FUTURHIST : A DSS for building future possible histories as policy scenarios

(J.-P. Bousset, September 1996)

FUTURHIS Software was designed and built in Prolog langage by P.Bousset, in the framwork of the "Task C of the AIR3-CT94-1545 project", to  help searchers to assess different Policies Scenarios by simulating the decision-making logics of an Actors's group with a variety of conditions (events, public actions (1), ...).

ngae

What is the PROBLEM ?

Firstly, Tasks A and B of the research project have shown all the rural Actors are not similarly concerned by the institutional actions, and which may be not similarly concerned by the possible future events : their points of view about the effectiveness of these actions depends on the relationships perceived or believed by them among their future situation (and/or the one of their region) and the actions they may use to achieve their goals. In other words, a rural area is plural.

Secondly, a rural area is complex, because the regional Actors are inter-dependent : e.g., the point of view/strategy of an Actor A about an action "a1" may be related to the positions/strategies of the Actors B, C, ...(i.e. the accointances of the Actor A) about the actions "a2", "a3", ...

Thirdly, future is plural. Notably, because time does not show probabilities of events and actions that could make up the future alternatives. In other words, we have no experience with events happening according to various probabilities : they either happen or they do not.

What is the PROPOSED METHOD ?

To cope with these problems, in accordance with Schwartz P. and Godet M., I propose (2) to bring the future down to a manageable scale by simulating the functioning of an Actors's system within/without some possible events and actions, and with different levels of relationships among the events, actions, and actors's goals. In other words, I propose (a) to see a policy as a theory about the role of public actions for changing a Society, and (b) to build possible future policies by "rolling the dice of life". These are possible future histories that we will call policies "scenarios".

As such, scenarios will be not predictions. The focus of the building of these possible future histories is not on forecasting the future, or fully characterizing its uncertainty, but rather on bounding the uncertainty. Scenarios will be thinking communication devices for  identifying which events and actions could be the cause of some inconsistencies in the current policies, which ones could be the cause of alliance and negotiation strategies among the progressive actors, as well as which ones could change the point of view of the opposing actors into desirable one. And thus identify some ways that could improve the effect of the current policies, now and/or in the future.

By offering a basis for discussing and structuring possible future regional problems and solutions, scenarios can be viewed as continuations of decision analysis. Instead of striving for supreme rationality within a necessarily simplified view of the world, they permit to use a semi-rationality in which intuition and analysis combine to manage highly complex tasks. They do this by providing a sense of general direction, without being a precision compass.

A similar distinction is relevant when comparing scenarios and traditional forecasting : scenarios try to highlight the reasoning underlying a forecast, with explicit attention to sources of uncertainty. In other words, scenarios complement traditional forecast and cost-benefit approach, either by presenting best and worse cases, or by making forecasts scenario dependent in an explicit manner.
 

How to BUILD SCENARIOS ?

STAGE A : Define the current Regional Panorama, i.e :
(a) identify the public actions in terms of decision variables (goals, means, actions, ...),
(b) identify the main regional Actors and their accointances,
(c) modelize the decision-making logics of these Actors.

The modelling of the Actors's Decision-Making Logic consists of formalising their theories about the roles of some actions (and notably, of public policies) for changing their situation and the one of their region.

To do that, Futurhis allows you to build as many cross-impact matrix (so called "Cognitive Map(3&4)") as Actors, in which the lines and columns represent the concepts used for describing the economics environment, the Actors's situation and their goals, as well as the actions used/could be used by the Actors for achieving their goals. And where the cells represent the Actors's beliefs about the sense and the magnitude (5) of the direct impacts of each concept-line on each concept-column.

So as to make the management of these relationships easier, the used concepts and attributes must be gathered in 4 types of variables :
.P-concepts, that reflect possible alternatives or options from which the actor select its actions,
.C-concepts, that denote events, contextual situations and problems that interact (including the points of view of other Actors(6)), and that may be directly related to one or more  P concepts,
.A-concepts, that refer to internal and external actions(7), and that are directly related to a C-concept or to an other action,
.O-concepts, that refer to the goals of the modeled actor, and that are directly related to the A-concepts.

STAGE B : Define events, actions, and relationships to be simulated, i.e. possible beginnings of the future history....

The second stage of the scenarios building consists of defining possible beginnings of the future history, i.e relevant changes in current events, actions, and relationships among concepts to be simulated.

