Some Applications of Complexity Theory to Health

Care
December 1, 1998
Prepared for
Business Enterprise Solutions and Technologies.
Veterans ealth !dministration
Department of Veterans !ffairs
Prepared b"
Science !pplications #nternational $orporation
ealth $are Technolog" Sector
1%&'% $ampus Point Dr.
San Diego, $a. 9&1&1
Tom (unnec)e (unnec)et*saic.com
The Complexity Crisis in Health Care
The hospital, as Peter Druc)er sa"s, is one of the most comple+ organi,ations in our societ". The
collecti-e efforts of the !merican health care industr" present a le-el of comple+it"
unimaginabl" greater than .ust a single hospital. $onfounding this comple+it" are serious global
public health, epidemics, ne/ technolog", research, economics, ethics, and political issues.
The result of all these interacting issues is a comple+it" crisis of enormous proportion.
Traditional approaches to deal /ith comple+it" are based the notion that s"stems can be
controlled b" creating more order /ith policies, controls, and specificit". The are a based on a
cogniti-e 0di-ide and con1uer0 approach. Di-ide a big problem into littler ones, and sol-e the
littler ones. The assumption is that sol-ing the di-ided and con1uered approaches /ill sol-e the
big problem. The immensit" of the health care crisis is such that there is no 0top0 from /hich to
do a top do/n anal"sis or create a top do/n po/er structure.
E+perience and modern notions of comple+it" theor" pro-ide an alternati-e.
Dee oc), founder and first $E2 of Visa #nternational, has dealt /ith the problem of
comple+it" /ithin the ban)ing industr". e has coined a term for thin)ing about s"stems and
organi,ations, and the relationship bet/een order and chaos3
Hock had read about the Santa Fe Institute and its work on "the edge of chaos": the notion that
healthy, adaptive systems will always exhibit a kind of dynamic tension between chaos and order It fit
in beautifully with the dynamic tension that he!d set up in "isa: encourage as much competition and
initiative as possible throughout the organi#ation $ "chaos" $ while building in mechanisms for
cooperation $ "order" Hock had even coined a new word to describe this kind of tension % system
that was both chaotic and ordered was "chaordic" &hus you had the ""isa chaord," the "brain
chaord," and so on
4rom this e+perience, he de-eloped a notion of 0$haordic s"stems30
'y (haord, I mean any self$organi#ing, adaptive, nonlinear, complex community or system, whether
physical, biological or social, the behavior of which exhibits characteristics of both order and chaos
)r, more simply stated, a (haord is any chaotically ordered complex
oc)5s -ision /as deri-ed from his highl" successful start up of Visa #nternational, and
benefits from man" "ears of e+perience dealing /ith large, comple+ s"stems.
Growing vs. Building Systems
2ne /a" of loo)ing at comple+ s"stems is thin) of 0gro/ing0 rather than 0building0 a
s"stem. 4or e+ample, a bridge is a -er" specific ob.ect. #ts beha-ior is predictable, and engineers
can calculate the stresses on an" component /ith great precision. The total bridge is the sum
total of its parts. #t can be 0built.0 $omple+it" is designed into the s"stem.
! garden, ho/e-er, is different. !lthough /e can spea) of building a garden, /e are
actuall" building the en-ironment in /hich the garden /ill gro/. 6e can build the beds of soil,
irrigation s"stem, drainage, etc., but /hen it comes to planting it, it is a gro/th process. 6e
create the initial conditions b" planting seeds, and then /e control the en-ironment as the" gro/,
but /hat actuall" gro/s is not as predictable or precise as building a bridge. Thus, /e spea) of
gro/ing a garden, rather than 0building0 tomato plants. $omple+it" emerges o-er time, the result
of the interaction of the plants, nutrients, en-ironment, and random occurances.
The $haordic s"stem can be seen as a combination of building the initial conditions and
tending the gro/th process. The e+act outcome of the process is not guaranteed, but rather a
result of the interactions bet/een the elements of the s"stem.
The patterns can be described as follo/s3
1. Creating an initial condition. These are the seeds or the 0primordial soup0 from /hich
the s"stem /ill emerge. 6hat is critical at this stage is not the si,e of the initial s"stem,
but rather that it contains the appropriate components for future gro/th.
&. Creating the fitness function. 6hat determines success in this en-ironment7 This is
critical in determining the e-olution of the s"stem. #f the s"stem re/ards a particular
beha-ior 8good or bad9, it can be e+pected to flourish as time goes on.
:. Determining constraints. 6hat are the boundaries, outside of /hich the s"stem /ill not
be allo/ed to stra"7
;. Providing a space for interaction. The elements need to be able to interact and e-ol-e
accordingl". <ote that the #nternet has pro-ided a much more highl" coupled
en-ironment /ithin /hich interaction can occur. #n certain t"pes of organi,ations, this
can be -ie/ed as an 0associati-e -orte+0 around /hich the s"stem re-ol-es.
=. Give it time. >ust as nine /omen can5t ma)e a bab" in one month, the emergent
properties of a comple+ s"stem cannot be detected immediatel".

