The Dynamics of Best Practices: A Structural Approach
Ignacio J Martinez!
University at Albany
Rockefeller College of Public A ffairs and Policy
135 Westem Avenue, Milne Hall 121-A, Albany, NY 12222
Tel: (518) 442-5238 Fax: (518) 442-5298
Email: imartinez@ gproyectos.com.mx
Luis F Luna
University at Albany
Rockefeller College of Public A ffairs and Policy
52E Knights Bridge, Guilderland, NY 12084
Tel/Fax: (518) 456-3456
Email: 118287@ albany.edu
Abstract
The Dynamics of Knowledge and Best Practices in a field is influenced by a series of factors,
amongst the most relevant are: (i) the dynamics of the population of practitioners of the field, (ii)
the theoretical and practical productivity of the practitioners, (iii) the policies related to
information and knowledge management in the field, (iv) the social judgment processes that take
place to consider a practice as a best practice, and to consider a practitioner as an advanced
practitioner and (v) the politics and power related forces acting in the field. The question that
drove the effort to build a formal simulation model of the Dynamics of knowledge and Best
Practices is: What underlying structure conditions the behavior observed in the dynamics of
knowledge in a field? Why the best practices in several fields are not necessarily related to the
best ideas? To address these issues a formal system dynamics model was built using Vensim. The
findings obtained with the model include the high impact of the mentoring and networking
activities in the development of knowledge and the critical influence of knowledge management
activities to consolidate knowledge in the field. A critical piece of understanding from the model
is how the dynamics are perceived and how this perception can be mistaken with the actual
knowledge.
Keywords: System Dynamics, Formal Simulation Models, Best Practices, Population Dynamics,
Modeling Process, Knowledge Management.
' Corresponding author.
Introduction
In the System Dynamics field, the single most powerful way of understanding the
behavior of a system is the creation of simulation models. The power lays in the capability of
taking into account the interactions of all variables in the model at the same time. In order to do
this, practitioners must determine the underlying structure of the system, capture it and construct
a model that replicates the behavior of the real system. System Dynamics also involves
understanding the world and helping other people (and groups of people) in that process. In order
to do that, SD practitioners should be able to build readable models from and for everyone.
Because the field has been expanding in type of systems modeled and number of
practitioners all over the world, the general practice of model development have been changing
over time. From the standpoint of some of the best SD modelers, the practice has been growing,
but the quality of the practice has been eroding through the time. Moreover, the lack of good
practices has promoted, from some experts point of view, a very limited growth in System
Dynamics core modeling knowledge (best practices). Despite the accomplishments of individual
practitioners, this practice diversity makes it difficult to broadly evaluate System Dynamics as a
modeling practice and can prevent the field from continued development (Scholl 1995). One way
to generate further insight on this matter is to build a formal model where the dynamics of best
practices are captured and understood in their relation with knowledge generation and practice
enhancement.
Altemative Models, Empirical Evidence Literature
Experience, Literature
‘a BO
Perceptions of Empirical and Infered
System Structure Time Series
Comparsonand STS Comparison and
System Reconciliations
Reconciliation Conceptualization
Structure Validating | Behavior Validating
Processes Processes
Model
Formulation
; agit Deduction of Model
Representation of, Behavior
Model Structure
Diagraming and
Description Tools
Figure 1. - The System Dynamics Modeling Process (Saeed, 1992, p.252)
Computing A ids
Figure 1 shows the SD modeling process (Saeed 1992) as a couple of loops of structure
and behavior validation. In the present paper this processes were followed in order to build a
model to analyze population, project and knowledge dynamics inside the SD field.
In any field of knowledge there is a beginning, growth, consolidation and presumably in
the long run (how long depends on the benefits of the field and the reinvention process of itself)
a decay leading to death. We can say that the SD Field is growing in number of practitioners,
2/2
type of practitioners, type of models addressed, software tools, purposes for modeling, etc. This
is generating fragmentation and the growth of different lines of thought around how System
Dynamicists should model and confront issues in the natural and social sciences. On this way,
the basic behavior of best practices generation and use is shown in figures 2a and 2b. As we
state above, some experts’ perception is that core knowledge has remained relatively constant
during the last years, but they believe that is desirable to this set of best practices to growth
(figure 2a). This reference mode could be enriched with the point of view of some knowledge
management experts (Pfeffer and Sutton 2000) who perceive that the use of core knowledge in any
field has and oscillatory nature. That is to A
say, people in the field use and forget to use core
the field core knowledge. Knowledge
(desired)
The best practices model
(knowledge!) has the purpose of (1) Gine
understand better the implications of the Knowledge
dynamics of best practices in a given field, (actual)
(2) raise consciousness on the relevance of
the issues amongst the individuals in the >
field, and (3) serve as policy making aid for 0 40
the community with respect to knowledge (a)
management and population dynamics issues. &
The model should be able to capture the Core
dynamics of the best practice in the field not Core Knowledae use Knowledge
only as a flow of them in time but the stock (desired)
of them in any given time and how do these
influence the promotion, learning, unity, Core
creativity and growth in a field of knowledge. ee
In the following sections, the structure >
and behavior of the Best Practices Model t 40
(Knowledgel) being developed are ()
presented. The model has been evolving and Figure 2, Basic dynamics of knowledge use and
changing over time following the iterative generation
nature of the process. Best Practices are
being considered in a sense as a part of the current paradigm (Richardson 1991) in the field that
they appear and are related to a number of factors including population dynamics and the
networking that occurs among individuals in the field. The effect of the way knowledge is
managed is being taken into account too. The paper concludes with some insights for policy
development within the field and with future directions of knowledge! further development.
3/3
The Best Practices Model (Knowledge!)
Model Overview
Knowledge is conformed by four main sectors (figure 3). The actual sectors of the
model are Human Resources, Project Development, Knowledge Generation and Knowledge
Management.
The Human resources sector contains the structure of population dynamics and mentoring
and training processes. Although the initial conceptual analysis includes the culture and values
of the people networks within the field, Knowledge1 does not include these dynamics. From the
authors’ point of view, the nature of network dynamics and knowledge transfer pose a challenge
similar to the one presented by Hines (Hines and House 2001) related with project managers’
learning. Thus, network dynamics; culture, values and knowledge transfer processes within
them can be modeled using different simulation tools like Agent-Based modeling.
Human Resources Projects
+ Novices 9 potaa mr ae >| © Projects
+ Intermediates diferent ievels + Networks
+ Experts of expertise to © dda
* Promotion work in projects generation
+ Training + Methods
+ Mentoring + Knowledge
+ Culture utilization
+ Values
A qeeeaemeemees: >
| Knowledge ! a 1
1 accessibility ' 1 1
| { be used in H Knowledge, 4 | insight generation
| training and enawieage! transfer within | 1 Insight g
1, mentoring transfer | networks. ' '
1 activities among H ¥v
H networks |
Hl Knowledge
Knowledge Management }--------- J
+ Learning
+ Identification + Forgetting
+ Elicitation ----- ~rechnear-7 7-7" + Knowledge
* Dissemination and applied accumulation
+ Repositories Knowledge + Knowledge
+ Technology to be stored obsolescence
and used
Figure 3. - Knowledge1 Sector Overview Diagram
The Project Sector of Knowledgel includes the processes of idea generation and group
collaboration to project development. People work together to develop projects, and the total
human resources available are the main constrain to project development. That is to say,
although people could develop a lot of ideas, they have a limited time to translate ideas into
actual projects.
4/4
The knowledge sector receives as a main input the quantity of projects successfully
finished. The main result of those projects is a set of insights conditioned by the proportion of
experienced people in the field. The model assumes that a fraction of this knowledge contributes
to the SD field, but other fraction contributes to other fields’ development. Another important
assumption is that knowledge becomes obsolete as a function of two different processes, the rate
of change of learning and human resources replacement. Learning rate of change is linked with
new knowledge generation, and it is assumed that the more learning, the more feasible this new
leaming replaces old one. On the other hand, the knowledge modeled in this sector refers to the
knowledge inside people’s heads. Thus, people who leave the field take with them their
cumulative knowledge.
The knowledge management sector represents the process of moving the people's
knowledge from their heads to publications. The basic assumption is that just a fraction of the
ideas is actually published. Finally, knowledge development perception is developed in terms of
the concrete publications in the field.
The Human Resources Sector
Population dynamics are modeled as an aging chain composed by three stocks of people
according to their level of expertise in system dynamics, junior, intermediate and advanced
(figure 4). The different stocks through the aging chain represent the level of understanding
about System Dynamics of the people in the field. Through this aging chain, people can move
into a deeper level of understanding or quit, moving to other activities. Our assumption is that
people at the junior and intermediate level quit because they get disappointed, and people at the
advanced level moves to other activities or fields because internal conflicts among them.