To do that, Futurhis allows you to identify key uncertainties whose resolution will significantly affect the most interesting variables by making the structural analysis of each cognitive map, and/or of the cognitive map of the studied Actors's Group (i.e the CM built by "piling" all the cognitive maps), and by searching for the causes of the motivity/dependence of each key variable (use the Decision Analysis.Search for ... menu (8).

STAGE C : Simulate the Decision-Making Process of each Actor facing each possible beginning of the future history, for identifying its possible future positions and strategies about the actions to be evaluated.

Many studies showing that, facing to problems and challenges, actors tend to gather and to use their resources for upon the society their points of view and solutions, the third stage consists of identifying the point of view and the possible strategy (i.e. interest and role) of all the actors about each action of each public institution (i.e. each studied programme), by simulating their decision-making logics within/without given contexts, actions, and with different level of relationships among concepts.

To do that, you must use the following way :
.for each possible event to be simulated
 (i.e. within/without some "A" and/or "C" concepts,
 and/or with different level of relationships among events, actions and goals)
 .for each Actor (and its accointances),
   .for each public action to be evaluated
     .search for the Actor's positions (9)), by using the forecasting analysis pulse process

Results of these simulations will be gathered in as many (KBF) Actors-Actions Positions matrix (10) as goals

In summary:

STAGE D : Identify the most contentious actions, by simulating a Global Evaluation Process

Because the public actions may have an impact on several goals, and all the actors may have not the same position about their effectivenness for achieving these goals, the highlighting of the actions that have to be changed (i.e. the 'what' of the possible future policies) has been achieved by running a discrete multicriteria analysis (DMA).

DMA consists : (a) of making a pairwise comparison among the ActorsÆs positions about the effectiveness of each action for achieving each goal, next (b), of ranking and clustering the actions from the frequency of their preference and their 'reject' (i.e. from their over and sub efficiency).

Normally, entry data of the DMA are in the Actors Positions Matrix (i.e. GAPMx.KBF files ; one for each goal X), which have been automatically generated by Futurhis when you have simulated  the forecasting analysis of each Actor about the impact of each action on its goals (ref. STAGE C).

Nevertheless, it is possible to run the DMA without modelling the Position-Making Logics of the Actors (i.e. if you have a lack of time for building the ActorsÆs Cognitive Maps ...), by inputing manually the positions taken by the actors in an GAPMX.APM file. Although, in this case, it is not easy to identify "in what context" and "why" actions could be changed, this way could be usefull for computing the global point of view of a focus group about the effectiveness of current and future public actions (and next, for simulating a discuss among its members ...).

Beside, note you can test the robustness of the final ranking by using the Evaluation.Define Story Beginning menu for modifying the actors's and goals's weights, as well as actions's preference and reject thressholds.

STAGE E : Identify the ways in which the most contentious actions could be changed into desirable ones, as well as which under-new actions could improve the regional situation, by simulating a concertation process

The fifth stage consists of searching for : (a) the actors that could be the best supporters of each action, (b) the actors that have to be better convinced of the effectiveness of a given action, and (c) the perceptions and beliefs that have to be changed. In other words, the 'who' and the 'how' of the possible future policies.

To do that, Futurhis allows you :

1.to classify the actors from the frequency of their preference/reject for each action, and to identify the possible alliance's strategies among them (12) and the political interest of a negociation (or the negociation progress) about the content of the contentious actions, by analysing their position distances (ref. Kendall Coefficient of Concordance W, HAC, and results of appropriate Strategies Analysis Processes).

2.to identify perceptions and beliefs that are the causes of the sub-efficiency/reject of the most inconsistent actions (i.e. those having to be changed), by analysing the results of the Forecasting Analaysis of the actors that have most "rejected" these actions (i.e. by scanning the a*.RES files corresponding to the actors having to be better convinced of the effectiveness of these actions (13)).

3.to test the political interest of the changes in these perception and believes, by modifying the appropriate Cognitive Maps, running again the Forecasting Analysis of the coresponding actors as well as the Global Evaluation Process, ... until you are satisfyied ...

How to DISCUSS SCENARIOS ?

These are procedures that would probably not occur to anyone without exposure to the use of cognitive maps. It is precisely this potential for generating accurate descriptions, rational explanations, and new (and even counterintuitive) propositions about persuasion and consensus-building in situations with actors of highly divergent political perspectives that recommends the cognitive map mining technics for the prospective studies.