Based on simple initial conditions, the s"stem e-ol-es to greater degrees of comple+it",
as components gro/ and interact according to the fitness function. The e-olution of the s"stem is
constrained b" the constraints, outside of /hich components ma" not stra".
oc)5s -ision is -er" timel" to the V!, as /ell as the health care industr" in general. Some of
his obser-ations are3
"&he better an organi#ation is, the less obvious it is," he says "In "isa, we tried to create an invisible
organi#ation and keep it that way It!s the results, not the structure or management that should be
apparent" &oday the "isa organi#ation that Hock founded is not only performing brilliantly, it is also
almost mythic, one of only two examples that experts regularly cite to illustrate how the dynamic
principles of chaos theory can be applied to business
*hat he read convinced him that the command$and$control model of organi#ation that had grown up
to support the industrial revolution had gotten out of hand It simply didn!t work (ommand$and$
control organi#ations, Hock says, "were not only archaic and increasingly irrelevant &hey were
becoming a public menace, antithetical to the human spirit and destructive of the biosphere I was
convinced we were on the brink of an epidemic of institutional failure" He also had a deep conviction
that if he ever got to create an organi#ation, things would be different He would try to conceive it
based on biological concepts and metaphors +embers are free to create, price, market, and service
their own products under the "isa name," he says "%t the same time, in a narrow band of activity
essential to the success of the whole, they engage in the most intense cooperation" &his harmonious
blend of cooperation and competition is what allowed the system to expand worldwide in the face of
different currencies, languages, legal codes, customs, cultures, and political philosophies
,o one way of doing business, dictated from head-uarters, could possibly have worked "It was
beyond the power of reason to design an organi#ation to deal with such complexity," says Hock, "and
beyond the reach of the imagination to perceive all the conditions it would encounter" Instead, he
says, "the organi#ation had to be based on biological concepts to evolve, in effect, to invent and
organi#e itself"
The Chaos Order Spectrum
2ne e+treme of orderl" s"stems came out of the industrial re-olution. 4rederic) 6inslo/ Ta"lor,
in-entor of 0scientific management0 thought that /e could 0scientificall"0 determine the one
correct /a" of doing things, and then ruthlessl" appl" this method to those doing the /or). !s he
said in 19%'3
"In our scheme, we do not ask for the initiative of our men *e do not want any initiative %ll we want
of them is to obey the orders we give them, do what we say, and do it -uick .sic/"
This )ind of thin)ing considers the organi,ation to be a machine, in /hich principles of
optimi,ation can be applied. Bureaucracies tend to operate in this manner, in /hich formall"
defined standards and procedures control the organi,ation. Each person in the organi,ation does
his or her .ob as if the" /ere interchangeable elements in a machine.
This ma" /or) for some t"pes of acti-ities, but health care does not operate according to
these orderl" conditions. #t cannot be dri-en b" 0one correct /a"0 st"le of thin)ing, for there is
immense -alue to be deri-ed from )no/ledge sharing, collaboration, education, research, and
man" other forms of conte+t?specific acti-ities /hich are uni1ue to the indi-idual. <either
human ph"siolog" nor the health care s"stem is amenable to understanding from the perspecti-e
of a precise, linear machine.
The other end of the spectrum is chaos, or completel" random and uncontrolled beha-ior.
Each element of a s"stem acts independentl", and /ithout regard to others. This is a feared state,
and man" rules and regulations are promulgated in an effort to pre-ent it.
Some/here bet/een these t/o e+tremes lies a common ground @ /hat Dee oc) calls
the chaordic model. #t is built on the notion of s"stems being in tension bet/een order and chaos,
and e-ol-ing o-er time. S"stems start simpl", and comple+it" emerges o-er time as a result of
interaction /ith their en-ironment, based on principles of natural selection against a fitness
function.
The Patient as a Billiard Ball
The intellectual ad-enture leading from the apple falling on #saac <e/ton5s head to
landing a man on the moon is one of humanit"5s great success stories. Ph"sics and engineering
ha-e gro/n immensel" based on <e/ton5s concepts. 4or man" generations, ph"sicists ha-e
h"pothesi,ed the beha-ior of billiard balls on a frictionless pool table. The" could appl" simple
la/s to predict the beha-ior the interaction /ith great precision. Their success is the en-" of
man" other disciplines, /hich hope to reduce the comple+it" of their problem set to something
so precise, repeatable, and predictable. The simple elegance of <e/ton5s F0+% formula eludes
other disciplines such as economics, sociolog", management, or health care.
ealth care is a particularl" comple+ discipline. Patients are not li)e billiard ballsA the"
are not nearl" as predictable or as interchangeable. There is no 2ne $orrect 6a" to deal /ith
health @ e-er"one has a different idea of /hat health is, and ho/ to achie-e it. The placebo effect
alone creates a /all of uncertaint" /hich deals a fatal blo/ to an" attempts of reducing health
care to simple linear e1uations.
The health care en-ironment is not li)e a frictionless billiard table, either. ealth care is
affected b" group beha-ior, public health issues, epidemics, the media, culture, and public
trends. 2ne cannot isolate health care to be the sum of the health of all of the indi-idualsA the" all
interact in comple+, adapti-e /a"s.
The health care industr" is fundamentall" different than sa", the oil industr". 6hereas the
global suppl" of oil is diminished b" e-er" barrel lifted from reser-es, there is no central suppl"
of health /hich is diminished /hen health care is pro-ided to an indi-idual. #n fact, the
impro-ement of an indi-idual5s health can impro-e the health of e-er"one else. 4or e+ample,
someone /ho resol-es a drin)ing problem not onl" impro-es their o/n health, but reduces the
potential for drun) dri-ing accidents, /hich affect the public as a /hole.
<or does one person5s impro-ement in health necessaril" come at the e+pense of
someone else5s3 it is not /hat economists call a ,ero sum game. E-er"one can become healthier
simultaneousl".