Knowledgel model assumption is that conflict is a function of the fraction of advanced
practitioners in the field, the bigger this fraction, the bigger the probability of conflict.
Time for juniors to ‘Time for intermediate
Siepactiant, to disenchant
Effect of conflict
Intermediate average over migration:
" Junior average maturation time Average
‘maturation time retirement time
Jutiod movie Intermediate movin Advanced
Geeta to other fields moving to other
omerne fields
—— Intermediat Advanced
New Advanced
Junior Intermediate sens
ractitioners
Growing fraction , momen, promotion retirement
‘Junior proportion Intermediate
proportion Advanced
Total people ed proportion
the field
e Probability of
conflicting approaches
Figure 4. The aging chain in Knowledge1
5/5
On the other hand, the average maturation times in the model are affected by the
adequacy of mentoring. That is to say, the proportion of juniors and intermediates joining
mentoring and training programs make these average maturation times shorter (the internal
dynamics of mentoring and training are described below). Moreover, maturation times are
related in opposite direction with the disenchanting times in the model, making the disappointing
times shorter for longer average maturation times (figure 5).
Effect of mentoring on
Junior maturation time
Disenchant time
nomal EIPT £ Intermediate Mat
Junior Maturation time normal
EJPT SE. time normal
Effect of mentoring on
Effect of jun Effect of int prom int maturation time
Time for juniors tog— ti
‘toatthenk prom time ‘ime for intermediate me Normal migration
to disenchant fraction
Effect of conflict
Intermediate average over migration:
i Junior average maturation time ‘Average
maturation time retirement time
jonni Intermediate moving Advanced
erase to other fields moving to other
omen fields
Be] lintermediat Advanced
New ‘Advanced
Junior Intermediate vane
ractitioners
Growing fraction , Bromotion, promotion retirement
Junior proportion Intermediate
proportion Advanced
Total people weet proportion
the field
ie Probability of
conflicting approaches
Figure 5. The aging chain with the average time to maturate and disenchant relationship.
The growing fraction of the field depends on field visibility. This field visibility is
defined as the ratio of the perceived change in the field and the perceived change in other fields,
and constitutes the connection point from the knowledge management sector to the human
resources sector (figure 6).
As it was claimed before, a very important activity inside the human resources sector is
mentoring and training (figure 7). In Knowledgel model, mentoring and training are modeled as
a set of programs with a limited capacity to people incorporation. The capacity of the programs
is determined by a fraction of advanced and intermediate practitioners willing to participate in
the training effort and a maximum capacity of people that they can help during a period of time.
On the program demand side, the assumption is that just a fraction of the people in the junior and
intermediate stocks is interested on joining a training program. All the intermediate available
effort on training is focused on the junior population, but the advanced mentoring capacity is
distributed between juniors and intermediate that want to be mentored. The allocation of
advanced resources is a policy decision represented in the model as a weight on junior, that is to
say, the fraction of advanced effort devoted to junior education.
6/6
Effect of mentoring on
Junior maturation time
Disenchant tinie
nomnal EPTE Intermediate Mat
Junior Maturation time normal
EJP E time nomal
Effect of mentoring on
ft of jun Effect of inf prom int maturation time
Time for juniors ta — prom time {me for intermediate in Nommal migration
waaint to disenchant frecion
Effect of conflict
Intemediate average over migration
‘maturation time
Average
Junior average
retirement time
‘maturation time
Intermediate movit Advanced
eter to ober fields roving to ber
fields
Junior
New ‘Advanced
Y Junior
Nor gm mos ooh — ae
an inal (Junior proportion Intermediate
i ‘Fropertion Advanced
robtpemeinl
_Bfatoffieaisitny TO DEORE Ne proportion
EFVGF£ on growing fraction
Probability of
conflicting approaches
Field vishility Perceived change in
~~ __ other fields
Increases in
Pesvetyed change i perceived knowledge
‘Time to avg actual perceived
knowledge
Figure 6. The aging chain with the effect of knowledge on the growing fraction
Thinking in an aggregate system that includes practitioners from several places around
the world, it takes some time to people for joining a program. Besides, partial saturation of the
training and mentoring programs increases the difficulty to join one, increasing the needed time
to find a program to join to. When the programs are fully saturated, it takes a very long time to
people for joining. Figure 8 shows the basic model structure of the junior practitioners joining to
mentoring programs that implements these assumptions (next page). People in the programs are
represented as two stocks; one for juniors joining to intermediate programs and a second one for
juniors joining advanced programs. The inflows to both stocks are closed when supply or
demand reaches the saturation point. Actually, figure 8 is a detailed view of the small feedback
loops inside the rectangle of figure 7. The loop inside the rounded corner rectangle on figure 7,
involves a similar process for the intermediate case.
7/7
EMJMTE
Beet of mening on
Junior maturation ume [uniormoving
‘ofbe elds
EMIMTE
fect of mentoring of
Junior Maturation Intermediate Mat
‘ime nomial time no
Junior average
‘maturation time
Intermediate movi
to other fields
Intermediate average
‘maturation time
int maturation time
Intermediatin
‘mentoring to
intermediate ratio
Intermediate in
mentoring
Advanced to
New
Tutor Intermediate
precioners promotion ‘womoton
Growing fraction
Frocion of inenediate
Frtion of advan
fede willing to mentor ling to mea
wae mo \
Mentoring Capacty
tenet ang vermenor Advared willing
Advanced fo
afm Advanced intermediate mentor
“Total Juniors‘ Thtermediate” Junior joining 4g" ete ee ec
‘mentoring capacity used mpeg,
programs
Advanced to junior
}— capacity used
"Junior joining
‘ment programs
Advanced to ultor
mentoring capacity
Weighton
junior
Figure 7. Adequacy of mentoring and its effects over average maturation times
Intermediate
_ mentoring capacity —~
Saturation of |—*—~
intermediate capacity
i Intermediate
{ ool capacity used
mentoring
\ ee 2
\,_ Average = nga
eo time leaving
Effect of intermediate
fF pnt saturation om —
EICST
Junior wing to be
mento .
_
Satiation of junior
re inmentoring
sith %
aS Advanced tn junior
st —~ rentorng capacity
Effetof Junior
ee > saturton of advanced
\ ee to junior capciy
ed, i
; pp hui ape
anor ining 5 i
nor joing me TJunleaving a’
7 oat q le
a ae pati |
i
\ [eerie mentoring Um
miawnaled
Time for jun to aoquire ‘adv mentoring
_lemediate mestar ofr uno om st
a Pa
Efecto avaced tg
t Tineto gia Junior saturation on ine
Tins Raa mentor nomad
Figure 8. Junior population joining to mentoring and training programs.
8/8
Project Sector
Projects are a result of ideas generated by people. Thus, the model considers that the
people in the System Dynamics field have a certain productivity in generating ideas that
constitutes the potential future projects (figure 9). The network effect is a fraction that reflects
people’s limitation to see the complete pull of ideas. That is to say, people idea generation and
people effort to transform simple ideas into feasible projects is bounded by the participation of
each person in a network of practitioners. A practitioner in a specific network produces ideas
related with previous work developed inside his/her network and devotes effort to maturate that
limited set of ideas. On this way, the mature project ideas represent all the ideas ready to
become actual projects.
Advanced
Intermediat Total people in
the field
Junior Project ideas Mature
Idea generation cooking Idea maturation broject ideas Ideas becoming
People idea
productivity
Network effect Cooking time
ee .
Average cooking
time
Figure 9. Project ideas generation.
Unfortunately, the outflow of ideas ready to go is constrained by a set of factors. The
main constrain modeled in Knowledge1 is people capability to work in projects (figure 10).
People in the field are assumed to be organized in work groups with a fixed project capacity, a
quantity of projects that each group can focus on per year. The ratio of projects in process and
this capacity constitutes the average time for mature project ideas to become actual projects. As
it is shown in the structure of figure 10, fraction of these projects will end with a success, but
another fraction will never be finished. This fraction is assumed to be a constant in the model.
9/9
Project success
fraction
Average project
duration
Work group Project ;
failure Total project
productivity Time to start new.
nies outflow
Average people Work group total a
per group capacity Projects in
Projects stating P>S* J Projects ending
Work groups
Advanced
Intermediates ae beanie in Project generation
ea per idea
Junior AS go Project ideas Mature
>
Tea generation | °°%!S | tdea maturation [ect
Ideas becoming
real
People idea
productivity
Network effect ge hing time”
Average cooking
time
Figure 10. Work groups’ capacity and projects in process.
