Modeling the Dynamics of Scientific Revolutions
Jason Wittenberg John D. Sterman
Department of Political Science Sloan School of Management
Massachusetts Institute of Technology Massachusetts Institute of Technology
Cambridge, MA 02139 Cambridge, MA 02139
jwittenb@athena.mit.edu jsterman@mit.edu
Introduction
Scholars have long attempted to understand the nature of scientific change. Is science characterized by the
steady application of universally-accepted norms of logical inquiry, or is it an enterprise that periodically
reconstructs itself from new fundamentals? One of the best-known examples of the latter view is Thomas
S. Kuhn’s Structure of Scientific Revolutions. Kuhn argues that new theories replace old ones rather
than build upon them, and in the process revolutionize science's very image of itself (1962: 84-85).
Scientific progress is seen not as a steady accumulation of truths, but “as a succession of tradition-bound
periods punctuated by non-cumulative breaks" (Kuhn 1970: 208).
Kuhn's theory has had enormous influence in the social sciences, but it is also of enduring interest in the
physical sciences (Barnes 1982; Lightman and Gingerich 1992). The notion of paradigm has, rightly or
wrongly, been used to legitimate alternative methods of research as well as to delegitimate dominant
modes of inquiry. Nonetheless, although ‘paradigm competition’ has become well-established in the aca-
demic lexicon, little is known about what such competition actually entails. How do internal and contex-
tual forces interact to shape and constrain the development of new paradigms? Why do some paradigms
last for centuries while others quickly wither?
Purpose
We address these questions with a formal dynamic model of paradigm competition. The model is based on
Sterman's (1985) model of Kuhn, but modified to represent explicitly the competition among different
paradigms. Although these models are inspired by Kuhn's work we do not claim to have fully captured
his theory. Translating the theory from its qualitative, highly abstract written form into an internally
consistent, formal model has involved many simplifications. Indeed, making explicit the causal connec-
tions that we and others readers of Kuhn routinely take for granted has required the introduction of conjec-
tures. Kuhn might even disagree with (Wittenberg 1992; but see also Sterman 1992, Radzicki 1992 and
Barlas 1992.) Nonetheless, formalization has advantages. Most discussions of Kuhn's theory are based on
ambiguous mental models, and Kuhn's work itself is textual, rich with ambiguity, multiple meanings,
and implicit assumptions. More important, Kuhn offers no calculus by which one can assess whether the
dynamics he describes can be produced by the causal factors he postulates. Formalization helps to surface
auxiliary assumptions so they can be debated and tested. We see formalization as complementary to the
work of philosophers and historians of science attempting to verify empirically theories of scientific
change (e.g. Donovan, Laudan and Laudan, 1988). Second, Kuhn's theory is one example of a broader
class of theories of revolutionary change. The model may provide insights into how revolutionary
upheavals occurs in other domains such as the social sciences (see Kuhn, 1970: 208-209; Gersick, 1991,
Tushman and Anderson 1986). Finally, the model applies nonlinear dynamics to sociological phenomena.
It describes emergent processes, and model behavior is at all times path-dependent.
A Theory of Paradigm Development
Rather than summarize Kuhn's theory here, we assume familiarity with Kuhn's work and the many inter-
pretations and altematives to it (e.g. Lakatos and Musgrave 1976). An important aspect of Kuhn's theory
for purposes of modeling is the life cycle of a “typical” paradigm. Kuhn describes a sequence of four
stages: emergence, normal science, crisis, and revolution (followed by the emergence of a new paradigm).