2ur econom" and management approaches are not set up to deal /ith this )ind of
beha-ior. The economist5s suppl" and demand theories deal /ith scarce commodities. ealth
does not ha-e to be a scarce commodit"A nor does it conform to the traditional notion of suppl"
and demand. !lcoholics !non"mous has impro-ed the health of hundreds of thousands of people
and reduced societal costs. Bet it is a -olunteer, no?cost program. There is no price, no
transactional -alue to the organi,ation. <o bills are generated, the health care industr"5s re-enues
are not increased, the CDP is not increased.
o/ do /e deal /ith the d"namics of such an organi,ation7 !! is a chaordic model, -er"
successful in transforming its members. Deduced to its basic, ho/e-er, the" pro-ide a
communit" /ithin /hich people communicate. o/ man" other health situations can be
addressed in a similar manner7 6hat can /e do to impro-e our health in other /a"s /hich
e+ploit the transformational po/er of communities /or)ing in a collaborati-e en-ironment7
o/ can /e encourage and support these transformations, instead of ignoring them7 6ithin the
ans/ers to these 1uestions lie a )e" to sol-ing our health care comple+it" crisis.

Other xamples o! Chaordic "ehavior
The World Wide Web
The 6eb began as a remar)abl" simple set of initial conditions, standards for the EDF,
TTP, and T(F. This formed the 0primordial soup0 from /hich the other aspects of the /eb
ha-e emerged. The designer of the /eb, Tim Berners?Fee, emplo"ed these techni1ues
intentionall". !lthough the original problem to be sol-ed b" the /eb /as the interchange of
information /ithin the Ph"sics communit", Berners?Fee built an infrastructure of far more
general purpose. Dather than .umping directl" into 0orderl"0 content, such as defining /eb sites
1?1%% for 1uantum ph"sics, 1%1?&%% for plasma ph"sics, etc., he created a 0chaotic0 space in
/hich an"one could publish an"thing the" /anted /ith an" name the" /anted, sub.ect to
#nternet naming con-ention constraints. This lead the emergence of ne/ technologies to deal
/ith that gro/th. Search engines emerged to compete for consumer attention to help them find
information the" need. 6eb 0Portals0 ha-e emerged to gi-e -ie/ers a common access, -irtual
communities emerged to allo/ people to meet others, shopping 0bots0 are emerging to allo/
consumers to find the lo/est cost product, and electronic commerce is rapidl" becoming an
acceptable acti-it".
6e can characteri,e the /eb as the follo/ing3
1. Initial conditions3 The minimal technologies to allo/ h"perte+t communication across
the #nternet. These /ere Eniform Desource Focator 8EDF9, to pro-ide a uni1ue name for
e-er" element on the /eb, the "perTe+t Transport Protocol 8TTP9, to e+change this
information, and "perTe+t (ar)up Fanguage 8T(F9 to allo/ documents to lin) to
each other.
&. Fitness Function3 That /hich people paid attention to sur-i-ed. $onsumer attention has
become a commodit" in the information age, and sites /hich attract this attention ha-e
sur-i-ed and become dominant in the /eb.
:. The Associative Vortex3 The #nternet Domain name s"stem, /hich allocates the names
of /eb sites, such as WWW.VA.GV
;. !mergent Properties. Search engines to allo/ a personal perspecti-e of /eb
information, 8!lta-ista9, portals /hich pro-ide editorial ser-ices 8BahooG9, electronic
commerce 8!ma,on.com9, and a host of other global enterprises.
=. Time. 4rom its inception in 1989 until 199;, the usage of the /eb /as limited to onl" a
relati-el" small group of #nternet e+perts. 4rom 199; until 199', the /eb became popular
as a 0obb"0 acti-it", entering its high gro/th stage. 4rom 199' on/ard, it has become a
per-asi-e part of modern media.
The VA Underground Railroad
#n 19H8, a group of V! people /ere gathered andIor hired b" Ted 25<eill and (art"
>ohnson, then /or)ing for the $omputer !ssisted S"stems Staff 8$!SS9 in central office.
!lthough the group /as originall" chartered b" central office, the decentrali,ed ambitions of the
group created man" tensions in the organi,ation, /ith the result of the central office leaders
being dismissed. o/e-er, the field personnel continued to /or) together in an 0underground0
capacit", /ith no formal organi,ation. ! small group met in 2)lahoma $it" V! in December,
19H8 to formulate a strateg" for a decentrali,ed computer s"stem for the V! and the federal
go-ernment. 2ut of this meeting came the initial conditions for /hat /ould later become the
Decentrali,ed $omputer Program 8D$P9 in the V!, $omposite ealth $are S"stem 8$$S9 in
the Department of Defense, and the Desource and Patient (anagement S"stem 8DP(S9 in the
#ndian ealth Ser-ice. The 19H8 group foresa/ a go-ernment?/ide health care information
s"stem, the role of minicomputers 8contro-ersial at the time9 being replaced b" microcomputers,
the need for a "ear &%%% compatible date, and the -alue of de-eloping public domain, open
s"stems solutions.