Knowledge sector
An important model presumption is that leaming takes place mainly by project
development. Although it is possible to argue that people leam from failure, it is supposed that
new insights to be added to the existing System Dynamics knowledge are owed to the successful
projects (figure 11). On this way, finished projects are translated into new insights. Because of
the interdisciplinary nature of the SD field, only a fraction of these insights enriches the SD field,
and another fraction promotes a deeper understanding in other fields. As it is shown in the
figure, knowledge has a lifetime reflected in the time to become obsolete.
On the other hand, insight productivity per project finished and the fraction of knowledge
transferred to System Dynamics are considered to be a function of the experienced fraction of
people in the field. In the case of insight productivity, experienced fraction is considered to be
equal to the sum of both, intermediate and advanced fraction. In the case of the fraction of
insights transferred to the SD field, experienced are considered to be only the advanced
population.
10/10
Fraction of Forgetting time
transferred insights
Project Applied
failure Insight tranfer knowledge] Tnsight
obsolescence
Projects in
P| progress
Projects starting! Projects ending
EEPIf
Insights per.
Effect of experienced project
proportion on insights Time to Become
Total insights pes
Insight per project a
nomeal & | Knowledge
Leaming Obsolescence
Experienced
fraction Nonmal SD ,
Imowledge fraction Fraction SD basic
eee \_Bepnteage
Reporsan Navanced
pnparnel Effect of advanced
proportion on basic
knowledge generation
EABKG f
Figure 11. Projects and knowledge generation in Knowledge1 model.
Although at the beginning of the modeling process the time for knowledge to become
obsolete was considered as a constant value. The actual version of the model considers this time
to depend on two different structures. First, it is assumed that the rate of change of knowledge
generation (learning) has an effect in the same direction that the obsolescence rate of change. In
different words, it is more likely for knowledge to be replaced by new knowledge when leaming
of new knowledge is faster. As a result, the faster the learning, the faster the knowledge
obsolescence. Second, knowledge modeled in this sector resides inside people’s heads. Then,
practitioners’ turnover has an effect on knowledge obsolescence in the same direction, the higher
the tumover, the higher the loss of knowledge or obsolescence (figure 12).
11/11
Fraction of
transferred insights
Forgetting time
Project gS
failure Insight tranfer
Insight
obsolescence
Projects startin Projects ending Time to perceive
Perceived EOIOf
enange of hg
field
EEPIE
Insights pe
Effect of experienced project
proportion on insights
Effect over insight
obsolescence Time to become
Total insights ahsolebs
Insight per project = ‘Avg time to become
normal : obsolete
Leaming | Knowledge] psofescence
Experienced
fraction Normal SD Effect of people living
knowledge fraction tion SD basic ea over obsolescence time
Intermediate Knowledge obsolescence time
Proportion = avantad EPLOKL Human resource, tmnt moving
proportion presen ava roiation index“
“Sie proportion on basic
knowledge generation
ae Junior moving to Advanced moving to
other fields other fields
EABKG f Total people in
the field
Figure 12. Knowledge sector, including knowledge generation and knowledge loses.
Knowledge Management Sector
Almost all the modern fields, professions and sciences assign an important amount of
effort to record the knowledge generated by all practitioners. In general, the process starts with a
good idea to be published and ends with a publication (ie. a paper or a book). Leaming
produces new ideas. Some of them are published locally (dissertations, working papers, reports),
and some of them are published widely (in books and joumals). The knowledgel model
considers in this version only the widely published ideas. This process is modeled with the
structure shown in figure 13 as a 4-stage process. Once an idea is considered good to be
published, a person or a group of people decides to write something about it, and it takes some
time to be done. After that, written works pass a revision process that takes time to be done too.
During these first stages it is possible that some publications fail. Publications that survive the
two initial stages wait a little time to become actual paper and ink. This finished works,
unfortunately, are thought to have a lifetime similar to the normal knowledge lifetime.
Perceived knowledge in the field is considered to be a smooth of the finished
publications. This perceived knowledge contributes to field visibility and field attractiveness
explained in the human resources sector.
12/12
sD
Knowledge | Actual Obsolescence
Fraction of new Knowledge obsolescence time
perceived as valuable to be Leaming
published
Time to integrate
ideas Perceived SD
Increases in | Knowledge
Insighisper_ perceived knowledg
pum Obsolescing
2 Publications i Publications i pp Publshable Piblbed
process revision "Aecépted materials materials
Deciding to Sending to publicatio Publshing
publish revision *<
“on t
Failing Total pub; "| Rejected Total
publications Pepers paper rate Publishing time
Writing time Receptance
time
Fraction finished Fraction accepted
Figure 13. Knowledge management sector.
The Behavior of the Model (Knowledge1)
Base Run
The basic behavior of the model is shown in figure 14. The knowledge related variables are (1)
SD Knowledge, (2) Perceived SD Knowledge and, (3) Published materials. SD Knowledge
represents the knowledge generated in the System Dynamics field related to System Dynamics
(core of the field); Perceived SD Knowledge represents the level of awareness of the people in
the field with respect to the level of knowledge available to be used. Published materials
represent the number of publications in the field.
KNOWLEDGE PROJECTS
400 publication 800
2,000 Insights
600
300 publication
1,000 aan i
200
200 publication | em
0 Insights 0
a a eT
‘Time (year) ‘Time (year)
Published materials : Base publication Project ideas cooking : Base Idea
SD Knowledge : Base Insights ‘Mature project ideas : Base Idea
Perceived SD Knowledge : Base Insights Projects in progress : Base Tdea
(a) (b)
Figure 14. Knowledge and projects in Knowledge
13/13
MENTOR CAPACITY POPULATION
c m0 2 3 40 50 6 70 8 9 100
Time (year) o mw 2 30 40
50
Time (year)
(a) (b)
PUBLICATIONS PEOPLE IN MENTORING
co 0 2 3 40 50 60 70 80 90 100
0 10 20 30 «440 «650 «G0 70 «8080 ~S«0 Time (year)
Time (year)
Publishable materials : B
Publications in process : B
Publications in revision : B
publication
publication
publication
(c) (d)
Figure 14 (cont). Population dynamics, mentoring and publications in Knowledge1.
Though the base run was intended to show equilibrium, the variations present are related to the
fact that the initial equilibrium values were assigned by graphical methods by replacing the
equilibrium state achieved by the model at the end of the run and not by the more adequate
analytical mathematical method that would include the complete set of relationships to achieve
equilibrium. On the other hand, for a relatively small and stable population, knowledge-related
variables will grow in a linearly like way.
Run 1 Changes in Time to Integrate Ideas
The first sensitivity analysis performed is related to changes in the time to integrate ideas. The
base time used is 50 years. This time represents the time that the average individual in the field
takes to integrate the ideas generated in the field that helps his knowledge perception of the field.
The change is introduced in time equals to 20 using a step function of size “time for sensitivity”
which varies from -49 to +49 using a random uniform distribution (all changes introduced in
sensitivity will use the same distribution). This means that the actual time to integrate ideas in
the model will vary from 1(50 + -49) year to 99 (50 + 49) years. This range was selected to
represent the fastest time possible to a doubling the normal time used.
We can see that the effect of changes of the time to integrate ideas in the population sector of the
model is almost entirely in the junior population (Figure 15). We observe an increment in juniors
14/14
followed by a decrement that lead to the original number of them approximately. The behavior of
the total people in the field follows the behavior of the juniors almost identically. The effect on
the intermediates and advanced is less important.
Sensi Time to perce Sensi Time to perceive
50% 75% 95% 00% 50% 75% 95% 00%
‘Total people in the field Advanced
204 40
170 =~ 35
40 30
no 25
80 L a" 20 - :
0 25 30 5 i009 25 30 5 700
Time (year) Time (year)
(a) (b)
Sens Time to percei Sensi Time to perceiv
50% 75% 5% 00% 50% 75% 95% 00%
Intermediate Junior
40 200
35 150
x0 100 a.
* n> °° | ed
20 0 25 50 5 100 ° 0 25 50 75 100
Time (yeas) Time (year)
(c) (d)
Figure 15. Population sensitivity runs for changes in time to perceive knowledge.
The high exposure of the field produced by the fast perceived change in the field attracts many
people, but in the long run these would leave due to a lack of mentoring capacity and
disenchantment of the field. The effect of doubling the time to integrate is not as nearly as
relevant then making it little. The doubling can deteriorate the total number of people of the field
for a while and then the population would become almost stable again.
We see a relatively low impact on the total knowledge generated (SD Knowledge) and the
perception of it (Perceived SD Knowledge). The changes in the time to integrate ideas do not
change significantly the behavior of these variables but change the final output of them (figure
16).
