For the past 15 years, we have carried out research in non-quantified modeling, with special emphasis on a method called general morphology or general morphological analysis (GMA). To date, we have employed this modeling method in over 100 projects, involving policy analysis, scenario development and strategy management.
What is general morphology? The term morphology comes from classical Greek (morphe
) and means shape
. Thus morphology is the study of the shape and arrangement of parts of an object, and how these parts "conform" to create a whole or Gestalt. The "objects" in question can be physical objects (e.g. an organism or an ecology), social objects (a social system or organisation) or mental objects (e.g. word forms, concepts or systems of ideas).
Specific forms of morphological analysis are used in a number of scientific disciplines - for instance, linguistics, zoology and geology - in which formal, structural relationships are as important as quantity and function. However, what I am presenting today is "general morphology" - i.e. morphological analysis that is not associated with any specific discipline.
The research program initiated in the early 1990's was aimed at developing a methodological framework for creating models of systems and processes, which cannot be meaningfully quantified. With this in mind, we began by attempting to develop an extended form of what is called typology analysis
(see below). Initially, we thought we were doing something new. However, we subsequently learned that extended typology analysis was invented as early as the 1930's by Fritz Zwicky, professor of astronomy at the California Institute of Technology - the famous Caltech in Pasadena.
Most of you have probably never heard of Zwicky, but 40 years ago he was a well known scientific personality. He coined the term supernova, was the first to hypothesize the existence of neutron stars, and is regarded by some as being the father of the modern jet engine. He developed a general form of morphological analysis in order - among other things - to categorise and hypothesize new types of astrophysical objects, to develop jet and rocket propulsion systems, and to study the legal aspect of space travel.
Slide 1: General morphological analysis
GMA is a general method for structuring and analyzing complex problem fields which 1) are inherently non-quantifiable; 2) contain non-resolvable uncertainties (both antagonistic and non-specified uncertainty); and 3) cannot be causally modeled or simulated in a meaningful way. Instead, a judgemental approach must be taken. The question is: can we put a judgemental approach - carried out in groups of subject specialists -- on a sound methodological and scientific basis? We think that this can be done with the non-quantified, but highly structured method of, morphological analysis.
Messes, problems and puzzles
Before going into morphological analysis as such, it is a good idea to discuss what MA is good for, and what it is not good for
. In his book from 1974 titled "Redesigning the Future", operational analyst Russell Ackoff defined three levels of complex problems.
Slide 2: Messes, problems and puzzles
The first level he called a mess
(also known as a Wicked Problem
). A mess is a complex issue, which does not yet have a well-defined form or structure. When you have a mess, you don't even know for sure what the problem is. Here is an example of a mess, that our National Rescue Services Agency asked us to help with some years ago: What are we going to do with the Swedish bomb shelter program now that the cold war has ended?
This is complex issue which concerns money, technology, ethics, politics, everything! And all of these different aspects must be treated together -and dealt with as a whole.
All of the really important issues in the world start out as messes. And all of us come into contact with messes on a daily basis.
The next level is what Ackoff calls a problem
. This is an issue that does have a defined form or structure; it is dimensioned; it has variables and we know something about how these variables interact. But it does not have any one, single, clear-cut solution. As long as it is a problem - in Ackoff's use of the term - it has many different, alternative solutions "depending on". Depending on, for example: how much money we have; what type of technology is going to be available; who is in political power; what the weather is going to be like? Since we may not know these things yet, we have to leave the problem's solution open to different hypotheses about how the future might turn out.
The last level is called a puzzle
. A puzzle is a well-defined and well-structured problem with a specific solution that somebody can work out.
Morphological analysis was explicitly developed to work at the level of messes
. More specifically, GMA is used to turn messes into (structured) problems. In the process, we build up a virtual problem laboratory where we can generate alternative solutions depending on different hypothesized input conditions. In a sense, we build a non-quantified input-output model, in which we can define independent and dependent variables, test certain conditions against others, and hypothesize relationships.