The group operated according to man" of the principles outlined in Dee oc)5s $haordic
-ision. #t /as a -er" small group of people, focused on -er" /ell defined 0core0 principles. Then
$hief (edical Director Donald $ustis sa/ an earl" -ersion of the soft/are, and 1uipped, 0#t
loo)s li)e /e ha-e an underground railroad here.0 4inall" ha-ing a name, the group had
membership cards printed up, and held ban1uets offering a/ard certificates for 02utstanding
Engineering !chie-ement on the Enderground Dailroad0 for de-elopers, and 0Enlimited Passage
on the Enderground Dailroad0 certificates for V#Ps and others s"mpathetic to the cause. The
soft/are e-entuall" spread to e-er" federal health care pro-ider in the Enited States, as /ell as
organi,ations in 4inland, Eg"pt, Pa)istan, <igeria, and Cerman". ! -er" small group of people,
/or)ing together /ith a /ell defined set of goals, created a -er" large s"stem.
6e can characteri,e D$P5s e-olution as follo/s3
1. Initial conditions3 $ore computer technolog" to allo/ de-elopment and distribution of a
decentrali,ed soft/are, a communit" of people /ith a strong interest in seeing it gro/.
The technolog" /as minimalistic, the entire computer s"stem being based on a single
language, /ith one data t"pe, 19 commands, and && functions. Participation in the
e-olution of the s"stem -ia user feedbac)A an electronic -irtual communit" 8(ail(an9 to
support the gro/th and e-olution of the technolog" and the communit".
&. Fitness Function3 S"stems /hich conformed to the core standard and met user needs
thri-ed.
:. Associative Vortex3 The data dictionar", /hich controls the definition of terms /ithin
the s"stem.
;. !mergent Properties. 4ar greater lateral communication /ithin hospitals and /ithin
groups across the countr". Focal staff used the s"stem to adapt to man" of the needs of
informal organi,ation. H= different application pac)ages emerged, based on core s"stem.
!s s"stem became larger, it became more 0orderl"0 and under the control of bureaucratic
procedures. $oncepts of e-olutionar" gro/th of core technolog" /ere lost, as /as user
communit" feedbac).
=. Time. 4rom 19H8 to 198&, the s"stem /as limited to a -er" fe/ sites, tended to b"
de-elopers. 4rom 198&?198;, the initial core s"stem /as deplo"ed to all V! sites. 4rom
198'?1998, the s"stem /as deplo"ed on a massi-e scale, /ith additional functionalit".
The #uthor$s xperience with Complexity Theory
The author had man" parallel e+periences to Dee oc), on a much more modest le-el.
e /as one of the founders of the original Enderground Dailroad /hile a V! emplo"ee from
19H8 to 198'. !fter participating and obser-ing the gro/th of the s"stem to other organi,ations,
he became interested in the issues of chaos and comple+it" theor". !fter -isiting the Santa 4e
#nstitute in 199;, he /rote an article entitled 0The 6orld 6ide 6eb and the Demise of the
$loc)/or) Eni-erse.0
e has been see)ing /a"s of appl"ing these concepts to health care since then. 2ne of
these concepts, called ealthSpace, is an attempt to define an information s"stem based on the
transformational -ie/ of the indi-idual, rather than the transactional -ie/ of the health care
enterprise.
#pplying Chaordic Principles to Health Care
The Associative Vortex
The Visa model can be considered to be an associative vortex of a group of competitors
/ho ha-e a need to /or) together despite the fact that the" are competitors. #n this case, the
-orte+ is the clearinghouse role of the organi,ation.
Dee oc)5s model for Visa can be considered an associati-e -orte+. ! loosel" coupled
association of competitors is dra/n together for their common benefit, and acting out of their
o/n self interest, agree to cooperate on a -er" specific set of interactions. The ensuing -orte+ of
acti-ities 0gro/s the pie0 allo/ing each to get a larger slice than if the" operated independentl".
An Associative Vortex for Health Care
The same d"namics can be applied to the problem of health. $urrentl", health care is
organi,ed as a set of competing, -erticall" integrated, closed enterprises. (ilitar" metaphors
abound3 pro-iders spea) of 0captured li-es0 to gro/ their memberships. $ommand and control
principles regulate the relationship bet/een the pa"ors and pro-iders. Principles of scientific
management are e-ident3 s)eptics call it 0medicine b" bod" part.0
The associati-e -orte+ /hich dro-e Visa /as the credit card clearinghouse function. !
chaordic -ie/ of health care creates an associati-e -orte+ around the concept of an indi-idual5s
health. The 1uestion is, /hat is at the center of the -orte+7 This is a profound 1uestion, the
ans/er to /hich could re-olutioni,e health care in !merica.
! )e" to this -orte+ is the notion of the health transformation of an indi-idual. The
transformation is understood in the conte+t of the indi-idual, not the enterprise. #t has a duration,
perhaps the entire life of the indi-idual. #t is multidimensional, goal oriented, and ma" not
necessaril" ha-e a defined end point. Transformations ma" be medical inter-entions 8a hip
replacement surger"9, or the" ma" consist of specific transactions 8a prescription9. The" ma"
consist of pure communication /ithin a trusted communit" 8attending !lcoholics !non"mous9,
or the" ma" consist of education 8reading about a disease on the /eb9, or collaboration
8participating in a support group9. The" ma" be pre-entati-e 8/ellness information9, or
)no/ledge based 8participating in a research stud"9.
The ma.orit" of the health care industr" is dri-en b" transactional considerations. ealth
care is dri-en as a se1uence of pa"ment transactions bet/een pa"or and pro-ider. Pro-iders ha-e
po/erful incenti-es to reduce cost and 0game0 the health care s"stem rather impro-e patient
health. ! common obser-ation is, 06e )no/ that our patients /ill turn o-er for another pro-ider
-er" 1uic)l", so /h" should /e be concerned /ith long term pre-entati-e health practices.0
The perspecti-e of the indi-idual, ho/e-er, is dramaticall" different from that of the
enterprise. 4or e+ample, the" /ould see great benefits to better management of ad-erse drug
e-ents, pre-entati-e medicine, and long?term health care considerations.