The emergence phase is characterized by the absence of commonly-accepted beliefs or standards governing
scientific activity. Conflict among paradigm-candidates is thus rooted in incompatible metaphysical
beliefs and logics of inquiry. Such conduct characterized electrical research before the work of Franklin and
his successors provided the field with a paradigm (Kuhn 1970: 13-15). Once a theory attracts nearly every
scientist in the field - thereby becoming a dominant paradigm — normal science begins. Here scientists
cease to debate fundamental methodological tenets, and, convinced that their paradigm is the proper way
to characterize reality, proceed to apply it to nature's puzzles. When clashes between theory and reality do
occur, they are more often than not resolved in favor of theory. Thus, for example, by the early twentieth
century physics had become so identified with Newton's Principia that no one questioned Newton's theory
even though there were persistent discrepancies between it and observations concerning the speed of sound
and the motion of Mercury (Kuhn 1970: 81). A paradigm enters crisis when enough unsolved puzzles are
recognized as important anomalies. Increasing numbers of scientists will devote their time to solving
these anomalies rather than other puzzles, and some will propose radical solutions. A revolution occurs
when a new paradigm based on such a radical idea is adopted, and science is reconstructed from new fun-
damentals, Einstein's theory of relativity is a well-known example of a revolutionary theory, in which
basic notions of space and time were fundamentally reconceptualized. Obviously the timing, length, char-
acter, and context of each stage differ from case to case. For example, a dominant paradigm in crisis may
quickly be replaced, or a crisis may deepen for decades as new theories fail to sprout or flower. The social,
political and cultural context, as well as chance factors (the existence of an Einstein, Bohr or Keynes)
may strongly condition the character and timing of the dynamics. Assessing the tension between situa-
tional and structural factors is one of the purposes of the model.
A Model of Paradigm Development
Sterman (1985) presents a dynamic model of Kuhnian paradigm change. The purpose of the model was to
test the dynamic consistency of Kuhn's theory by assessing whether the causal processes Kuhn postulates
can produce the dynamics he describes. To do so, the work deliberately focused on the internal dynamics
of a single paradigm and ignored the explicit dynamics of competition. Wittenberg (1992) argued, how-
ever, that the model insufficiently accounts for paradigm competition, We thus Construct a multi-
paradigm model in which the structure of Sterman's model is replicated for each of the competitor
paradigms, and additional structure is added to specify how paradigms interact.
‘The model creates a simulated ecology of interacting paradigms, each representing a community of practi-
tioners, recruitment and defection from that community, as well as the intellectual activities of the mem-
bers such as formulating and solving what Kuhn calls puzzles, recognizing and trying to reconcile anoma-
lies, and conceiving new theories, The model accounts for attitudes and beliefs of the practitioners within
each paradigm through constructs such as ‘confidence in the paradigm’ and the time required to recognize a
phenomenon as an anomaly which challenges the theory.
The structure of Sterman’s original (1985) model is retained with few modifications; readers are directed to
that work for a complete description of the model structure. Here we provide a brief outline. The essence
of Sterman’s dynamic hypothesis is the notion that the average difficulty of the puzzles to be solved by
the paradigm increases as the cumulative number of puzzles solved grows. This ‘paradigm depletion’ rep-
Tesents the idea that each paradigm is a limited model of reality which may apply well in the domain of
phenomena it was originally formulated to explain, but will be harder and harder to apply as scientists
extend it to new domains. Newtonian mechanics worked brilliantly for macroscopic, slow masses, but
was harder to apply successfully to the domains of the very small or very fast. As the difficulty of
puzzles grows, puzzle solving may slow and more unsolved puzzles may become recognized as anoma-
lies. If the stock of anomalies grows too large, the confidence practitioners have in the truth or utility of
the paradigm may fall, initiating a self-reinforcing collapse as anomalies destroy confidence, and falling
confidence increases the ability and willingness of practitioners to see the gaps in the theory.
The focal point of the model is a construct called ‘confidence in the paradigm’. Confidence determines
how anomalies are perceived, how practitioners allocate their research, and recruitment to and defection
from the paradigm. It represents the basic beliefs of practitioners which structure reality, encompassing
both logical, cultural, and emotional factors. Confidence is defined between 0 (absolute conviction the
paradigm is false, nonsensical, superstition) through .5 (maximum uncertainty as to its truth) to 1
(absolute conviction the paradigm is truth). Pressures leading confidence to change arise both from within
a paradigm and from comparisons with other paradigms. Confidence tends to decline when the number of
anomalies exceeds an acceptable level, or when progress in puzzle-solving slows. The impact of anoma-
lies and progress is mediated by the level of confidence itself. High levels of confidence preclude rapid
changes in confidence because practitioners, utterly committed, resist any evidence contrary to their
- 828 -
beliefs. Practitioners with only lukewarm commitment, lacking firm reasons to accept or reject the
paradigm, are far more likely to change their confidence if any significant evidence appears.