The relationship between messes, problems and puzzles is summed up beautifully by Michael Pidd in his book "Tools for thinking" (1996):
"One of the greatest mistakes that can be made when dealing with a mess is to carve off part of the mess, treat it as a problem and then solve it as a puzzle -- ignoring its links with other aspects of the mess." (p. 40)
This type of mistake is made all the time, because we humans do not like to be confronted with messes for any length of time. Inherently, we are puzzle solvers. We want to get out of the mess as quickly as possible. However, this can have dreadful consequences.
When we do a morphological analysis, we want to examine the whole mess first, stalk out its boundary values and study its possible internal relations - before going on to generate alternative solutions, and then to solve puzzles.
Typologies and morphologies
Slide 3 (below) is an example of a so-called fourfold table. When you have two simple variables - each variable expressible in two ways, e.g. yes or no - you put them together in a fourfold table and check out the possible combinations they produce. This is a fourfold table for Landsteiner's ABO blood-type system. In this system, you either have base substance A in your blood or not, AND you either have base substance B or not. Four possible blood types are thus defined -- or hypothesized. At one extreme, if neither A nor B is present, you have blood type 0, the universal donor. At the other extreme, if both are present, you have type AB - the universal receiver.
Slide 3: Four-fold table (ABO-system)
A four-fold table of this kind is the simplest form of a typology. A typology is a simple model based on the possible combinations obtained between two or more variables, each variable containing a (finite) range of discrete values or conditions. Each of the possible combinations of variable-values is called a constructed type
Observe a few things about this typology. Like a little Cartesian coordinate system, it utilizes (two) spatial dimensions - height and breadth - to represent the ranges of the two variables. A typology of this sort is the simplest possible form of a model. You may not think of it as a model, but it is. It has inputs and outputs; and it gives you the possibility to hypothesize relationships and generate theory.
Slide 4 (below) represents the same model, but in the form of a morphological field
instead. In a morphological field, you put your variables up at the top of the columns, and list their values underneath. So, in Slide 5, we see two fully equivalent models of the ABO-system, with equivalent representations of blood type A.
Slide 4: Simplest morphological field
Slide 5: Equivalent field configurations
But here is the point: As long as you are only working with two variables, typologies are fine. But what if you want to work with more than two variables? Remember: a classical typology uses spatial co-ordinates to represent its ranges of values. What happens if you want a three-variable model?
The three-dimensional typological field on the left side of Slide 6 (below) is called a Zwicky box. Zwicky used it to demonstrate the advantages of morphological fields over classical typological fields. On the right is the corresponding, three-dimensional morphological field. Note that the blue cells in the morphological field represent the blue point on the x-, y- and z-axes in the typological box.
Slide 6: Two forms of 3-dimensional configuration spaces
Four-dimensional fields (employing so-called embedded variables) are used in typology construction, but this is about the effective limit. However, morphological fields are not constrained by spatial dimensions, and are thus not limited to three or four variables.
We work with morphological fields of up to a dozen dimensions, which may define hundreds of thousand, or millions, of different constructed types or configurations.
Fritz Zwicky worked in the 1930s, 40s and 50s when there were no modern computers. He died in 1974 and general morphology more or less stagnated at that point. Today, fast, small computers with advanced graphic interfaces have revolutionized morphological analysis. Now, with computer support, we can develop complex, non-quantified inference models and scrutinize the many thousands of relationships in a time frame that was impossible for Zwicky.
General morphological analysis as a process of collective creativity
We use GMA primarily for developing scenario and strategy laboratories, for structuring and analyzing policy spaces and for relating means and ends in strategic and operational planning. It is also excellent for carrying out so-called positional or stakeholder analysis - which we feel is a necessary complement to cost-benefit analysis. Our clients include the Swedish Total Defense authorities, Swedish EPA and Ministry of the Environment, the National Rescue Services, the International aid and Development Agency (SIDA), Dstl (U.K.), DoD (U.S.) and a number of international academic and non-governmental organizations.
For our part, we developed computer support for morphological analysis - the so-called MA/Carma process - with the following conditions in mind:
- The method should be process and group oriented. In other words, it is the process that the subject-specialist group goes through in doing morphology that is the most important result, not what comes out of the computer.