Perverse %ncentives in Today$s Health Care System
*hy, .he/ asked time and time again, are organi#ations, whether governmental, commercial,
educational or social, increasingly unable to manage their affairs1 *hy are individuals increasingly
alienated from the organi#ations of which they are part1 *hy are commerce and society increasingly
in disarray1
#n toda"5s health care s"stem, there are a number of incenti-es /hich force an other/ise
/ell?meaning pro-ider to act against the o-erall benefit of the patient and the general health of
the population. These become part of the fitness function of the en-ironment /hich selects
successful pro-iders. #n the case of per-erse incenti-es, ho/e-er, the selection process forces an
e-olution to/ards the negati-e. Some e+amples of these per-erse incenti-es are3
Image vs. Reality
Each pro-ider, in competing for ne/ members, is incenti-i,ed to pro-ide a public image
of a caring, trust/orth" institution that /ill ta)e care of its members. o/e-er, it also has a
strong incenti-e to cut cost. Therefore, the pro-ider is induced to spend mone" in public relations
campaigns, /hich reduces mone" a-ailable for patient care. Thus, one of the top applications of
information technolog" toda" is to identif" high?ris) patients. !lthough it is illegal to 0disenroll0
high ris) patients, pro-iders ha-e strong incenti-es to encourage these patients to lea-e b"
ma)ing it difficult to ma)e appointments or generall" ma)e things difficult for them.
#f pro-ider ! in-ests hea-il" in ad-ertising and cuts bac) in patient care e+penditures, it
/ill attract patients a/a" from Pro-ider B, /ho in-ests lightl" in ad-ertising and hea-il" in
patient care. #n the absence of mechanisms for ob.ecti-e feedbac) to the customers, the s"stem is
dri-en b" the per-erse incenti-e that the mar)et /ill re/ard those pro-iders /ho pro-ide the
inferior health care.
edical record as !egal "efense
#n toda"5s litigious societ", the medical record has become the core of man" malpractice
suits. !s a result, ph"sicians are )eenl" a/are that an"thing the" put in the record can be held
against them. Since e-er" medical procedure has elements of ris) and uncertaint", the medical
record becomes a target for 0(onda" morning 1uarterbac)ing0 in those instances in /hich the
/orst occurred. The fear of ad-erse outcomes and malpractice suits thus dri-es the medical
record to become a defensi-e document, protection in a court of la/ rather than a document to
impro-e the patient5s health.
!lthough some might argue that this forces pro-iders to be more careful, it stiffles much
of the communications -alue of the record. #t is as if antagonistic la/"ers /ere interposing
themsel-es bet/een the ph"sician and patient. The fear induced b" this intrusion restricts the
benefits of patientIph"sician communication and discussion, and dri-es health care into a
defensi-e, mechanistic mode of operation.
Cost vs. #enefits of $reventative Care
#n a highl" mobile population, each pro-ider organi,ation )no/s that their members /ill
li)el" mo-e on to other groups. This /ould dri-e each them to ignore the longer term benefits of
pre-enti-e health care, under the assumption that the" /ould be funding cost reductions for their
competitors as members mo-e a/a".
The Transactional %allacy
(an" ha-e critici,ed the notion of the gross national product as a measure of the
goodness of an econom". The C<P is calculated b" summing up the transactions of an econom",
/ith no .udgement of the 1ualit" of the transactions. #n-estments in public education and
in-estments in enlarging our prisons are both contributors to the C<P, ho/e-er, their long term
effect on societ" is radicall" different. #ncreased tobacco sales and increased lung cancer
treatment both add to the C<P. #f someone /ere loo)ing at these statistics /ithout understanding
the long term situation, it could lead to the conclusion that /e should sell more cigarettes and
treat more lung cancer in order to impro-e the C<P.
! root cause of this fallac" is the notion of the transaction. !n interaction is e-aluated at
a 0snapshot0 in time, and a monetar" -alue is attached to it. Transactions are aggregated
according to predefined categories. Enderl"ing the concept of the transaction is the assumption
of linearit"3 that the /hole is e1ual to the sum of the parts. Transactions /or) /ell for ban)
accounts. !n account is precisel" the sum of the transactions for a month, accounting for mone"
is a strictl" linear process. ealth care, ho/e-er, is not a linear s"stem. The health of a patient is
not the sum of the transactions incurred b" the patient o-er the last month. The interactions of
people, diseases, public polic", global tra-el, and epidemics are all highl" non?linear.
edia "riven edicine
The impact of communications media is gro/ing rapidl". #t is in the best interest of drug
manufacturers, for e+ample, to use the media to create demand for their products.