The external factors affecting confidence encompass the way in which practitioners in one paradigm view
the accomplishments of other paradigms. We distinguish between the dominant paradigm, which we
define as that paradigm commanding the allegiance of the most practitioners, and alternative paradigms.
Confidence in an alternative paradigm tends to increase if its number of anomalies is less than that of the
dominant paradigm, or if it has solved more puzzles. It tends to decrease if the dominant paradigm has
fewer anomalies or more solved puzzles. Alternative paradigms compare themselves with one another as
well as with the dominant paradigm. Confidence in an alternative paradigm tends to decrease if it has
more anomalies or fewer solved puzzles than the largest of the other alternatives.
According to Kuhn, the normal science is puzzle solving. In the model, the rate at which scientists
formulate and solve these puzzles are solved depends on the number of practitioners, the fraction of their
time devoted to puzzle solving, and the intrinsic difficulty of the puzzles. Under normal conditions a
puzzle, once formulated and attacked, will be solved in fairly short order, adding to the cumulative
stockpile of knowledge generated by the paradigm. But as the intrinsic difficulty of puzzles grows, a
growing number will resist solution long enough to be recognized as anomalies.
Anomaly recognition is a subtle psychological process, mediated in the model by confidence. Confidence
influences the perception of anomalies in two ways, Confidence determines the degree to which
practitioners are conditioned to see reality as consistent with their paradigm. Increases in confidence will
slow the recognition of anomalies since practitioners are becoming more blinded by the paradigm, and
thus take a longer time to recognize the problems that do arise as anomalies. Anomalies may sometimes
be resolved into the theory, thus ending a potential threat to the paradigm. The rate at which anomalies
are resolved depends on the number of practitioners in sanctioned research, the fraction of those involved
in anomaly resolution, and the‘average difficulty of anomalies. Anomalies are assumed to be more diffi-
cult to solve than puzzles, and as the difficulty of puzzles increases, the difficulty of anomalies rises as
well. The fraction of practitioners involved in anomaly resolution depends on the balance between the
number of anomalies and the acceptable number. The acceptable number of anomalies is the number that
can be tolerated without losing confidence in the paradigm. If the number of anomalies increases, addi-
tional scientists are drawn into anomaly resolution in an attempt to solve the major outstanding problems
challenging the theory, as for example the Michelson-Morley experiment drew forth many efforts to rec-
oncile Newtonian theory with the constancy of the speed of light with respect to relative motion.
The population bf practitioners committed to each paradigm is endogenous, increasing with recruitment
and decreasing with retirement of elder scientists and defection of others to competing paradigms. We
assume for,simplicity that the total population of scientists in all paradigms is constant: Scientists who
leave one paradigm enter another; and entry of young scientists is balanced by retirement of the old. The
assumption of constant total population simplifies the interpretation of the. results but is in no way
essential to the main conclusions; it can easily be relaxed in future versions. Practitioners defect based on
their confidence relative to the confidence of those in the dominant paradigm. The greater the (negative)
discrepancy between a challenger's confidence and confidence in the dominant paradigm, the larger the
proportion of the challenger's practitioners that will defect. The overall magnitude of the defection is de-
termined by the number of practitioners in the paradigm. Recruitment is proportional to a paradigm's rela-
tive attractiveness and its total number of practitioners, The greater a paradigm's attractiveness, the greater
the proportion of defectors it will recruit. Attractiveness is proportional to the number of practitioners
since large paradigms are assumed to get more funding, train more students, and have a larger voice in
tenure and other peer-career decisions than small paradigms. Attractiveness also depends on the confidence
of the paradigm’s practitioners. Here confidence measures the excitement, enthusiasm, and progress flow-
ing from a successful endeavor — scientists are naturally drawn to outstanding examples of achievement.