- It should be generic. GMA is a general method for non-quantified modeling. It sets no specific preconditions on the working group. You start with a blank slate, and the group successively builds up a non-quantified model.
- It should be transparent. There are no black boxes. All the cards are on the table. You cannot hide anything. Another way of saying this is:
- It should leave an audit trail. You should be able to trace what you have done and how you have come to your conclusions. Although judgemental processes will never be as traceable as, for example, a mathematical proof, MA allows for as much traceability as is possible under such circumstances.
- It should be easy to update. If we work with a complex problem area, and come back to our client a year later, and somebody says .. "a new dimension or variable has become important" .., we can build this new parameter into the prior work, without having to start from scratch.
How, then, does one carry out a morphological analysis? First, you have to have a mess to work with. The example that I am going to use for this presentation is the "bomb shelter mess" I mentioned earlier - i.e.: What are we going to do with the Swedish bomb shelter program, now that the cold war is over?
The second thing you need is a small group of subject specialists -- no more than 5-7 people. Ideally, the group should be heterogeneous, representing different aspects of the issue involved. In the "shelter group", we engaged people representing financial, technical, political, military, security policy and ethical aspects of the issue.
A morphological analysis is carried out in a number of iterative steps, in which the subject-specialist group goes through a number of analysis-synthesis cycles. The time needed for such a study depends on a number of factors: the severity of the "mess", the level of ambition, extent of coverage and/or resolution, the desired end-result, etc. For instance, the end-result can vary between simply structuring the problem field, to developing a expert validated, non-quantified input-output model.
The first step (Slide 7, below) is to define the primary parameters of the problem complex. For instance, concerning the shelter issue: Where are we going to build shelters? Who and what do we actually shelter? Size and degree of cramming? What are we going to do with new construction and maintenance? Finally, what is the general philosophy behind the bomb shelter program? (This is only a segment of one of the fields developed in the shelter study. In reality there were more parameters than those shown here -- including economic and threat parameters. We use this reduced field as a pedagogical example.)
Slide 7: Define important dimensions of problem complex
Defining the 6-8 primary parameters of a problem complex may seem relatively simple, but in fact is a difficult and fascinating task. We have carried out some 80 projects in morphology during the last 10 years, and have learned that it takes considerable time and a lot of discussion to agree upon the most important variables. However, 6-8 parameters - if chosen and molded carefully - suffice to get a good grip on even the most complex problem areas. If you want more detailed coverage, more parameters can be added later. However, there is no such thing as 100% coverage. Reality presents us with infinite dimensionality. We are working with models of reality, which are never complete.
The second step (Slide 8): For each parameter, define a spectrum of values - or what we call conditions - which represent alternative solutions to the particular issue that the parameter expresses. Sometimes this is a scale, as in "Geographical priority"; sometimes a complex binary combination such as "Size and degree of cramming"; sometimes small scenarios or idea-packages such as the four points under "Shelter philosophy". If you are working with a scale, start with the boundary values. In general, it is good practice to draw out the boundaries of the parameters as far as possible, and attempt to define the extreme limits of each variable. (Things that you think are extreme or silly now, have a habit of turning into reality tomorrow.)
Slide 8: Define spectrum of possible "values" for each dimension
The totality of the parameters and their respective values is a morphological field
. When you get a good field defined, the working group is really happy, because this represents your problem universe. It might be six variables or eight or ten. It is good to keep the field relatively small at the beginning. You can always expand it later.
There is an enormous amount of potential information in a well-developed morphological field, especially as concerns how the different parameter conditions are related to one another. Remember, this is not a table. It is a multi-dimensional configuration space. Within this space, we want to start defining configurations (corresponding to constructed types
in a typology) -- that is, combinations of conditions, which represent different formal solutions to the problem complex. For instance, here is one solution (Slide 9, below).
How many solutions or configurations are there in a morphological field? This is easy to calculate: simply multiply together the number of conditions under each parameter. In this case 4x4x4x3x3x4 = 2,304 possible configurations. This is a relatively small field. Normally, when we work with 6-8 parameter fields, we may have millions of possible (formal) configurations. The point is, that there are far too many to look at and check by hand. Somehow, we must be able to reduce the number of configurations in a field, so that only those that meet certain criteria remain. The main criterion is, that a configuration is internally consistent, i.e. that is does not contain conditions that are mutually contradictory.