(learly the problem of H(" .Hepatitis ( "irus/ will re-uire a responsible partnership of public and
private organisations 2nfortunately, the Schering (orporation has already breached the public trust
with a 2S newspaper campaign that appears designed more to creating H(" hysteria than public
understanding In the advertisement, the company gives the impression that anyone who has had their
ears$$or any other body part$$pierced, has a tattoo, or who has "shared a ra#or, toothbrush, or any
item that could carry blood", is at risk for H(" infection and should be tested "&o put it bluntly", the
company warns that "every living, breathing human being can get hepatitis ($$even you" 3ast year,
Schering sold 4567 million worth of the H(" treatment interferon a$8b in the 2S% alone, which
helped make interferon its second biggest product worldwide &he advertisement has infuriated 2S
public health officials, as well it might If we are to make progress against this perplexing epidemic,
careful and disinterested voices must prevail
ntropic vs. &orphic Changes
#magine a species /hich had four fingers, but no thumb. #f that species e-ol-ed a thumb
/hich gre/ out of the bac) of the hand, the thumb /ould ta)e energ" from the animal, and "ield
-er" little benefit. o/e-er, if the same thumb emerges opposite the e+isting four fingers, it
/ould allo/ the animal to grasp things better, and it /ould impro-e the species5 sur-i-al.
The same thumb, in t/o different conte+ts, had entirel" different results. The thumb on
the bac) of the hand could be called entropic, /hile the opposing thumb could be called
morphic.
$hanges to our s"stems can be -ie/ed from this perspecti-e as /ell. !cti-ities /hich
suc) energ" out of the organi,ation can be said to be entropic @ the "ear &%%% bug, for e+ample.
!cti-ities /hich increase the adaptabilit" of the organi,ation can be said to be morphic @ the
increase in business en.o"ed b" Dell and $isco through their /eb?based ordering s"stems, for
e+ample.
Chaos and 'ractals in Health Care
#magine an ancient Cree) using the Doman numeral s"stem. is numbering s"stem
/or)ed -er" /ell for counting a limited number of sheep @ one sheep, one scratch, t/o sheep,
t/o scratches, etc. e might e-en be able to calculate the area of a field b" counting off paces
and appl"ing some geometrical anal"sis. E-entuall", ho/e-er, the concreteness of his numbering
s"stem /ould cause him to falter as problems e+plode /ith comple+it". Toda", e-en a child /ith
limited )no/ledge of the decimal s"stem could perform calculations /hich /ould astound the
ancient Doman numeral user. Bet, to tr" to e+plain the -alue of our modern s"stem /ould be a
confounding e+perience 806h" do "ou need the concept of ,ero7 6e can all see that there is
nothing there.09
#n man" /a"s, health care informatics toda" is in the same situation as the ancient
Cree)s. 6e ha-e a s"stem /hich /or)s /ell for some -er" limited operations, "et e-er"/here
/e turn, /e find e+ploding comple+it". 6e find that fi+ing one problem creates a /ealth of ne/
ones.
6e need to loo) to/ards ne/, more abstract models /ith /hich /e can address a broader
set of problems /ith a more concise set of concepts. Fi)e tr"ing to teach the ancient Cree)
decimal arithmetic, this is not an eas" transformation to achie-e. #t /ill first appear as abstract
and /ea) compared to the current model. #deall", ho/e-er, after the ne/ concepts ta)e hold,
things /hich once appeared comple+ can be made simple again.
$omputers and communications and the /eb ha-e radicall" affected our notion of scale
and communit" beha-ior. 2ld notions of ho/ /e address problems, and the speed at /hich the"
appear and disappear ha-e to be discarded.
! successful model /ill deal /ith this comple+it" b" addressing the intrinsics of the
problem, then letting technolog" scale the solution to the broadest range of participants. This
re1uires a ne/ set of tools for dealing /ith comple+it".
$haos theor" and fractals are one such mechanism. Dather than -ie/ing s"stems as fi+ed
entities /hich operate /ith mechanical precision and predictabilit", the" are -ie/ed as comple+
and adapti-e. 2ne dimension to this problem is the notion of scale in s"stems.
Scale as a (imension
Time and space are familiar dimensions /hen -ie/ing ob.ects. 6e can be at a precise
point on the earth @ a beach, for e+ample @ at a precise time @ perhaps at sunset. 6ith these
coordinates, /e can precisel" locate oursel-es.
But there is a missing dimension to this @ that of scale. The same beach ma" be -ie/ed
from the perspecti-e of a satellite, a lo/ fl"ing airplane, a standing adult, a cra/ling infant, an
ant or the -ie/ through an electron microscope. #n the time and space dimension, it is the same
beach. But the scale dimension re-eals entirel" different -ie/s, from the e+tremel" coarse -ie/
of the satellite to the e+tremel" fine -ie/ of the electron microscope.
This perspecti-e is an e+ample of fractal dimension3
"Fractal dimension becomes a way of measuring -ualities that otherwise have no clear definition: the
degree of roughness or brokenness or irregularity in an ob9ect % twisting coastline, for example,
despite its immeasurability in terms of length, nevertheless has a certain characteristic degree of
roughness +andelbrot specified ways of calculating the fractal dimension of real ob9ects, given some
techni-ue of constructing a shape or give the irregular patterns he had studied in nature &he claim
was that the degree of irregularity remains constant over different scales Surprisingly often, the claim
turns out to be true )ver and over again, the world displays a regular irregularity:In the mind;s eye,
a fractal is a way of seeing infinity"
This sense of regular irregularity displa"s itself in the human bod". 4or e+ample, e-er"
cell in the bod" is no more than three or four cells a/a" from a blood -essel. Bet the -essels and
blood ta)e up little space, no more than fi-e percent of the bod". Ci-en the bod"5s continuous
change of gro/th and repair, it is impossible to construct a fi+ed map of blood -essels in the
bod"A it is a constantl" adapting and changing 0landscape.0 #t /ould be impossible to specif" the
precise la"out of blood -essels at the D<! le-elA it must instead specif" a repeating process of
rules of bifurcation and de-elopment.