The most significant difference between the original and present models is the explicit representation of
the creation of new paradigms. We model the creation of a new paradigm as a stochastic event whose
probability depends upon the distribution of practitioner activities in the currently dominant paradigm
among normal science (puzzle-solving), anomaly resolution (the attempt to reconcile anomalies with the
current paradigm), and other activities (described by Kuhn as including philosophical reconsideration of
the paradigm and other activities which are not sanctioned by the dominant paradigm). In general, each of
these activities may result in the creation of a new paradigm, but the probability that a new paradigm is
created as a result of a practitioner year of effort devoted to each activity may differ. Thus:
PA, = Pps * PPS: + Par * PAR: + Poa * POA;
= probability a new paradigm is created (per year)
actitioners in the dominant paradigm engaged in puzzle-solving (practitioners)
probability of creating a new paradigm per practitioner year of effort in anomaly resolution
Poa = Probability of creating a new paradigm per practitioner year of effort in other activities
Following Kuhn, we assume that normal science is unlikely to produce new paradigms, focused as it is
on solving puzzles within the context of the existing paradigm. Other activities are more likely to pro-
duce a new paradigm, while effort devoted to anomaly resolution is most likely to result in the creation of
tadical new theories which can form the basis for a new paradigm. Thus Par > Poa > Pps- In the model,
the distribution of effort among these three activities is endogenous. Thus the probability that a new
Paradigm will be created in any time period is endogenous and will vary as practitioner effort changes in
Tesponse to.the changing health of the dominant paradigm. Once a new paradigm is launched, we assume.
it begins with a small number of practitioners (five), a confidence level equal to .5 (neutral), a very small
stock of solved puzzles and no initial anomalies. The newly launched paradigm must then compete for
members against other existing paradigms and will succeed or fail to the-extent it can (1) solve puzzles
and resolve anomalies such that confidence in that paradigm grows; and (2) prove more attractive than
other paradigms against which it might be competing. Note that it is possible, and indeed given the prob-
abilities we assume, likely, that during a period of crisis, when many practitioners in the dominant
paradigm abandon puzzle solving, that the probability of creating a new paradigm may rise and remain
high long enough for more than one new paradigm to be launched. In this.case, the newly created
paradigms will vie for ascendancy not only against the dominant paradigm but against one another.
Exploring the Dynamics of Paradigm Development
In order to disentangle the internal and contextual factors underlying paradigm change we first present a
simulation in which only one new paradigm may emerge out of the crisis of the previously dominant
paradigm (figure 1), We assume the same high potential explanatory power used in Sterman (1985). We
begin the simulation with a dominant paradigm in the full flower of normal science, with 100% of the
Practitioners committed to that paradigm, a high level of confidence, and few anomalies. However, as
puzzles gradually become more difficult to solve, anomalies slowly accumulate, eventually leading to
crisis and a drop in confidence. The new paradigm is created when confidence in the dominant paradigm
falls below 0.7, in this case in year 124.75. Figures 2 and 3 illustrate the details of paradigm 2's life
cycle. In the early period (years 125-170), confidence rises dramatically, since initial puzzle-solving
progress is great and anomalies are low. The paradigm, initially untested, proves itself capable of solving
puzzles, and thus attracts more practitioners, further boosting confidence. The virtuous circle of rising
confidence, faster recruitment and puzzle-solving, leading to further boosts in confidence in the new
paradigm bootstraps the new paradigm and accelerates the decay of the old as it is increasingly starved of
practitioners, until the new paradigm dominates the entire field (about year 190), signalling the beginning
of normal science organized around a different underlying metaphor, method, and metaphysics.
Normal science, a period of high productivity in which practitioners engage primarily in puzzle-solving
and are blinded to potential anomalies by their faith in their paradigm, occurs approximately between
years 190 and 300. As the paradigm is elaborated and solved puzzles grow, however, puzzles become
more difficult to solve and anomalies slowly accumulate. Although the fraction of all practitioners
committed to the paradigm remains high throughout the period, confidence peaks in year 240 and slowly
falls, as does the fraction of practitioners engaged in sanctioned research (puzzle solving).