15/15
Sensi Time to perceiv Sensi Time to perceiv
50% 75% 95% 1.007 50% 75%|M5° 0095
SD Knowledge Perceived SD Knowledge
2,000 400
1,500
1,000
500
° 0 25 50 6 100 0 25 50 75 100
Time (year) Time (year)
(a) (b)
Sensi Time to perceiv Sensi Time to perceiv
50% 75% | 95% 1007: 50% 75% |S 0095
Published materials: Publications in process
600 8
450 6 = —
0 Fy 50 75 100 0 25 50 5 100
Time (year) Time (year)
(c) (d)
Figure 16. Publications and knowledge sensitivity runs for changes in the time to perceive
knowledge
Lastly, we see that the impact on published materials is low and that on publications in process is
relatively high. The impact is smoothened though the aging chain of publications.
Run 2 Changes in Normal Growth Fraction
The second sensitivity analysis performed is related to changes in the normal growth fraction.
The base fraction used is 10%. This fraction represents the normal growth, on the average, that
the field experiments per year. The change is introduced in time=20 using a step function of size
“normal growth for sensitivity’ which varies from -0.09 to +0.09 using a random uniform
distribution. This means that the actual growth in the model will vary from 0.01 (0.10 + -0.09) to
0.19 (0.10 + 0.09). This range was selected to represent the lowest positive growth feasible to a
doubling the normal growth used.
16/16
Sensi Growth ‘Sensi Growth:
50% 75% (NN 95% 1007 50% 75%|M5° 0095
SD Knowledge Perceived SD Knowledge
2,000 400
1,500 300
1,000 200
0 6 50 5 100 0 25 30 5 100
Time (year) Time (year)
(a) (b)
Figure 17. Knowledge sensitivity runs for changes in the growth fraction.
Changes in growth clearly influence SD Knowledge (figure 17) in a significant way this is
derived of the activities performed by a larger number of individuals in the field. The growth
generates large variability in the population variables as shown in figure 18.
Sensi Growth. ‘Sensi Growth:
50% 759% (N95% 100% 50% 75% (NNN959% 005
Junior Intermediate
2,000
400
1,500 300
41,000 200
eis A ii _A
0 0 -
0 50 0 rr 30 8 Too
Time (year Time (year)
(a) (b)
Sensi Growth: ‘Sensi Growth:
50% 7506 NF 95% 1.00% 50% 75% [lllo59% 00%
‘Advanced Total people in the field
400 000
300 3,000
200 2,000
0 [ 0 =
0 ro 0 75 io 0 2 30 ra Too
Time (year Time (yea)
(c) (d)
Figure 18. Population dynamics for changes in the growth fraction.
As we could expect the most affected population variable is the junior variable, it is interesting to
see the way the growth affects similarly the advanced and the intermediate. As a consequence the
publications in process and the published materials are affected (figure 19).
17/17
Sensi Growth ‘Sensi Growth:
50% 75% (NN 95% 1.007 50% 75%lM5° 0095
Publications in process Published materials
20 600
15 im 450
10 300
5 150
° 0 25 50 6 100 ° 0 25 50 75 100
Time (year) Time (year)
(a) (b)
Figure 19. Publication sensitivity runs for changes in the population growth fraction.
Run 3 Changes in Mentoring Capacity per Mentor
The third sensitivity analysis performed is related to changes in the mentoring capacity per
mentor. The base capacity used is 3 people in mentoring for each mentor. This capacity
represents the maximum number of people that the average advanced individual in the field can
mentor at the same time. The change is introduced in time=20 using a step function of size
“mentoring capacity for sensitivity” which varies from -2 to +18 using a random uniform
distribution (all changes introduced in sensitivity will use the same distribution). This means that
the actual mentoring capacity per mentor in the model will vary from 1(3 + -2) person to 21 (3 +
18) persons. Increases in mentoring capacity could be owed to improvements in mentoring
techniques or to the use of distant learning or other information technologies.
Sensi CapMent: ‘Sensi CapMent.
50% 75% NS 95% 100% 50% 75% (N05 % 00%
SD Knowledge Perceived SD Knowledge
2,000 400
1,500 300
1,000 200
so eae “
© 0 25 50 6 100 e 0 25 50 75 100
Time (year) Time (year)
(a) (b)
Figure 20. Knowledge sensitivity runs for changes in the mentor capacity.
According to this model, the changes in mentoring capacity do not have a significant impact on
SD Knowledge nor the Perceived SD Knowledge (figure 20). The effect of the changes in
capacity is not sufficient by itself to influence the behavior of the knowledge variables, and it
could be expected for a relatively stable population in the field. Changes in this variable could
be more critical in growing periods.
18/18
Sensi CapMent- ‘Sensi CapMent.
50% 759495 % 1007 50% 75%|§5° 00%
Publications in process Published materials
6 400
55 350
5 — 300 a
45 250
4 0 25 50 6 100 200 0 25 50 75 100
Time (year) Time (year)
(a) (b)
Figure 21. Publications sensitivity runs for changes in the mentor capacity.
Publications in process is influenced in a very
slight way and by the time they get to be Grapl forintermedis promotion
published materials the influence has vanished
through the smoothing effect of the aging chain °*
(figure 21). is NORD) 6Zccnnnnsnnnnnn
The population variables of the model are
influenced by these changes having a similar 04020 304050807080 80
impact across stages of the aging chain. The most Tae ven,
influenced can be considered the advanced _ temsiicprm Capea) fensr
: . ids omsn: Capel 18 Penta
variable where we can see that according to the — '™si#»mnwies: ti Pe
model, the more people the advanced mentor the . . .
more advance there are in the field through time. Figure 22. Intermediate promotion
This effect of growing through mentoring is due
to the number of intermediate moving to advanced through mentoring (figure 22).
Sensi CapMent Sensi CapMent
50% 75% NF 95% 100% 50% 75% NN05% 00%
Total people in the field Advanced
120 40
115, ——. 315
105 325
100 0 25 50 6 100 30 0 25 50 5 100
Time (year) Time (year)
(a) (b)
Figure 23. Population sensitivity runs for changes in the mentor capacity.
19/19
Graph forTotal people in the field Graph for Advanced
o mw 2 30 40
50 70 80 90 100 o i 2 30 40 30 60 70 a0 90 100
Time (year) Time (year)
Sensi CapMent Sensi CapMent
50% 75% (95% 1.0075 50% 75%|N5° 0096
Intermediate Junior
40 60
35 55
30 50
5 a asf a
205 40
Fo} 50 75 100 0 25 50 5 100
Time (year) Time (year)
(e) (f)
Figure 23 (cont). Population sensitivity runs for changes in the mentor capacity.
We present the sensitivity graphs and the normal graphs using four runs to identify specifically
where, in the continuum, are the limits of the behavior. We can see that the run with Capacity in
(+18) described by the top line tells us the story of growing through mentoring and the run with
Capacity (-2) described by the bottom line tells us the case of low mentoring in the field (figure
23).
Run 4 Changes in Weight on Junior
The fourth sensitivity analysis performed is related to changes in the weight on junior. The base
weight used is 0.2 or 20% of the advanced capacity to mentor goes to junior individuals, the rest
goes to intermediate. This parameter represents the willingness of advanced people to interact
directly with juniors and mentor them. The change is introduced in time 20 using a step function
of size “weight for sensitivity” which varies from -0.19 to +0.79 using a random uniform
distribution. This means that the actual weight on junior in the model will vary from 0.01 (0.2 + -
0,19) to 0.99 (0.2 + 0.79). This range was selected to represent the lowest weight possible to the
highest weight being a virtual 100%.
20/20
‘sensi on weight sensi on weight
50% 75% 95% 100% 50% 75% (iNN05% 00%
Total people in the field ‘Advanced
120 40
115, 315
110 35
105 325
100 25 50 6 100 “0 25 50 75 100
Time (year) Time (year)
(a) (b)
‘sensi on weight sensi on weight
50% 75% 95% 100% 50% 75% NN05% 00%
Intermediate Junior
40
60
55
50
45
6 50
Time (year)
(c)
5 100
40
0 25 30
Time (year)
(d)
100
(
Figure 24. Population sensitivity runs for changes in weight on junior.
We can see that the changes on the weight on junior affect principally the advanced people and
the juniors (figure 24). The increments achieved are not dramatic because the actual capacity per
advanced remains constant in this run.
sensi on weigh sensi on weight
30% 75% (iN}95% 100° 50% — 75%%(NN959% 005
SD Knowledge Perceived SD Knowledge
2,000 400
1,500 300
41,000 200
500 100
100 0 25 100
50
Time (year)
(b)
30
Time (year)
(a)
Figure 25. Population sensitivity runs for changes in weight on junior.
The effect on SD Knowledge and perceived knowledge is very small (figure 25). The higher
disposition to work with juniors is not enough by itself to influence the development of SD
Knowledge in this model.
