Information Technology Reference
In-Depth Information
In this setting, the Producer manufactures a fixed quantity of product each step in
the simulation, but that quantity is weighted across the 'active set' according to the
Producer's value system. There are three weighting rules that we have explored:
1. Minimize cost—essentially focused on exploiting the benefits of learning-by-
doing.
2. Maximize performance—a proxy for providing the most utility to Consumers
without regard to cost as long as it is below the feasibility threshold.
3. Maximize performance/cost ratio—a middle ground between (1) and (2). This
corresponds to profit maximization in micro-economic models.
The graphs in Fig. 11.5 show cumulative production by product design index for
a single run for each of the three rules but with the same initial conditions. The
stopping condition for each run was either 3,000 time steps or producing 500 units
of the design with maximum utility under that rule. Thus the production quantities
are different between runs because they reached the stopping condition at different
times.
The diagrams in Fig. 11.6 show how design trajectories differ under the three
Producer rules. All 112 possible designs are shown in the space as a grey dot
(potential), or other dot as described in Fig. 11.6 . The layout of the space is
suggestive of proximity between designs from the Producer's point of view—i.e.
proximity of production recipes. This layout was created using distance to
12 nearest neighbors and a graph layout algorithm—spring-electrical embedding.
Other methods were tried, including Multi-dimensional Scaling, but none produced
a
Cumulative Production by Product Design
Quantity
Prod u ced
400
300
200
100
0
1
112
Product Design Index
Fig. 11.5 Results of three runs with different valuation rules for Producer acting alone (Setting 1)
(a) Rule 1 (b) Rule 2 (c) Rule 3
 
Search WWH ::




Custom Search