The fractal model allo/s us to add a dimension of scale to our thin)ing. 6e can describe
and deal /ith regular irregularities, and /e can allo/ the same ob.ect to be described in -er"
coarse or -er" fine terms. (ost importantl", /e can )no/ /hether /e are spea)ing at a specific
le-el of precision, or dealing /ith the scale independent regularities.
$onsider the term 0episode.0 !n" t/o English spea)ers can use the /ord easil", and
probabl" communicate /ith shared meaning. The V! spent a decade, ho/e-er, attempting to
come to a nationall" accepted meaning of 0episode0 in the health care setting. 2ther
organi,ations are also in the throes of similar definitional processes, and also finding that the"
re1uire "ears of collaboration. The" are see)ing e-er?finer definitions of the term in their
particular speech communities. !t a coarse le-el, the" /ould all agree, "et as the precision
becomes finer, the" ha-e progressi-el" more difficulties coming into agreement.
2ne could -ie/ this process as acting li)e D<! /hich is tr"ing to e+plicitl" control the
precise la"out of blood -essels in the bod". E-er" gro/th phase /ould ha-e to be e+plicitl"
coded, and e-er" possible in.ur" /ould be classified so that a specific response could be pre?
programmed. Bod" gro/th /hich did not conform to the predefined blood -essel map /ould not
thri-e. Bodies /hich suffered in.uries not anticipated /ould simpl" not sur-i-e.
The bod" does not function this /a"A blood -essels are an e+ample of an adapti-e s"stem.
Their gro/th adapts to their en-ironment. #f Bill, an adolescent bo", shoots up 1& inches in one
"ear, his blood -essels adapt accordingl". #f he suffers a gash or bro)en bones pla"ing football,
his blood -essels adapt to the in.ur" to repair an" damage.
!t the coarser le-els of description, the process is 0sur-i-al of the species,0 and 0sur-i-al
of Bill.0 !t finer le-els of description, the problem becomes 0repair a bone,0 and 0)eep gro/ing
finer bifurcations of blood -essels until the" are /ithin : or ; cells of e-er" other cell.0 This
control is not based on a single, top do/n mechanism based on preprogrammed methods. #t is
adapti-e and fle+ible3
"*ith all such control phenomena, a critical issue is robustness: how well can a system withstand
small 9olts:% locking$in to a single mode can be enslavement, preventing a system from adapting to
change )rganisms must respond to circumstances that vary rapidly and unpredictably< no heartbeat
or respiratory rhythm can be locked into the strict periodicities of the simplest physical model< and the
same is true of the subtler rhythms of the rest of the body"
The human bod" is a s"stem far from stabilit" or e1uilibriumA a li-ing organism /hich
achie-es e1uilibrium is dead. #t is constantl" adapting and changing in response to internal and
e+ternal influences.
(an" of our s"stem design methods are based on the implied assumption that the s"stem
is stable. The" assume that a predetermined set of re1uirements can be reduced to a fi+ed set of
specifications, /hich in turn can be tested, integrated, optimi,ed, and deli-ered to the
organi,ation that originall" named the re1uirements. #n some cases, such as sending a man to the
moon, this /or)s remar)abl" /ell. The moon5s orbit didn5t change, the la/s of ph"sics sta"ed
constant, and /e did not change these facts b" performing the flights.
2ur health care s"stem and human ph"siolog", ho/e-er, are far from e1uilibrium. The"
are constantl" changing and adapting to internal and e+ternal forces. The" beha-e as chaotic
s"stems, according to rules /hich are far remo-ed from the simple mechanistic ones applied to
getting to the moon. De1uirements created = "ears ago for a health information s"stem /ould
surel" not be current toda"A .ust as re1uirements named toda" /ould not be applicable = "ears
from no/. #nstalling a computer s"stem /ill change the organi,ation, creating entirel" ne/
information needs in the future.
The turmoil /e see in health care is not .ust 0going through a phase0 after /hich things
/ill settle do/n. #t is indicati-e of the fundamentall" chaotic nature of the s"stem /e are dealing
/ith. !d-ances in computers and communications /ill create additional turmoil in the "ears
ahead. #t is not a linear, predictable s"stem, in /hich the /hole is e1ual to the sum of its parts.
The /hole of our health is far greater than the sum of its parts, and deals /ith an incredibl"
comple+ /eb of causalities. #f our health care s"stem e-er reaches e1uilibrium, it /ould be dead.
$haos theor" offers much to this discussion. #t allo/s us to thin) about 0regular
irregularities0 across scale, not .ust time and space. 4rom the perspecti-e of a senator on the ill,
health care represents trillions of dollars o-er decades. 4or an emergenc" room ph"sician, it
in-ol-es ma)ing decisions in seconds @ during /hich the lac) of a J&= item can ma)e a life or
death difference. 4or an indi-idual patient, it in-ol-es some of the most intimate trust and
pri-ac" issues in all of our societ". The health care s"stem couples business, go-ernment,
medical science, information, mass media, culture, religion, famil", and personal belief s"stems
in a unpredictable biological, political, and en-ironmental /orlds. #n addition to all these
comple+ities, the human bod" itself is an unpredictable, chaotic realm.
4rom this perspecti-e, it is impossible to ma)e simple linear causal statements. !s
comforting as it might be to someone suffering from toda"5s lust for precision, /e cannot sa"
that a specific case of cancer /as caused 1;K b" smo)ing, 19K b" smog, =K b" nuclear fallout,
&K b" li-ing near po/er lines, 1HK b" lifest"le choices, and so on. 6e can, ho/e-er, at some
degree of coarseness, tal) about these contributions to the cancer process.