It is not thus intended to generate the best final action/policy of tomorrow. But by making more transparent both positive and negative ripple effects of any events or actions, this approach may generate some focus of attention that have cognitive implications in the building of the future policies. And By taking into account the diversity of the beliefs of the actors, as well as the possible conflicts that could appear among them, this methodology takes place in the process and comprehensive evaluation methods (14) : it is near to the Judgment Impact Matrix (JIM (15)).

So, the most interesting stage of the scenario approach consists of the comparison of a large number of more or less "risked" scenarios, rather than in the searching for the "best plausible" scenario.

To do that, for each pair of scenarios x-y, compare :
.the utility rank-order of each public action,
.the possible position of each actor about each institutional action,
.the possible alliance strategies among the actors,
.the possible evolution paths of the perception and believes of the actors having to be better convinced of rhe effectiveness of the actions, as well as the politico-economics challenges of these evolutions,

How to LINK SCENARIOS to POLICY FORMULATION ?

1. Firstly, see scenarios as futures against which you intend to evaluate current strategy. Where each ambiguous path constitutes a possible negotiation point, that may be the starting point of possible coalitions around the "progressive" actors, for solving some current and/or future problems, and/or for avoiding some future ones. And where institutional actions that support the alliance strategies of the "progressive" regional constitutes ways by which the effectiveness of the policies could be improved.

2. Secondly, integrate the scenarios in the strategic planning processes of the public institutions by using the two following steps :
.build a Key-Success-Factor matrix, where each line i and column j respectively represents one of the studied institutional actions/strategies and one of the possible beginnings of the future regional history, and where each cell KSF(i,j) represents the point of view of the searcher about the rank-order of the effectiveness of the action i within the context j;

.then, ask the Actors to rank-order the key success factors that are crucial for achieving their goals (ref. Delphi approach).

Once the KSF matrix has been completed, it can be checked if some KSF occur more than once in a given row (implying robustness) or in a given column (implying synergies). Such multiple occurrences of KSF might in turn provide a basis for a sound strategic vision, which focus on doing a few key things very well.

Nevertheless, the translation of each scenario from the "history" level to the "political" level has to be partly intuitive, through the mind of the searchers ...

(1) i.e. programmes
(2) ref. unpublished working papers : "Un système Expert pour réfléchir sur l'améliorations des politiques actuelles des institutions publiques", Bousset J.-P., Caen, Septembre 1994 ; "Prospective Stage : Needs of the Expert System", Bousset J.-P., Heathrow, December 1994 ;"The Orkney's future histories : an example of cognitive map mining using Futuris", Bousset J.-P., Kikwall, June 1995; "About the Prospective Model", Bousset J.-P., Galway, November 1995; "Scenario thinking", Bousset J.-P., Morella, March 1996.

(3) R.Axelrod, " Structure of decision, cognitive maps of policy elites ", Princeton Univesity Press, 1976
(4) W.R..Zhang, S.Chen, W.Wang and R.S.King, "A cognitive-map based approach to the coordination of distributed cooperative agents", IEEE, vol. 22, no1 Jan/Feb 1992
(5) by using Negative Positive Neutral fuzzi values (e.g. 0, +0.5, -1.5, ...)
(6) the acronym of the concept position must be organized as "ckxy", where xy represents respectively the second and the last letters of the CMap File k.

(7) e.g. each action of the studied public institutions (i.e. studied policies)
(8) ref."The Orkney's future histories : an example of cognitive map mining using Futuris",Bousset J.-P., Kikwall, June 1995;
(9) i.e. the strategies that could allow to build one of the possible continuation of the future regional history
(10) M.Godet in "De l'anticipation à l'action, manuel de prospective et de stratégie " Paris, Dunod ; and J.Hart in " Structure of decision : Comparative cognition", R.Axelrod Ed.,

Princeton Univesity Press
(11) ref. Roy, B., "Des critères multiples en recherches opérationnelle", Paris, Economica (1985) ; Saaty, T.L., The Analytical Hierarchy Process", Pittsburgh, University Press (1988).
(12) "About the Prospective Model", Bousset J.-P., Galway, November 1995
(13) ref."The Orkney's future histories : an example of cognitive map mining using Futuris", Bousset J.-P., Kikwall, June 1995;
(14) E. Monnier, CEOPS-Europôle, Vaulx en Velin ; Hill, B., Young, A.A. and Brookes, G., 1989 : Alternative Support Systems for Rural Areas. Report to the DoE and MAFF. Wye College.

(15) ref. McAllister D.M., "Evaluation in Environmental PLanning", Ed Th MIT Press, Cambridge, Massachussetts, 1980