Slide 9: Single field configuration of 2304 possible
Take a look at Slide 10 (below). If everyone is going to have the "same shelter quality", then you cannot have any "geographical priority". This is a logical contradiction. Thus any and all configurations that contain any of these pairs of conditions can simply be thrown out. They are logically inconsistent. In this case, the field is reduced by 432 configurations, c. a fifth of the total number of formal configurations.
Slide 10: Contradictory parameter values
Classical morphological fields are full of contradictions, both logical and empirical, which must be identified and weeded out. In fact, most morphological fields can be reduced by up to 90 or even 99 percent. This reduction leaves us with a manageable number of configurations - i.e. solutions - to examine and work with.
How do you reduce the field? You do this by comparing each condition with every other condition, and asking the question: Can these two conditions coexist? This is done by doing a cross-consistency assessment with the help of a cross-consistency matrix (Slide 11).
Slide 11: Cross-consistency matrix
The cross-consistency assessment does three things. Firstly, it functions as a garbage detector
. The garbage it detects is in the form of poorly defined concepts, concepts with different meanings for different participants; different terms meaning the same thing, etc. This and other forms of garbage are immediately revealed when put to the test of a cross-consistency-assessment. Then, one must go back to the morphological field and review the concepts, redefining and adjusting, adding and subtracting, parameters and conditions. When everyone is in agreement about what the content of all the cells in the field mean, and the cross-consistency assessment starts working, then you have an acceptable prototype field.
Secondly, the cross-consistency assessment represents a deep dive into the problem complex. Everyone working through the field in this way learns something new - since the whole point is to relate different aspects of the problem complex to each other. No one has a total mastery of the type of complex problem fields we usually work with. The cross-consistency assessment forces a dialogue between different areas of knowledge.
Finally, when you have completed the assessment, the third function is to reduce the field. We execute a function in the software developed to support morphological analysis, and the computer does in a matter of seconds, that which took Fritz Zwicky weeks to do by hand. The field is reduced and we are left with a list of the surviving configurations. This list represents the solution space
of this particular problem complex. (Note that the cross-consistency matrix can be changed and recalculated, in order to perform a sensitivity analysis.)
Slide 12: One configuration within solution space
In this case, with our "bomb shelter exercise", we are left with 125 solutions out of the original 2304 formal configurations (Slide 12). We could go right down the list and look at one after the other. However, with computer support, we can do much more. For instance, we can define any particular variable as a driver, or even select several variables as multiple drivers.
For example: If you want to see all the solutions that include the condition "Everyone gets the same shelter quality", we can treat this particular condition as an "input" by locking it, and obtain a resultant outcome space (Slide 13a). In a morphological field, any parameter (or multiple parameters) can serve as a driver (or "independent variable"). Slide 13b shows a multi-driver input (red) and a clustered output.
Slide 13a: Solution space (blue) for single driver input (red)
Slide 13b: Solution space for multiple driver input
This field segment was presented as an exercise only. Let's take a look at something more complex.
Complex Scenario-Strategy Fields
GMA projects typically involve developing conceptual laboratories, which allow us to relate (external) scenario fields to (internal) strategy alternatives. We have developed such laboratories, inter alia, as instruments for generating threat scenarios for the Swedish National Defense; for identifying alternative long-term social evolutionary trends for the Swedish Nuclear Waste Management agency; and for evaluating risk (including terrorist threats) and alternative preparedness measures for the Swedish Nuclear Power Inspectorate and the Radiation Protection Agency.
We have found GMA especially suitable in the area of environmental policy studies, in which different strategies can be pitted against a range of possible futures scenarios. A case in point is a study done for the Ministry of the Environment concerning the development of an Extended Producer Responsibility system in Sweden.