21/21
‘sensi on weight sensi on weight
50% 75% (95% 1007 50% 75465 0095
Publications in process Published materials
6 400
55 350
5 300
45 250
4 0 25 50 6 100 200 0 25 50 75 100
Time (year) Time (year)
(a) (b)
Figure 26. Publication sensitivity runs for changes in weight on junior.
Publications in process and published materials are not very much influenced by the changes in
the relative attention of advanced to juniors or intermediate.
Run 5 Changes in Network Effect
The fifth sensitivity analysis performed is related to changes in the network effect. The base
effect used is 0.50. This effect has two effects, (1) is a multiplier for idea generation and (2)
creates the adequate environment to reduce the maturation time of ideas due to interchange and
crossbreeding of ideas amongst individuals and networks. We thinks that this variable is one of
the relevant ones to clarify further and explore new ways to properly model it. The change is
introduced in time=20 using a step function of size “effect for sensitivity” which varies from -
0.49 to +15.5 using a random uniform distribution. This means that the actual network effect in
the model will vary from 0.01(0.50 + -0.49) to 16 (0.50 + 15.5). This range was selected to
represent the most difficulty to interact in networks represented by a virtual cero (0) to a high
interact ability represented by a 16. This parameter, up to now, does not have a real counterpart
to identify. This is a conceptual tool to reflect the notion of the importance of the networking
activities.
sensi for net effect ‘sensi for net effec
50% 75% (95% 1007: 50% 75% (NOS 0095
SD Knowledge Perceived SD Knowledge
4,000
600
3,000 450
2,000
1,000
0
0 Fy 50 75 0 25 30 75 100
Time (year) Time (year)
(a) (b)
Figure 27. Knowledge sensitivity runs for changes in Network effect.
22/22
The model resulted to be extremely sensitive to this parameter generating very high changes in
the variables of knowledge, population and publications. This is evidence of the importance of
pursuing further investigation in this matter for the future.
sensi for net effect sensi for net effect
50% 75% (N95 % 1.007: 50% 7595 0095
Total people in the field Advanced
400 60
300 50
200 40 ZA
100 — 30
0 - 20 ° =
0 5 30 5 T00 0 5 30 % Too
Time (year) Time (year)
(a) (b)
sensi for net effect ‘sensi for net effect
50% 75% (N95 % 1.007 50% 7595 0095
Intermediate Junior
60 400
50 300
40 200
30 ai 100
a 0 25 50 5 100 ¢ 0 25 50 oe} 100
Time (year) Time (year)
(c) (d)
Figure 28. Population sensitivity runs for changes in Network effect.
Run 6 New Parameters and Changes in Time to Integrate Ideas
This run will use a combination of changes derived of the previous runs. We came to know that
changes in time to integrate ideas attracted many juniors to the field, which will be
acknowledged by changing three parameters, and then perform a sensitivity analysis on time to
integrate ideas identical to that of run number one. The parameters to change are (1) mentoring
capacity per mentor from 3 to 6 persons, (2) weight on juniors from 20% to 50% and, (3)
network effect from 0.5 to 1 recognizing the added activity to the field through doubling the
mentoring. The base run was performed with a time to integrate ideas of 20 years.
23/23
KNOWLEDGE
POPULATION
800 publication
4,000 Insights
400. publication
2,000 Insights
© publication
0 Insights 0 c
o 1 2 30 40 50 60 70 80 90 100 o i 2 3 40 60 70 80 90 100
Time (year) Time (year)
Publish materials: Run6Base————— publication _ teint ni non
SD Knowledge : Run6Base———— Insights Junior: RunfiBase| ——— ss
Perceived SD Knowledge : Run6Base——— insights Tae people in hella: Rune rene
(a)
PUBLICATIONS
(b)
MENTOR CAPACITY
o wo 20 3 4 50 6 70 80 90 100
Time (year)
Publishable materials : Run6B. publication
Publications in process : Run6B publication
Publications in revision : Run6B: publication
(c)
o 1 2 30 40 50 7080 90 100
Time (year)
(d)
Figure 29. Model behavior with a combined set of variable changes (new base mun).
As we can see, the behavior of the model indicates a constant growth in the population sector and
the knowledge variables. The publications grow, and then get to a period of stagnation to then
continue growing. The field is enjoying a continuous growing capacity in mentorship with this
combination of levels (figure 29).
SensiRun6 SensiRun6
50% 75% (N)95% 100° 50% 75% (N95 00%
SD Knowledge Perceived SD Knowledge
4,000 600
3,000 450
2,000 300
1,000 150
0 L f) -
0 25 30 75 Too ) 25 30 75 100
Time (year) Time (year)
(a) (b)
Figure 30. Sensitivity runs for changes in time to perceive knowledge.
24/24
Graph for SD Knowledge Graph for Perceived SD Knowledge
4,000 600
3,000 450
2,000 300
1,000 150
0 0
o wo 2 30 40 50 6 70 0 90 100 o 0 2 30 40 50 60 70 80 9 100
Time (year) Time (year)
SD Knowledge : Run6Ba Insights Perceived SD Knowledge : Run6Base Insights
SD Knowledge : Bast Insights _Perveived SD Knowledge : Base Insights
(a) (b)
Figure 31. Comparison between the base run and run 6.
SD Knowledge is increased with respect to the base run and exhibits low sensitivity to the
changes in time to integrate ideas while in perceived change there is a larger influence (figure 30
and 31).
SensiRun6 SensiRun6
50% 759% (N95% 1005 50% 75% (NNI959%S 00°
Total people in the field ‘Advanced
1,000 200
750 150
500 100
250 30
0 L 0 - =
0 25 50 5 100 0 25 50 5 100
Time (year) Time (year)
(a) (b)
SensiRun6 SensiRun6
50% 75% (N)95% 1.005 50% 75% NNI959% 00°
Intermediate Junior
200 600
150 450
100 300
50 150 =
0 L 0 - =
0 25 50 5 100 0 25 50 5 100
Time (year) Time (year)
(c) (d)
Figure 32. Population sensitivity runs for changes in time to perceive knowledge.
Population variables are clearly affected by these changes generating large varying outcomes
depending on the time to integrate ideas (figure 32).
25 / 25
Run 7 Systematic Exploration using a full factorial va Design of Experiments in standardized
factors X;,
Lastly, we will use a systematic approach to explore the best combination of factors varying
three of them in two different predetermined levels. This approach seeks to identify further ways
to systematically approach the sensibility analysis of the model and a way to identify
combinations of factors that would generate the greater response possible (Kleijnen 1995). The
array that we will use allows us to design experiments to test the combinations. The total number
of experiments is eight and the variable used to measure efficiency of the experiment will be the
amount of SD Knowledge accumulated at year 100°. We could have selected any variable as
measure of efficiency or even a combination of them using a weighted average function or a
multiple attribute model (MAU) to measure the efficiency of the model in each experiment. The
latter represents the best way to measure the output. We chose the first for simplicity sake and
due to the nature and purpose of the model. The full factorial 2° is represented as the matrix x
shown below and the results of the experiments will be collected in the vector containing the
elements x, leading us to the final maximum ¥,,,..,. The elements in the matrix represent the
levels at which the variables will be tested in each experiment. We will use “high level” for 1
and “low level” for 0.
Oo 119 Oud
bo Beh
fl 0 a i
ARE
i it
0 0 g
oof BB
The three variables used for the experimental design are (1) mentoring capacity per mentor
{3,6}, (2) weight on juniors from {0.2,0.8} and, (3) time to integrate ideas from {10,40}. For this
experiments the network effect will remain constant and equal to 1 and the normal growing
fraction will remain constant and normal in 0.1. The next matrix contains the actual levels for the
variables and the results of the experiments measured in the units of the output reference variable
(SD Knowledge).
2 This simple measure of the efficiency of the model will generate simple solutions that will allow us to understand
how to proceed further.
26 / 26
(6 0.7 209 (p,7409
1 0.7 20H H.537
‘ 04 201 rea
0.4 20 321
= O- "4" =
X=05 0.7 380” (aoith> om Expl
iB 0.7 35H (5934
0.4 350 U925U
Boe od Band
To test the totally decision dependent variables we made four new experiments. We argue that
the number of people that mentor can see and the relative weight that he puts into juniors and
intermediate are decisions that are made by the mentor. This is that the levels of the variables
represent different preferences of the advanced. We replicated four experiments shown in the
matrix o three times to “test” how the “integrative” environment can affect the preferences of
the advanced related to which altemative generates a more robust outcome changing the “time to
integrate ideas” from 10 to 20 and finally to 30 years as a measure of possibilities.
We generated three sets of results of the four experiments using a full factorial designs of two
variables with two levels. The results are shown below.