Practitioners In Paradigms 1 through 3
Without Competition _
gE 4 in i
227 0.8 4
553
Ess 0-6 4
SRE
ee 0.4 4
SeE~ 0.2 4
£8
0 T
100. 150. 200 250 300
Years
2
é
Lifeeyele of Paradigm 2
Without Competition 5
ge & 1 t be : — - 50 3
2 Ey Confidence In 3 >
Begzio.s as Vv Practitioners F P| zo 33
esegs DEee
Boece 0-6 Fo0g #588
“aos et
BBs 880.4 | F20888 28
ZEsE Bock
SES 80.2 | netatve bio 8835
3S — §- Anomalie: om
£6 8 9 zg 5
wun 100° 150 200° 250° 300. 360° 400 >
eee a Years
Lifecycle of Paradigm 2 >
25 1 7" Without Competition | 25 258
o : Nop
885 too gea®
ees Average Ditticult 2952
c= a YY oo =
855 of Puzzles F153s<5
Los sexe
she E10 s°ae
ooo 258
coe Puzzte-Solving B3eS
S55 Progress E5 28ha
Ba Sak
gs 0 oO e7s
100° "150°" 200° ' 250° "300° 350" "400 &
é
w
Years
By year 300 the paradigm is in crisis due to high anomalies and slowing progress. The positive feed-
backs which had previously caused membership to rise now cause it to decline. The progress of normal
science has increased the difficulty of puzzles, since practitioners have begun to apply the paradigm
beyond the scope for which it was created. This leads to an increase in anomalies, causing practitioners to
leave puzzle-solving, eroding progress and decreasing confidence. Practitioners, increasingly sensitive to
- 831 -
the paradigm's limitations, become more apt to see difficult puzzles as anomalies, thus further increasing
anomalies and decreasing confidence in another positive feedback.
As the number of practitioners engaged in normal science falls and the number in anomaly resolution and
other activities rises, the probability that a new paradigm will be created gradually grows. Around year
320 a new paradigm is in fact launched (figure 1). Since the new paradigm emerges during the crisis of
paradigm 2, it quickly gains adherents while paradigm 2 loses members. Confidence and membership in
paradigm 3 now accelerate sharply through the same processes at work earlier for paradigm 2, and the life
cycle is completed as paradigm 2's confidence and membership fall eventually to 0. The many positive
feedbacks described above create the self-organizing dynamic by which uncommitted and unorganized Prac-
titioners coalesce into a highly focused paradigm with productive normal science. The same positive
feedback processes operate in the opposite direction during the crisis period to accelerate the collapse of a
paradigm which has accumulated sufficient anomalies for confidence to start to fall. Many if not most of
these feedback loops involve processes internal to the paradigm. These loops were captured in the
original model of Sterman (1985). In addition, having extended the model to explicitly account for com-
peting paradigms, several additional positive loops which operate between competing paradigms are now
Tepresented (figure 4; note that the negative loops which ensure global stability are not shown). These
loops reinforce the internal loops such that the overall behavior of a single paradigm going through its
life cycle remains qualitatively similar to the original model.
Solved jf Solved Puzzlesia__. Solved’
+ Puzzles of Avs. B
+ Anomalies aN ed
anbmaty 2) CRA Cr
wn PA '
4 Sontigence + ~| Confidence _
in inB .
Puzzle-Solving Puzzle Solving
Progress NU Ie} +
+
‘) Relative Gs
Aaractvendas
of Avs. B
Practitioners Practitioners
inA _' + yf ‘ce +p inB
+ ky) Net Net +
Recruitment Recruitment
toA
toB
Figure 4. Some of the positive feedback loops captured in the model which create path-dependent
behavior, These loops rapidly differentiate paradigms which might initially be quite similar, and can
amplify small fluctuations in local conditions to macroscopic significance.
- 832 -
We now simulate the model with fully endogenous competition among ‘paradigms. Figures 5 and 6 show
a simulation incorporating both internal and competitive pressures. As in the first simulation, paradigm 1
is initialized in normal science, but now new paradigms are launched with a probability depending upon
the vitality of the dominant paradigm. We further assume practitioners engaged in anomaly resolution
have a greater likelihood of creating a new paradigm than those engaged in other activities, while practi-
tioners in puzzle-solving are assumed never to produce a new paradigm. In these simulations the model is
completely deterministic except for the probabilistic process by which new paradigms are created, and all
paradigms have identical structure, parameters, and potential explanatory power.