Extended producer responsibility
) imposes accountability over the entire life cycle of products and packaging introduced on the market. This means that firms, which manufacture, import and/or sell products and packaging, are required to be financially or physically responsible for such products after their useful life cycle. They must either take back spent products and manage them through reuse, recycling or in energy production, or delegate this responsibility to a third party -- a so-called producer responsibility organization (PRO), which is paid by the producer for spent-product management.
In studies of this type, it is advantageous to develop two complementary morphological fields: one in which we can systematically develop different EPR strategies, depending on variables which can be more or less controlled (i.e. an "internal world" field); and the systematic development of a number of different possible futures scenarios based on factors which cannot be directly controlled by those developing the EPR-system (i.e. an "external world" field). These two fields can then be linked by cross-consistency assessments in order to establish which strategies would be most effective and flexible for different ranges of scenarios.
Slide 14 (below) is one of the fields developed for the EPR study. It represents a scenario matrix linked to three strategy alternatives developed from a strategy matrix. In this example, eight scenarios are defined which cover a wide range of global possibilities (listed in the first parameter on the far left). The three strategy alternatives (listed on the far right) were linked by cross-consistency assessments to each of the attributes in the scenario field. A forth strategy alternative -- "No strategy available" -- represents the case in which none of the three given strategies measures up to the demands of a given scenario.
In this configuration, strategy B is shown to be able to cope with 3 of the 8 hypothesized scenarios, whereas there is no strategy available to cope with 2 of the scenarios (light blue).
Slide 14: A scenario-strategy matrix for Extended Producer Responsibility study
Finally, the example in Slide 15 (below) is one of the fields developed in a study done for the Swedish Agency for Economic Crime and a number of other law enforcement organizations. We wanted to look at the complex structure of economic crime and relate this to different methods of mitigation. This was a real mess: economic crime concerns everything from me cheating on my income tax to organized drug trafficking to state-led international economic sabotage. The issues involved include: What types of crimes are involved? Who are the victims? What are the different methods for perpetrating such crimes? What types of controls or sanctions are available? What type of legislation could be used in alternative cases? And so forth.
Slide 15: Sample matrix for Economic Crime study
We have not yet carried out a morphological analysis concerning Information Technology and its relationship to developing social processes -- which is the subject of this seminar. It is hard to say how such a study would precede at this point, but it would be an enormously interesting undertaking.
* * *
Dr. Tom Ritchey is a former Research Director for the Institution for Technology Foresight and Assessment at the Swedish National Defence Research Agency (FOI) in Stockholm. He is a methodologist and facilitator who works primarily with non-quantified decision support modelling -- especially with General Morphological Analysis (MA), Bayesian Networks (BN) and Multi-Criteria Decision support (e.g. AHP). Since 1995 he has directed more than 100 projects involving computer aided MA for Swedish government agencies, national and international NGO:s and private companies. He is the founder of the Swedish Morphological Society, Director of Ritchey Consulting and a founding partner of the U.K. based Strategy Foresight Partnership.
(Download this article in PDF-format)
- Ackoff, R. (1974) Re-designing the Future, London.
- Bailey, K. (1994) Typologies and Taxonomies - An Introduction to Classification Techniques, Sage University Papers: Sage Publications, Thousand Oaks.
- Coyle, R. G. (1994) Crawshay, R. and Sutton, L.: "Futures Assessments by Field Anomaly Relaxation", Futures 26(1), 25-43.
- Coyle, R. G. (1995) McGlone, G. R.: "Projection Scenarios for South-east Asia and the Southwest Pacific", Futures 27(1), 65-79.
- Coyle, R.G. and Yong, Y. C. (1996) "A Scenario Projection for the South China Sea", Futures 28 (3), 269-283.
- Doty, D. H. & Glick, W. (1994) "Typologies as a Unique Form of Theory Building", Academy of Management Review, Vol. 19, No.2.
- Pidd, M. (1969) Tools for Thinking, Wiley, London.
- Rhyne, R. (1981) "Whole-Pattern Futures Projection, Using Field Anomaly Relaxation", Technological Forecasting and Social Change 19, 331-360.
- Rhyne, R. (1995a) "Field Anomaly Relaxation - The Arts of Usage", Futures 27 (6), 657-674.