0.70 (2.9900
0.74, HL545 H
040 [b,134[]
04H H.s17H
0.70 (e.7409
0.7h #537}
o.4fl [b,o920)
04H Hus21H
IL
H> Xicmax)
Time =40 years
Time=20 years
0.79 [p.402q
u u
o.7f_, Asm
oat) [1.9650]
oa H.402H
Il
Baths Goods ess
> Xi¢maxy
Time=80 years
27/27
The complete set of results is shown including the mean and the standard deviation of the results
for each experiment using different levels of “time to integrate ideas” as general conditioners of
the results.
cucre.990 2,740 2.4029 (p,711 295p
HHLS45 1,537 15771 fa.553. 214
O= Q
(BITb134 2,092 19650) [p,064 880
HitH.317 1321 1402 H347 48}
We can see that experiments 1 and 3 generate the larger average output in SD Knowledge. This
two experiments use 6 as the number of people that each mentor can see at a time. Experiments 2
and 4 generate lower average outputs but with significantly lower standard deviations. This gives
us the idea that a preference related to lower people per mentor can be more robust across
different “integrative” environments. This is that independently of the environment in which the
mentorship is taking place, lower numbers of mentors generate more stable SD Knowledge in the
end. To be sure, about the added robustness, we included a metric of dispersion over the mean of
the results of the experiments to correct for the effect of the lower average results of experiments
2 and 4. The results are shown below.
oo ().1088])
P. 2 ppowst
O° pw [).04261)
With this new results we confirm that the “best” preference to maximize the average result (SD
Knowledge) across different levels of “ability to integrate ideas” would be the one related to
experiment number 1. This is to have 6 people per mentor and to devote 70% percent of the
mentorship to Juniors in the field. If the idea is to generate a preference that would be less
variable (more robust) across different integrative environments the alternative would be the one
related to experiment 2 having 3 people per mentor and also devoting 70% of the time with
juniors in the field. This tells us that the orientation to juniors as the targets of intensive
mentorship is highly recommended. The number of people per mentor should be analyzed with
respect to the risk taking strategy of the mentor or the field.
Conclusions
The use of a model approach for the research of knowledge dynamics makes a lot of
sense. The exploration of best practices and the creation and dynamics of knowledge threw light
at some important processes of the different fields of knowledge. Among others, (1) mentoring,
(2) networking, (3) knowledge management and, (4) mechanisms of perception and integration
in the fields are crucial. This can be of major impact on the cohesiveness and alignment of the
communities if studied and acknowledged as important. The way people perceive knowledge and
the time that passes between knowledge is generated and one has the opportunity and capacity to
28 / 28
integrate the ideas and “perceive” it can lead people to think that knowledge is not being
generated, that the field is static. What the model shows is that knowledge is growing very
rapidly but the perception of it cannot keep up with this growth, and most probably will not do it
in the future. We argue that the networking effect increases dramatically the capacity of the
individual to learn but also generates a giant blockade that prevents people to see beyond their
networks of influence generating operating blindness that prevents leaming. Many people think
that the knowledge that they know of is the only knowledge out there, and by doing so they rest
momentum to the whole field of knowledge that they belong.
Implications for Future Research
No model is right, because all of them constitute simplifications of reality. However, it is
possible to leam and get insights about reality through the analysis of these simplifications as it
was shown in the preceding sections. Thus, this model dualism leads System Dynamicists to get
conclusions about every simple model, but always produces a parallel set of dynamic behaviors
and structures to be added to the model. The main areas of interest that are not addressed in this
study are:
(1) The knowing - doing gap (Pfeffer and Sutton 2000) understood as the gap between
perceived knowledge in the field and the actual utilization of the core knowledge in
practice.
(2) The influence of power and politics in the generation of new knowledge and practice.
(3) Network dynamics and knowledge transfer inside them. This area could imply the
collaboration of a SD model with an Agent-Based model.
(4) Project initiation is constrained by financial resources too. Future versions of the
knowledge model can explore the effects of funding on knowledge generation.
(5) Psychologists think that previous knowledge can both, facilitate and limit the
learning capability. It could be interesting to explore the structure of the impact of
perceived and actual SD knowledge on leaming.
(6) The dynamics of System Dynamics in its contribution to other fields is only partially
addressed by knowledge1, and it constitutes another way to go.
(7) Improving knowledge management activities is related with the effectiveness of
mentoring and training programs.
(8) In the present model, core knowledge is the result of a continuous process of ideas
and insights selectivity through the knowledge management effort. Benchmarking
and best practices approaches consider a different approach not implemented in the
model.
29/29
References
Hines, J. and J. House (2001). “The Source of poor policy: controlling learning drift and
premature consensus in human organizations.” System Dynamics Review 17(1): 3-32.
Kleijnen, J. P. C. (1995). “Sensitivity analysis and optimisation of system dynamics models:
regression analysis and statistical design of experiments.” System Dynamics Review
11(4): 275-288.
Pfeffer, J. and R. I. Sutton (2000). The knowing-Doing Gap: How smart companies tum
knowledge into action. Boston, Massachusetts, Harvard Business School Press.
Richardson, G. P. (1991). Feedback Thought in Social Science and Systems Theory. Waltham,
MA, Pegasus Communications.
Saeed, K. (1992). “Slicing a complex problem for system dynamics modeling.” System
Dynamics Review 8(3): 251-261.
Scholl, G. J. (1995). “Benchmarking the system dynamics community: research results.” System
Dynamics Review 11(2): 139 to 155.
30/30
Appendix - Model equations.
JB DI RIDE ODI IE
-Human Resources - Aging chain
LERCH SSO DOERR SAH
(002) | Advanced = INTEG( Intermediate promotion -
Advanced moving to other fields
- Advanced retirement , Advanced initial )
Units: People
(003) Advanced initial = 35
Units: People
(004) Advanced moving to other fields = Advanced *
Normal migration fraction
* Effect of conflict over migration
Units: People/year
(005) Advanced proportion = Advanced / Total people
in the field
Units: Dmnl
(006) Advanced retirement = Advanced / Average
retirement time
Units: People/year
(007) | Average project duration = 1
Units: year
(008) Average retirement time = 20
Units: year
(009) — Disenchant time normal = 8
Units: year
(010) | ECOM £ ( [(0,0)-
(1,5)],(0,1),(0.1,1),(0.2,1),(0.3,1),(0.4,
(0.6,1.8),(0.7,2.5),(0.8,3.6),|
Units: Dmal
(011) Effect of conflict over migration = ECOM f (
Probability of conflicting approaches
)
Units: Dmnl
(012) Effect of field visibility on growing fraction =
EFVGF f ( Field visibility
)
Units: Dmal
(013) Effect of int prom time = EIPT f ( Intermediate
average maturation time
/ Intermediate Mat time normal )
Units: Dmal
(014) — Effect of jun prom time = EJPT f ( Junior average
maturation time / Junior Maturation time normal
)
Units: Dmal
(015) Effect of mentoring on int maturation time =
EMIMT f ( Intermediate in mentoring to intermediate ratio
Units: Dmnl
(016) Effect of mentoring on Junior maturation time =
EMJMT f ( Juniors in mentoring to junior ratio
Units: Dmal
(017) | EFVGF £ ( [(0,0)-
(2,2)],(0,1),(0.2,1.03),(0.4,1.1),(0.6,1.2),(0.8,1.33)
(1,1.5),(1.2,1.67),(1.4,1.8),(1.6,1.9),(1.8,1.97),(2,2))
Units: Dmnl
(018) — EIPT £( {(0,0)-