The assumption that all new paradigms have the same potential explanatory power is deliberately made to
highlight the processes of competition among newly launched contenders. In the simulation the crisis of
a dominant paradigm may, depending on the random process governing paradigm creation, result in one or
several new paradigms being launched within the crisis period when the probability of paradigm creation
is large. Variations in the timing, length, and character of the life cycles across paradigms can only be
due, therefore, to contextual factors, specifically the number and health of paradigms against which a
newly created paradigm must compete.
Practitioners in Paradigms 1 through 20
identical Potential Explanatory Power
Ze 1
22
58a 0.81 J eC VNALTAIAA wl 7
See v a
E28 0.6 4 L
SBs 0.4 4 E
seé
BES 0.2 4 2 18 E
ro 0 : 1
ae 0 500 1000 1500
Years
Confidence in Paradigms 1 Through 20
— ‘ Identical Potential pase, Power
3 1
za
fe 0.8 4 E
Ee 0.6 4 E
es
3s 0.4 4 L
88 0.2 | L
5 0
me Years 1000 1500
We simulate the model for 2000 years to lessen the influence of initial transients and to exhibit the range
of possible fates for new paradigms. Figure 5 shows a succession of dominant paradigms in which the
initial paradigm gives way to successors whose life cycles follow approximately the pattern seen in the
previous simulation. Because the timing of paradigm creation is fully endogenous and stochastic, there is
- 833 -
considerable variation in the lifetimes despite the equal explanatory power of all paradigms. Paradigm 2,
for example, lives for 375 years, while paradigm 9 survives a mere 115.
What is most interesting about figure 5, however, is not what it displays but what it conceals. Not all
new theories succeed, Paradigms 3, 4, 11-14 and 16- 20 never become dominant. Beneath the apparently
orderly succession of paradigms seen in figure 5 lies considerable turmoil, as newly launched paradigms
compete for dominance. Many face early extinction. Note the slight fluctuation in practitioners during the
dominant phases of paradigms 1, 10 and 15, the telltale sign of paradigm candidates which are launched
and rapidly fail. Figure 5 illustrates what Kuhn calls the invisibility of revolutions, where the linear and
cumulative character of normal science portrayed in the textbooks conceals the messy, uncertain and con-
tentious character of actual scientific practice (Kuhn 1970: 136- 143).
Consider paradigm 12, launched in year 951.25. In a world without competition it would grow as growth
of solved puzzles and lack of anomalies raised confidence, thus attracting more recruits. However,
paradigm 10 dominates science at the time and is far more attractive than the newcomer. Indeed, in year
950 paradigm 10 is still in the midst of normal science. With 10’s greater numbers and higher confidence
a new paradigm stands little chance of survival. By year 975 paradigm 12 is dead.
Consider now paradigms 15 and 16, launched in years 1039 and 1042.25, respectively. Although they
emerge only 3.25 years apart, during paradigm 10's crisis, they suffer very different fates: paradigm 15
comes to dominate the field, while paradigm 16 perishes after a brief spurt. Here the contingency of out-
comes on situational factors is decisive. Significantly, paradigm 15 does not succeed because of its head
start in attracting practitioners: in year 1045 it actually has fewer than paradigm 16! The difference in
their destinies lies in their levels of confidence. Consider the year 1055. Paradigm 15, though equal in
size to paradigm 16, is more attractive to adherents of crisis-ridden paradigm 10 because its adherents,
having had a 3 year lead over paradigm 16 in solving puzzles, have been able to consolidate and articulate
their paradigm more coherently and persuasively than their chief rivals. The small advantage held by
paradigm 15 at time 1055 is amplified as success begets success through the positive loops shown in
figure 4. The greater the confidence of the paradigm, the more focused the puzzle-solving activity and the
higher the rate of progress, further “ boosting confidence; the greater the confidence in the paradigm, the
less able practitioners are to recognize anomalies, thus the lower the number of relative anomalies and the
higher confidence becomes, The greater the confidence in the paradigm, the greater the recruitment of prac-
titioners and the smaller the defection rate, increasing the size and political power of paradigm 15's com-
munity vis 4 vis paradigm 16, further benefiting paradigm 15 in the competition for resources, students,
control of journals and conferences, and so on, By year 1105 paradigm 15 dominates science, while
paradigm 16 has withered, and if remembered at all, is viewed as a blind alley, foolish error, or curiosity.
