First developed by Bertrand de Jouvenel in the 1970s, the procedure given here is a summary
of the description in Hugues de Jouvenels 2000 article . This
article was part of a special issue of Technological Forecasting and Social
Change devoted to scenarios.
Unlike most of the other procedures in this collection, Hugues de Jouvenel did
not give a short summary of the procedure. As a consequence, in addition to the
basic steps, there are several Notes that give advice, warnings, etc. that
appeared in reference .
Define the problem and choose the time horizon
Note: A good timeline should take into account: (1) inertia and the need to smooth over
temporary fluctuations; (2) the time required to implement a strategy; (3) the
degree of rigidity and motivation of the actors.
Build the system and identify key variables
Preferably, the list of variables is drawn up by more than one person.
Note: Do not mix up run-of-the-mill variables with highly
specific ones; include as accurate a definition of each variable as possible.
Analyze the relations between the variables (e.g., with a cross-impact matrix)
Collect the data and draft the hypotheses
For each driving variable ask: (1) What is the past development of this variable? (2) What is
its trend? (3) What potential curves or breaks could interfere with the trend?
To answer the questions, must address five issues: (1) Which indicators best represent
the driving variable? (e.g., GNP per capita as an indicator of standard of living);
(2) What data are available, either qualitative or quantitative? (3) What historical time
period should be studied to develop a trend? (The trend is often very sensitive to
the choice of time period.) (4) How to interpret past developments? Will the causal factors
of the past continue into the future? (5) On whose opinions are these conclusions based?
Scenarios consist of: (1) the base (current state and trends); (2) possible paths into
the future, built using an if this, then that format; (3) the last images at
different periods, the results of the paths.
Note: Generating the final image is not more important than generating
the paths leading to it.
Note: Do not give a snapshot image of a future year without any
consideration of the path leading to it.
Note: Approximation is good, but vagueness is bad. For example, economic
growth should reach X% between 2000 and 2005 - does that mean it probably will?
Or that it is best if it does?
Note: The scenarios described here are exploratory (or forecasts), rather than
strategic (or backcasts).
Develop strategic choices, based on the insights given by the scenario exercise.
 de Jouvenel, Hugues. 2000. A Brief Methodological Guide to Scenario Building,
Technological Forecasting and Social Change65, 37-48.