(2,6)],(0,5),(0.110092,3.97368),(0.299694,2.65789), (0.556
575,1.86842)
(0.788991,1.31579),(1,1),(1.21101,0.982456),(1.41284,0.9
21053)
(1.60245,0.877193),(1.81651,0.789474),(1.98777,0.68421
1),(5,0.25)
)
Units: Dmal
(019) EJPT £( [(0,0)-
(2,6)],(0,5),(0.110092,3.97368),(0.299694,2.65789),(0.556
75, 1.86842)
(0.788991,1.31579),(1,1),(1.21101,0.982456),(1.41284,0.9
21053)
,(1.60245,0.877193),(1.81651,0.789474),(1.98777,0.68421
1),(5,0.25)
)
Units: Dmal
(020) | EMIMT f( [(0,0)-
(1,2)],(0,2),(0.1,1.98),(0.2,1.95),(0.3,1.9),(0.4,1.6)
(0.5,1.2),(0.6,0.95),(0.7,0.75),(0.8,0.6),(0.9,0.55),(1,0.5) )
Units: Dmnl
(021) | EMJMT £ ([(0,0)-
(1,2)],(0,2),(0.1,1.98),(0.2,1.95),(0.3,1.9),(0.4,1.6)
(0.5,1.2),(0.6,0.95),(0.7,0.75),(0.8,0.6),(0.9,0.55),(1,0.5) )
Units: Dmnl
(022) Field visibility = Perceived change in actual
perceived knowledge / Perceived change in other fields
Units: Dmnl
(023) Growing fraction = Normal growing fraction *
Effect of field visibility on growing fraction
Units: I/year
31/31
(024) Increases in perceived knowledge = ( ( Published
materials / Insights per publication
) - Perceived SD Knowledge ) / Time to integrate
ideas
Units: Insights/year
(025) — Insights per publication = 1
Units: publication/insight
(026) Intermediate = INTEG( Junior promotion -
Intermediate moving to other fields
- Intermediate promotion , Intermediate initial )
Units: People
(027) Intermediate average maturation time =
Intermediate Mat time normal *
Effect of mentoring on int maturation time
Units: year
(028) Intermediate in mentoring to intermediate ratio =
Intermediate in mentoring
/ Intermediate
Units: Dmnl
(029) Intermediate initial = 27
Units: People
(030) Intermediate Mat time normal = 4
Units: year
(031) Intermediate moving to other fields =
Intermediate / Time for intermediate to disenchant
Units: People/year
(032) Intermediate promotion = Intermediate /
Intermediate average maturation time
Units: People/year
(033) Intermediate proportion = Intermediate / Total
people in the field
Units: Dmnl
(034) | Junior = INTEG( New practitioners - Junior
moving to other fields - Junior promotion
Junior initial )
Units: People
(035) | Junior average maturation time = Junior
Maturation time normal * Effect of mentoring on Junior
maturation time
Units: year
(036) Junior initial = 44
Units: People
(037) Junior Maturation time normal =8
Units: year
(038) Junior moving to other fields = Junior / Time for
juniors to disenchant
Units: People/year
(039) Junior promotion = Junior / Junior average
maturation time
Units: People/year
(040) Junior proportion = Junior / Total people in the
field
Units: Dmal
(041) Juniors in mentoring to junior ratio = Total
Juniors in mentoring / Junior
Units: Dmal
(042) New practitioners =Total people in the field *
Growing fraction
Units: People/year
(043) Normal growing fraction = 0.1
Units: I/year
(044) Normal migration fraction = 0.05
Units: I/year
(045) Perceived change in actual perceived knowledge
= SMOOTH ( Increases in perceived knowledge
, Time to avg )
Units: Insights/year
(046) Perceived change in other fields = 5
Units: Insights/year
(047) Perceived SD Knowledge = INTEG( Increases in
perceived knowledge , 80
)
Units: Insights
(048) Probability of conflicting approaches = Advanced
proportion
Units: Dmal
(049) Published materials = INTEG( Publishing -
Obsolescing , 210)
Units: publication
(050) — Time forintermediate to disenchant = Disenchant
time normal * Effect of int prom time
Units: year
(051) Time forjuniors to disenchant = Disenchant time
normal * Effect of jun prom time
Units: year
(052) Time to avg =1
Units: year
32/32
(053) Time to integrate ideas = 50
Units: year
(054) Total people in the field = Advanced +
Intermediate + Junior
Units: People
Zp bo pS pS Sa a obo REESE IR
Control
bobo bbbopokiibiiiobbbbo sitet
Simulation Control Parameters
(056) FINAL TIME =100
Units: year
(057) INITIAL TIME =0
Units: year
(058) SAVEPER =1
Units: year
(059) TIME STEP =0.125
Units: year
poco obbbopoiiibiicbbiono ciate ik
Knowledge
Jp rooibbbopoiiobiobiebbioociak ice
(061) Applied knowledge = INTEG( Insight tranfer -
Insight obsolescence , 21.28
Units: Insights
(062) Avg time to become obsolete = 20
Units: year
(063) EABKG f([(0,0)-
(1,1)],(0,0),(0.1,0.02),(0.2,0.1),(0.3,0.2),(0.4,0.33)
,(0.5,0.5),(0.6,0.67),(0.7,0.8),(0.8,0.9),(0.9,0.98),(1,1) )
Units: Dmnl
(064) EEPI£ ( [(0,0)-
(1,1)],(0,0.1),(0.1,0.12),(0.2,0.2),(0.302752,0.280702)
,(0.4,0.38),(0.5,0.5),(0.6,0.67),(0.7,0.8),(0.8,0.9),(0.9,0.98)
L,1) )
Units: Dmnl
(065) Effect of advanced proportion on basic
knowledge generation = EABKG f
( Advanced proportion )
Units: Dmal
(066) Effect of experienced proportion on insights =
EEPI f ( Experienced fraction
Units: Dmnl
(067) Effect of people living over obsolescence time =
EPLOKL f ( Human resource rotation index
)
Units: Dmal
(068) — EOIO £( [(0,0)-
(2000,2)],(0,0),(244.648,0.0350877), (495.413,0.0789474)
(770.642,0.140351),(984.709,0.192982),(1168.2,0.350877)
(1333.33,0.789474)
(1486.24,1.31579),(1645.26,1.7193),(1798.17,1.91228),(2
000,2) )
Units: Dmnl
(069) | EPLOKL f( [(0,0)-
(40,2)],(0,0.1),(5,0.25),(10,0.5),(15,0.75),(20,1),
(25,1.25),(30,1.5),(35,1.75),(40,2) )
Units: Dmnl
(070) Experienced fraction = Advanced proportion +
Intermediate proportion
Units: Dmnl
(071) Forgetting time = Time to become obsolete
Units: year
(072) Fraction of transferred insights = 1 - Fraction SD
basic Knowledge
Units: Dmal
(073) | Human resource rotation index =Total people in
the field / ( Junior moving to other fields
+ Advanced moving to other fields + Intermediate
moving to other fields)
Units: year
(074) Ideas becoming real = Mature project ideas /
Time to start new projects
Units: Idea/year
(075) — Insight obsolescence = Applied knowledge /
Forgetting time
Units: Insights/year
(076) — Insight per project normal = 4
Units: Insights/Project
(077) Insight tranfer = Fraction of transferred insights *
Total insights
Units: Insights/year
(078) Insights per project = Insight per project normal *
Effect of experienced proportion on insights
Units: Insights/Project
(079) Normal SD knowledge fraction = 0.2
Units: Dmal
33 / 33
(080) Perceived change of the field = SMOOTH (
Leaming , Time to perceive
Units: Insights/year
(081) Project failure =( 1 - Project success fraction ) *
Total project outflow
Units: Project/year
(082) Project generation per idea = 1
Units: Project/Idea
(083) Project success fraction = 0.5
Units: Dmal
(084) Projects ending = Total project outflow * Project
success fraction
Units: Project/year
(085) Projects in progress = INTEG( Projects starting -
Project failure - Projects ending
, 160)
Units: Project
(086) Projects starting = Ideas becoming real * Project
generation per idea
Units: Project/year
(087) Time to perceive =5
Units: year
(088) Total project outflow = Projects in progress /
Average project duration
Units: Project/year
(089) Work group total capacity = Work group
productivity * Work groups
Units: Project/year
JH bbb bb bb Socal babar
-Knowledge Management
JERE C OR EOD EEA SOD HII
(091) Acceptance time = 1
Units: year
(092) Accepted publications = Fraction accepted *
Total paper rate
Units: publication/year
(093) Actual obsolescence time = SD Knowledge /
Obsolescence
Units: year
(094) Deciding to publish = Learning * Fraction of new
Knowledge perceived as valuable to be published
* Insights per publication
Units: publication/year
(095) Effect over insight obsolescence = EOIO f (
Perceived change of the field
)
Units: Dmal
(096) Failing publications =( 1 - Fraction finished ) *
Total pub rate
Units: publication/year
(097) Fraction accepted = 0.5
Units: Dmnl
(098) Fraction finished = 0.8
Units: Dmal
(099) Fraction of new Knowledge perceived as
valuable to be published = 0.5
Units: Dmal
(100) Fraction SD basic Knowledge = Effect of
advanced proportion on basic knowledge generation
* Normal SD knowledge fraction
Units: Dmnl
(101) | Leaming = Fraction SD basic Knowledge * Total
insights
Units: Insights/year
(102) Obsolescence =( SD Knowledge / Time to