Note that the death of paradigm 16 is not due to its intrinsic weakness, since it has the same puzzle-
solving potential as paradigm 15 and all others.
The simulation illustrates the subtle interplay between endogenous feedback processes and contextual,
situational factors in determining the dynamics and succession of paradigms. The basic life cycle of
paradigms is determined by the feedback loop structure of the system as discussed above and highlighted
in figure 4. The positive feedbacks which boost confidence and rapidly produce a focused community
from a promising but incoherent new idea create the rapid growth of new paradigms as they bootstrap
themselves into normal science. These same loops are responsible for the resistance of the dominant
paradigms to challenges, as high confidence suppresses the creation and progress of any new theories.
The same loops then create the accelerating collapse of the dominant paradigm once it begins to ex-
perience depletion of the root metaphor which defines it. The prevalence of positive feedback processes in
the dynamics, however, means that contingent situational factors such as the number of practitioners in
the dominant paradigm, their confidence level, the number of solved puzzles in anomalies of the dominant
paradigm, as well as the number of other competing paradigms and their membership, confidence, and
accomplishments strongly condition the fate of new paradigms. While it is obvious that the creation of a
new theory is intrinsically unpredictable, the simulation shows clearly that the likelihood any given new
paradigm grows to dominance or rapidly becomes extinct is strongly contingent on the environment into
which it is launched-- an environment which in turn depends on the entire history of the paradigms which
precede it. The prevalence of positive feedback processes in paradigm development and decay means that
the evolution of the system as a whole is strongly path-dependent.
Although not explicitly modeled here, the fact that each new paradigm differs in intellectual content, pos-
sibly including fundamental epistemic and metaphysical assumptions, means that the succession of
paradigms-- of world views-- is unpredictable, contingent on the prior history of science, and need not re-
flect an arrow of progress or even of consistency as paradigms of equal or even greater intrinsic "merit"
may become extinct while weaker paradigms grow to dominance solely as a function of the context into
which they are launched,
To illustrate, figures 7-8 show a simulation in which the intrinsic puzzle-solving capability of each
paradigm is different. Specifically, the rate at which puzzle-solving becomes difficult as cumulative
puzzles accumulate (the paradigm's "inherent potential") is chosen randomly, The evident differences in
the duration of paradigm dominance cannot be explained solely by variation in the paradigms’ intrinsic
potentials. Thus, paradigms 13 and 18 have approximately the same potential, yet paradigm 13 outlives
paradigm 18 by 175 years. The simulation shows how the context into which a new paradigm is launched
may dominate its intrinsic puzzle solving capability. For example, paradigm 16, endowed with a poten-
tial approximately twice that of paradigm 13, fails because it is launched while paradigm 13 is still attrac-
tive enough to retain practitioners. Paradigm 16 is thus not able to recruit any practitioners and build a
coherent body of knowledge. The sensitivity of outcomes to context is further exemplified by the fate of
paradigm 17. It has less explanatory power than any one of the paradigm candidates 14 through 16, yet
nonetheless becomes the successor to paradigm 13.
