite that simple."

Once the ice is broken, and you get them to talk, you'll want to start writing on the board, to create a public short-term memory. Don't try to draw the semantic models (see Chapter 10) during the meeting; I have found that to be distracting to the business at hand, which is mostly brainstorming.

Enumerating Types

Once you've identified a "type" of thing that seems to be somewhat central to the topic, such as "loans" from the mortgage example, it often helps to ask, "What types of ‘loans’ are there?" People seem to be much more inclined to contribute when they are enumerating variations on the type theme.

Splitters and Lumpers

In taxonomy, as in life generally, there have always been "splitters" and "lumpers". The splitters are most impressed by the differences between entities, and they therefore advocate small taxonomic entities and tend to make more species, more and narrower genera, etc. The lumpers are more impressed by similarities than by differences, and they will always prefer larger species and fewer, broader genera.^[56]

This process will turn over a lot of stones, very rapidly. At some point you want to shift to one of the following techniques to elaborate further.

Definitions

You will want to periodically ask for definitions of the terms being brought forward. You will find that most definitions are trite and aren't much use semantically, but they are good jumping off points for further clarifications. In particular, you will want to challenge why something isn't more like something else; you might say, for example, "So, ‘explanation of benefits’ is really just a set of error messages to explain why certain procedures were outside the scope of the particular medical plan." (This led to a lengthy conversation because most people in health care are convinced that explanation of benefits is something important and specific because they deal with it on an almost daily basis.)

Making Distinctions

As people list variations on some basic type, periodically you will want to interject with some distinctions. This is far easier than it sounds, and it generates a lot of new data that the participants often hadn't thought of themselves.

We were discussing receivables (monies owed to us) in the context of workers' compensation systems. One of the main sources of receivables in this domain is inadvertent overpayment of claims. One of the questions posed was, "What is the difference between a receivable on a claim overpayment and a pension overpayment?" This led to a series of comments that unearthed the fact that in some cases the pension overpayment concerned someone who is no longer living. This creates a very different scenario than the typical claims overpayment, where it is very likely that the payer has a continuing payment obligation against which the debt can be netted.

Using Semantic Primitives

We are building a set of semantic primitives based loosely on the work of Charles Peirce^[57] and Anna Wierzbicka.^[58] Peirce had a system of dividing everything in the world into a categorization involving 12 orthogonal axes of categorization. Wierzbicka's study of language has led her to believe that there are a small number of semantic primitives that are so fundamental that all people know them; they cannot be fully defined but are experienced as a by-product of being human, and all other concepts are built on them.

When the conversation slows, we ask people to attempt to categorize anything they've discussed into one of the following:

• Person

• Identifiable biologic entity (pet, tree, etc.)

• Organization

• Building

• Discrete made object (equipment, parts, etc.)

• Homogeneous made object (chemicals, etc.)

• Historical event

• Measurement

• Planned activity

• Intellectual property

• Document

• Agreement

• Relationship

• Transaction

• Category

• Generative rule

• Constraint

Most items are one of these or a composite of them. In almost all cases, the act of attempting to assign a proposed "type" of entity to a semantic primitive results in greater understanding of what is being said. It is important to make some of these distinctions semantically as the concepts are uncovered, otherwise the participants are going to be unsure of what is being discussed.

For example, in a discussion of work flow tasks, you need to ask whether the participants are discussing a task that has occurred (recording what really happened as opposed to what was supposed to happen), a task scheduled or assigned, or merely the category of a task (editing, for example, as opposed to editing a specific document, which could be a scheduled task). This may sound obvious, but this is exactly where most systems get confused.

Going to Extremes

It is also useful to test the boundaries. In a discussion about the definition of patient, you might ask, "When a patient dies, is he or she no longer a patient? When did they cease to be a patient?" "When a woman gives birth in the hospital, are there two patients or one?" "Does the patient have to be a human? A living thing?" "Is a tissue sample a patient? If not, what is it?"

Completeness/Wholeness

Somewhere around an hour and a half into this, you will want to begin checking for coverage and completeness. Given what you know about the topic, are there areas that haven't been discussed yet? You might want to write these on the board and include them in the conversation later.

^[56]Wim Vader, "Splitters and Lumpers," personal communication via email, May 6, 1998.

^[57]C. S. Peirce, On A New List of Categories. Proceedings of the American Academy of Arts and Sciences, 1868. http://www.peirce.org/writings/p32.html

^[58]Anna Wierzbicka, Semantic Primes and Universals. Oxford: Oxford University Press, 1996.