ScenLab is a high-end software tool that supports the so called consistency and
robustness analysis in the scenario process. It uniquely incorporate disruptive
events Wild Cards into the data analysis and is the only software that is capable
of processing even large projects on a desktop computer.
ScenLab is a product of evolve:IT, Scientific Software
Systems. evolve:IT is specialized in the development of genetic algorithm solutions.
The development of ScenLab was initiated and supported by Z_punkt GmbH, The
Foresight Company. Z_punkt is a leading european, german-based think tank for
Based on the research question key factors for the field of interest are identified.
Key factors are the main driving factors for the developments in the field of interest.
These key factors usually have more than one possible future projection, that is more than
one way how they could evolve in the future. For example the key factor 'economic growth'
can have the future projections 'high growth', 'low growth' and 'recession'. One of the
future projections of a key factor might be more plausible than others. To account for that
a plausibility value is assigned to each future projection of a key factor.
The aim is now to find the future projections of each key factor that are likely to occur
in the same future. Lets take the key factor from above and a second key factor 'Employment'
with the future projections 'full employment' and 'high unemployment'. It will be very unlikely,
that the future projections 'recession' and the future projection 'full employment' will occur
in the same future development. A scenario containing both these future projections will be
called inconsistent. But the idea of scenario building is to find consistent future developments.
How can we find consistent scenarios from the future projections of the key factors? In the above
example this is easy since we only have five future projections. But in a typical scenario process
we would have 15 to 30 key factors with 2 to 4 future projections each. In this case it is not
feasible to find all possible consistent sets scenarios – we need software support. Instead of
finding scenarios and the checking if they are consistent we can how ever determine which of the
future projections of one key factor are consistent with which of the future projections of all
the other key factors. This is done by comparing all the future projections with each other and
assigning paired consistency values to them. What we get is a matrix with consistency values.
Now ScenLab processes this consistency matrix to find the most consistent sets of future
projections – called projection bundles. These projection bundles are the base for writing
Furthermore ScenLab uniquely incorporates the plausibility values assigned to the future
projections to find not only consistent but consistent and plausible – robust – scenarios.
Even in medium sized projects the accumulated amount of data grows so fast that a standard algorithm
run on a desktop computer does not have enough computational power to find all the possible projection bundles.
Standard software implementing the consistency analysis usually breaks down if the number of key factors exceeds 20-25.
Using a genetic algorithm works around this problem by specifically searching for good projection bundles.
This unique approach allows to process even large projects with ScenLab.
A genetic algorithm uses the Darwinian concept of the survival of the fittest on a virtual population.
It starts out with a random population of individual, each representing a solution of the problem that
is to solve. In our case that is each individual represents a projection bundle.
Now, we know that a good projection bundle does not contain two future projections that are
inconsistent and has a high plausibility. We can use this prior information to assign a fitness value
to each of our projection bundles.
Based on these fitness values a environment is modeled in which fitter projection bundles have higher
chance of 'survival'. That is, using the genetics concepts of cross-over and mutation, they have a better
chance of propagating their information to a next generation. After a few hundred generations the
population will consist only of fit individuals, that is of projection bundles that full fill our
requirements of being consistent and plausible.
The number of projection bundles found by the consistency or robustness analysis is usually still quite large.
But the aim of a scenario process is to find three to five scenarios. ScenLab supports finding these scenarios
by various tools for analyzing the set of projection bundles.
The list of projection bundles can be shortened without loosing essential information by using several reduction modules.
The distribution of projection bundles can be visualized in a plot using a technique called multi-dimensional scaling (MDS).
This helps finding similar projection bundles.
Also, ScenLab is able to perform a cluster analysis on the set of projection bundles. This allows the user to identify
a good number of scenarios and to find the projection bundles that are relevant for these scenarios.
To compare projection bundles visually ScenLab incorporated a morphological matrix.
Wild Cards are disruptive events in that are usually highly unlikely to occur, but have high impact on the possible futures.
ScenLab uniquely incorporates Wild Cards that are defined by the user into the analysis and its data analysis tools enable to
find such projection bundles that represent scenarios that are robust to the occurrence of Wild Cards.