- Rhyne, R. (1995b) "Evaluating Alternative Indonesian Sea-Sovereignty Systems", Informs: Institute for Operations Research and the Management Sciences.
- Ritchey, T. (1991) "Analysis and Synthesis - On Scientific Method based on a Study by Bernhard Riemann". Systems Research 8(4), 21-41. (Download REPRINT)
- Ritchey, T. (1997) "Scenario Development and Risk Management using Morphological Field Analysis", Proceedings of the 5th European Conference on Information Systems (Cork: Cork Publishing Company) Vol. 3:1053-1059.
- Ritchey, T. (1998) "Fritz Zwicky, 'Morphologie' and Policy Analysis", Presented at the 16th Euro Conference on Operational Analysis, Brussels.
- Ritchey, T. (2003a) "Nuclear Facilities and Sabotage: Using Morphological Analysis as a Scenario and Strategy Development Laboratory". Adapted from a Study for the Swedish Nuclear Power Inspectorate, and presented to the 44th Annual Meeting of the Institute of Nuclear Materials Management - Phoenix, Arizona, July 2003. (Available for download at www.swemorph.com/downloads.html)
- Ritchey, T. (2003b) "MA/Carma - Advanced Computer Support for Morphological Analysis". (See: MA/Carma).
- Ritchey, T. (2004a) "Strategic Decision Support using Computerised Morphological Analysis", Presented at the 9th International Command and Control Research and Technology Symposium, Copenhagen, September 2004. (Available for download at www.swemorph.com/downloads.html).
- Ritchey, T. (2004b) "Wicked Problems. Structuring Social Messes with Morphological Analysis". Adapted from a lecture given at the Royal Institute of Technology in Stockholm, 2004. (Available for download at www.swemorph.com/downloads.html).
- Ritchey, T. (2005) "Futures Studies using Morphological Analysis". Adapted from an article for the UN University Millennium Project: Futures Research Methodology Series (Available for download at www.swemorph.com/downloads.html).
- Ritchey, T. (2006a) "Problem Structuring using Computer-Aided Morphological Analysis". Journal of the Operational Research Society, Special Issue on Problem Structuring Methods, (2006) 57, 792-801. (Available for download in PDF for JORS subscribers at: (Available for download at www.swemorph.com/downloads.html).
- Ritchey, T. (2006b) "Modeling Multi-Hazard Disaster Reduction Strategies with Computer-Aided Morphological Analysis". Reprint from the Proceedings of the 3rd International ISCRAM Conference, Newark, NJ, May 2006. (Available for download at www.swemorph.com/downloads.html).
- Ritchey, T. (2007) "Threat Analysis for the Transport of Radioactive Material using Morphological Analysis". Adapted from a paper presented at the 15th International Symposium on the Packaging and Transportation of Radioactive Materials, PATRAM 2007, Miami, Florida. (Available for download at www.swemorph.com/downloads.html)
- Ritchey, T. (2009) Developing Scenario Laboratories with Computer-Aided Morphological Analysis. Presented at the 14th International Command and Control Research and Technology Symposium, Washington DC. (Available for download at www.swemorph.com/downloads.html)
- Ritchey, T. (2010) "Specifying Training and Instruction Requirements using Morphological Analysis". Paper presented at ICELW 2010, Columbia University, New York. (Available for download at www.swemorph.com/downloads.html)
- Ritchey, T. (2011) Wicked Problems/Social Messes: Decision support Modelling with Morphological Analysis. Springer: Berlin. (First dedicated book to be published on GMA. See description at: http://www.swemorph.com/book.)
- Ritchey, T, Stenström, M. & Eriksson, H. (2002) "Using Morphological Analysis to Evaluate Preparedness for Accidents Involving Hazardous Materials", Proceedings of the 4th LACDE Conference, Shanghai, 2002. (www.swemorph.com/downloads.html).
- Zwicky, F. (1969) Discovery, Invention, Research - Through the Morphological Approach, Toronto: The Macmillan Company.
- Zwicky, F. & Wilson A. (eds.) (1967) New Methods of Thought and Procedure: Contributions to the Symposium on Methodologies, Berlin: Springer.