Practitioners in Paradigms 1 through 20
Random Potential Explanatory Power
Ze 1
oa H
Soa 0.8 4: / o ¢ f f E
ses ia] 2
E28 0.6 4 E
ba 3 A 0.4 4 -
see
SE>. 0.2 4 E
no 0 ; : r +
ro] 0 400 800 1200 1600 2000
* Years
Confidence in Paradigms 1 Through 20
E 1 Random Potential Explanatory Power
ce Ca ‘
BF og 17) * OVAL 7 E
ae \ iby 20
Eg 0.6 4 E
ce
£3 0.4 4 bE
ge
38 0.2 4 E
5° 0
2 0 400 800 1200 1600 2000
Fig & Years
Conclusion
The present work extends Sterman’s original (1985) model to portray explicitly the endogenous emer-
gence of and competition with new paradigms. Results show that consideration of competing paradigms
does not alter the essential dynamics of the paradigm life cycle, lending some confidence that the feedback
Processes captured in that model are robust in their ability to generate the collective behavior associated
with emergence, normal science, crisis and revolution as Kuhn describes it. The addition of explicit com-
petition among paradigms, however, adds significant new insight into the importance of situational con-
tingencies in the succession of paradigms. Because paradigm growth and success are strongly conditioned
by multiple positive feedback processes, historical contingencies can be decisive in determining which of
several newly launched paradigms survives. The simulations show clearly that the health of existing
paradigms at the time a new paradigm is launched, other initial conditions surrounding the emergence of a
new theory, and inherently unpredictable events associated with a small number of individuals may be
more important in determining the fate of any particular paradigm than its intrinsic explanatory power,
logical force, or other ‘rational’ factors.
Indeed, the simulations show that historical context can easily cause a paradigm with greater ultimate po-
tential to be eclipsed by a weaker one. The model thus identifies specific processes by which phenomena
Kuhn highlights-- such as anticipations and the invisibility of revolutions — might arise. The model,
however, is clearly highly simplified and cannot capture the full scope of sociological, intellectual, cul-
tural, and other factors which impinge on activities as basic to society as scientific theory-building. We
do not argue here that this model captures all the subtleties of Kuhn's theory, nor even that it represents a
correct or comprehensive model of scientific activity. Plainly it does neither. Rather, we seek to demon-
strate that it is both desirable and possible to capture in a formal model the causal hypotheses embodied
in written theories of scientific endeavor which are alleged by their authors to produce the dynamics as
those authors see them. The process of formalizing such hypotheses demands a discipline which surfaces
inconsistencies, implicit assumptions, glosses and errors in the mental simulations authors necessarily
perform to infer the dynamics of science from their theories of its structure. Such an endeavor is worth-
while as a complement to historical studies and other analyses. As in Sterman (1985), complete docu-
mentation of the model is available; we invite others to replicate, critique, revise, and extend the model to
model and test views of scientific activity which differ from ours.
References
Barlas, Yaman (1992) Comments on 'On the very idea of a system dynamics model of Kuhnian science’.
System Dynamics Review. 8(1), 43-47.
Bames, Barry (1982) T.S. Kuhn and Social Science. New York: Columbia University Press.
Donovan, Arthur, Larry Laudan and Rachel Laudan, eds. (1988) Scrutinizing Science. Dordrecht, The
Netherlands: Kluwer Academic Publishers.
Gersick, Connie J.G. (1991) Revolutionary Change Theories: A Multilevel Exploration of the
Punctuated Equilibrium Paradigm. Academy of Management Review. 16(1), 10-36.
Kuhn, Thomas S, (1962) The Structure of Scientific Revolutions. Chicago: University of Chicago Press.
1970. The Structure of Scientific Revolutions. 2nd ed. (Chicago: University of Chicago Press.
Lakatos, Imre and Alan Musgrave (1976) Criticism and the Growth of Knowledge. Cambridge:
Cambridge University Press.
enone, ae and Owen Gingerich (1992) When Do Anomalies Begin?, Science, 255, 7 February,
Radzicki, Michael J. (1992) Response to ‘On the very idea of a system dynamics model of Kuhnian
science’. System Dynamics Review. 8(1), 49-53.
Sterman, John D. (1985) The Growth of Knowledge: Testing a Theory of Scientific Revolutions with a
formal model. Technological Forecasting and Social Change 28(2),93-122.
Sterman, John D. (1992) Response to 'On the very idea of a system dynamics model of Kuhnian science’.
System Dynamics Review. 8(1) 35-42. .
Tushman, M. and Anderson, P. (1986) Technological Discontinuities and Organizational Environments.
Administrative Science Quarterly, 31, 439-465.
Wittenberg, Jason (1992) On the very idea of a system dynamics model of Kuhnian science. System
Dynamics Review. 8(1) 21-33.
- 836 -
CC BY-NC-SA 4.0