become obsolete ) * Effect over insight obsolescence
Units: Insights/year
(103) Obsolescing = Published materials / Actual
obsolescence time
Units: publication/year
(104) Publications in process = INTEG( Deciding to
publish - Failing publications
- Sending to revision , 5)
Units: publication
(105) Publications in revision = INTEG( Sending to
revision - Accepted publications
- Rejected papers , 4)
Units: publication
(106) Publishable materials = INTEG( Accepted
publications - Publishing , 2)
Units: publication
(107) Publishing = Publishable materials / Publishing
time
Units: publication/year
(108) Publishing time =1
Units: year
(109) Rejected papers =( 1 - Fraction accepted ) *
Total paper rate
Units: publication/year
34/34
(110) | SD Knowledge = INTEG( Leaming -
Obsolescence , 100)
Units: Insights
(111) — Sending to revision = Fraction finished * Total
pub rate
Units: publication/year
(112) Time to become obsolete = Avg time to become
obsolete * Effect of people living over obsolescence time
Units: year
(113) Total insights = Insights per project * Projects
ending
Units: Insights/year
(114) Total paper rate = Publications in revision /
Acceptance time
Units: publication/year
(115) Total pub rate = Publications in process / Writing
time
Units: publication/year
(116) Writing time =1
Units: year
Job Go ob bbbboaroiirebeb bonito
-Human Resources - Mentoring
JERI ASSIA AAA A AA
(119) Intermediate in mentoring = Advanced to
intermediate capacity used
Units: People
(120) Total Juniors in mentoring = Advanced to junior
capacity used + Intermediate capacity used
Units: People
(122) Advanced mentoring capacity = Advanced
willing to mentor * Mentoring Capacity per mentor
Units: People
(123) Advanced to intermediate capacity used =
INTEG( Int joining mentoring programs
- Int leaving mentoring , 8)
Units: People
(124) Advanced to intermediate mentor capacity =
Advanced mentoring capacity
* (1 - Weight on junior )
Units: People
(125) Advanced to junior capacity used = INTEG(
Junior joining ment programs
- Jun leaving adv mentoring , 19)
Units: People
(126) Advanced to junior mentoring capacity =
Advanced mentoring capacity *
Weight on junior
Units: People
(127) Advanced willing to mentor = A dvanced *
Fraction of advanced willing to mentor
Units: People
(128) Average mentoring time = 4
Units: year
(129) Avg int leaving mentoring = SMOOTH ( Int
leaving mentoring , Time to avg
)
Units: People/year
(130) | Avg jun leaving adv mentoring = SMOOTH (
Jun leaving adv mentoring ,
Time to avg )
Units: People/year
(131) Avg junior leaving mentoring = SMOOTH ( Jun
leaving mentoring , Time to avg
)
Units: People/year
(132) EACS £ ( [(0,0)-
(1,1e+010)],(0,1),(0.1,1.03),(0.2,1.11),(0.3,1.25),(0.4,1.53)
(0.5,2),(0.6,2.86),(0.7,5),(0.8,10),(0.9,34),(1,1e+010) )
Units: Dmal
(133) EAJS £ ( ((0,0)-
(1,1e+010)],(0,1),(0.1,1.03),(0.2,1.11),(0.3,1.25),(0.4,1.53)
(0.5,2),(0.6,2.86),(0.7,5),(0.8,10),(0.9,34),(1,1e+010) )
Units: Dmal
(134) Effect of advanced capacity saturation on time =
EACS £ ( Saturation of advanced capacity
)
Units: Dmal
(135) Effect of advanced to junior saturation on time =
EAJS f ( Saturation of advanced to junior capacity
)
Units: Dmal
(136) Effect of intermediate capacity saturation on time
=EICS f ( Saturation of intermediate capacity
)
Units: Dmal
(137) Effect of intermediate saturation on time = EIS f (
Saturation of intermediate on mentoring
)
Units: Dmnl
35/35
(138) Effect of Junior saturation on time = EJS f (
Saturation of junior in mentoring)
Units: Dmnl
(139) EICS £ ( {(0,0)-
(1,1e+010)],(0,1),(0.1,1.08),(0.2,1.11),(0.3,1.25),(0.4,1.53)
(0.5,2),(0.6,2.86),(0.7,5),(0.8,10),(0.9,34),(1,1e+010) )
Units: Dmal
(140) —_ EIS f ( [(0,0)-
(1,1e+010)],(0,1),(0.1,1.03),(0.2,1.11),(0.3,1.25),(0.4,1.53)
(0.5,2),(0.6,2.86),(0.7,5),(0.8,10),(0.9,34),(1,1e+010) )
Units: Dmnl
(141) EJS £( [(0,0)-
(1,1e+010)],(0,1),(0.1,1.03),(0.2,1.11),(0.3,1.25),(0.4,1.53)
(0.5,2),(0.6,2.86),(0.7,5),(0.8,10),(0.9,34),(1,1e+010) )
Units: Dmnl
(142) Fraction of advanced willing to mentor = 0.5
Units: Dmal
(143) Fraction of intermediate willing to be mentored =
Units: Dmnl
(144) Fraction of intermediate willing to mentor = 0.2
Units: Dmal
(145) Fraction of junior willing to be mentored = 0.8
Units: Dmnl
(146) Int joining mentoring programs = ( Advanced to
intermediate mentor capacity
- Advanced to intermediate capacity used ) / Time to
acquire an advanced mentor
+ Avg int leaving mentoring
Units: People/year
(147) Int leaving mentoring = Advanced to
intermediate capacity used / Average mentoring time
Units: People/year
(148) Intermediate capacity used = INTEG( Junior
joining mentoring programs
- Jun leaving mentoring , 14)
Units: People
(149) Intermediate mentoring capacity = Intermediate
willing to mentor * Mentoring Capacity per mentor
Units: People
(150) Intermediate willing to be mentored =
Intermediate * Fraction of intermediate willing to be
mentored
Units: People
(151) Intermediate willing to mentor = Fraction of
intermediate willing to mentor
* Intermediate
Units: People
(152) Jun leaving adv mentoring = Advanced to junior
capacity used / Average mentoring time
Units: People/year
(153) Jun leaving mentoring = Intermediate capacity
used / Average mentoring time
Units: People/year
(154) Junior joining ment programs =( Advanced to
junior mentoring capacity
- Advanced to junior capacity used ) / Time for jun to
acquire advanced mentor
+ Avg jun leaving adv mentoring
Units: People/year
(155) Junior joining mentoring programs = (
Intermediate mentoring capacity
- Intermediate capacity used ) / Time to acquire an
intermediate mentor
+ Avg junior leaving mentoring
Units: People/year
(156) Junior willing to be mentored = Junior * Fraction
of junior willing to be mentored
Units: People
(157) Mentoring Capacity per mentor = 3
Units: 1
(158) Saturation of advanced capacity = Advanced to
intermediate capacity used
/ Advanced to intermediate mentor capacity
Units: Dmal
(159) Saturation of advanced to junior capacity =
Advanced to junior capacity used
/ Advanced to junior mentoring capacity
Units: Dmal
(160) Saturation of intermediate capacity =
Intermediate capacity used / Intermediate mentoring
capacity
Units: Dmal
(161) Saturation of intermediate on mentoring =
Advanced to intermediate capacity used
/ Intermediate willing to be mentored
Units: Dmnl
(162) Saturation of junior in mentoring = ( Intermediate
capacity used + Advanced to junior capacity used
) / Junior willing to be mentored
Units: Dmal
36 / 36
(163) Time forjun to acquire advanced mentor = Time
to get a mentor normal
* Effect of advanced to junior saturation on time *
Effect of Junior saturation on time
Units: year
(164) Time to acquire an advanced mentor = Time to
get a mentor normal * Effect of advanced capacity
saturation on time
* Effect of intermediate saturation on time
Units: year
(165) Time to acquire an intermediate mentor = Time
to get a mentor normal
* Effect of intermediate capacity saturation on time
* Effect of Junior saturation on time
Units: year
(166) Time to get a mentor normal = 2
Units: year
(167) Weight on junior = 0.2
Units: Dmal
ZOE EERO DOES AI
Project Sector
ZED S EERE O DEER ODA IIE
(169) | Average cooking time = 1
Units: year
(170) Average people per group = 5
Units: People/group
(171) Cooking time = Average cooking time / Network
effect
Units: year
(172) Idea generation = People idea productivity *
Total people in the field
* Network effect
Units: Idea/year
(173) Idea maturation = Project ideas cooking /
Cooking time
Units: Idea/year
(174) | Mature project ideas = INTEG( Idea maturation -
Ideas becoming real ,
600)
Units: Idea
(175) Network effect = 0.5
Units: Dmnl
(176) People idea productivity =3
Units: Idea/(People*year)
(177) Project ideas cooking = INTEG( Idea generation
- Idea maturation , 320
)
Units: Idea
(178) Time to start new projects = Projects in progress /
Work group total capacity
Units: year
(179) | Work group productivity =2
Units: Project/(year* group)
(180) Work groups =Total people in the field /
Average people per group
Units: group
37/37
CC BY-NC-SA 4.0