#f /e ta)e a fractal -ie/ of health care, /e need to loo) at the 0regular irregularities0 in
the s"stem. 6hat are the properties in the s"stem /hich are constant across scale7 6hat patterns
can /e recogni,e as /e ,oom in from the satellite -ie/ of the beach, to si+ foot high le-el, to the
microscopic7 6hat are the properties /hich persist regardless of the coarseness or fineness of
our perspecti-e7
6e /ill call these scale independent properties the intrinsics of the s"stem. These
properties are critical to the successful understanding of a chaotic s"stem.
)exicons and *omenclatures
The past decades ha-e seen an e+plosion in the number and di-ersit" of standards
organi,ations, attempting to impose order in the communication of medical information. Some
commercial le+icon ser-ices ha-e o-er 1 million terms.
These le+icons are not complete @ there is still much /or) to be done. (edical
informatics is approaching a comple+it" crisis, in /hich the attempts to create a static, pre?
defined le+icon for the process of health care is unable to )eep up /ith the immediate needs of
the industr". Dather than creating more standards /ith more interfaces bet/een the -arious
-ersions and approaches, an alternati-e /ould be to focus on means of communicating /ith
fewer standards /hich are more adapti-e to their communities of interest. B" focusing on the
concept of shared meaning instead of static le+icons, and e+ploiting the po/er of connecti-it" of
the /orld /ide /eb, this research applies principles of comple+it" theor" and linguistics to
create an e-olutionar" en-ironment in /hich participants /ill naturall" con-erge on a common
understanding of the discipline.
This approach is counterintuiti-e to most approaches in medicine. o/e-er, the process is -er"
much e-ident in natural languages. 4or e+ample, until 19H9, <icaragua had no schools for the
deafA deaf children /ere ne-er part of a single communit". !fter the formation of the first school,
the children spontaneousl" created their o/n language, albeit a primiti-e pidgin. The ne+t
generation of students pic)ed up the language and created a full?fledged sign language. The
language 8#dioma de Signos <icagLnse9 spontaneousl" standardi,ed itselfA there /ere no
standards committees, policies, or top do/n hierarchical controls imposed on the students.
The lesson is that creating a communit" /ith a need to communicate /ill create language
necessar" to share meaning. This language is an e-olutionar", adapti-e process that necessaril"
begins simpl". $omple+it" emerges as a b"?product of the e-olution, not as part of the initial
conditions 8as is the case /ith our current attempts -ia standards9
The problem to the current 0failure to communicate0 in health care is principall" due to the lac)
of connecti-it" and communit" in the health care setting. #t is not due to the lac) of standards,
nomenclatures, and interface paraphernalia, of /hich there is a gro/ing e+cess.
Sha)espeare /as able to create some of the greatest and most enduring /or)s of English
literature /ith onl" &9,%%% different /ords. The current e+plosion of medical le+icons must be
met /ith less, rather than greater specificit" and atomicit". The principles of chaordic design can
deepl" influence the flo/ of language and the /a"s /e create shared meaning in our
communications.
Some %ntrinsics o! Health Care
Transformations
The notion of transformation is a )e" intrinsic to understanding the health care s"stem.
The senator /ants to transform the health care s"stem to pro-ide lo/er cost, higher 1ualit", more
e1uitable health care to !mericans. The mother of a ne/born /ants to transform her bab" into a
health" adult. T/el-e step programs /ant to transform their members a/a" from their addicti-e
beha-ior. Pharmaceutical companies /ant to introduce drugs /hich /ill assist the bod" in
transforming from a disease state to a health" state.
! transformation ma" be defined as a flow of purposeful activities within an interpretive
community. 2ne /a" to understand this concept is to contrast it /ith its cogniti-e predecessor,
the transaction.
#magine getting a flopp" dis), ha-ing t/o files of about &=M b"tes. 2ne file,
B#FF#<C.D2$, contains a "ear5s /orth of credit card billing information. The other,
$2<ST#TET#2<.D2$ contains a cop" of the ES constitution. 4rom one perspecti-e, the" are
.ust t/o collections of b"tes of information.
Foo)ing at B#FF#<C.D2$, /e find orderl", precise information. !fter the beginning
balance, /e ha-e precisel" defined transactions /hich can be added together to arri-e at the
closing balance. There is little or no uncertaint" in the statement. #f /e do not understand a
charge, /e could call the compan" for a cop" of the original receipt and resol-e an" ambiguities.
Each transaction has a date attached to it, and is -alued b" a single dollar amount. Simple rules
define ho/ /e aggregate the transactions do/n the bottom line. The credit card billing statement
is an e+ample of transactional processing.
Foo)ing at $2<ST#TET#2<.D2$, ho/e-er, /e find much less order and precision. #t is
a rather 0coarse0 document, loo)ing at !merican go-ernment from a -er" high, and sometimes
ambiguous le-el. 6e cannot add up all the articles to get to the 0bottom line.0 There is much
uncertaint" in the constitution, so much that the courts must continuall" interpret it. There are no
simple rules b" /hich one can predict the meaning of the constitution. 6e cannot assess /hat
percentage of Ceneral (otors re-enues this 1uarter /ere due to the constitution. #t is impossible
to do a costIbenefit anal"sis. 6e ha-e no /a" of )no/ing if /e ha-e the 0optimum0 constitution,
nor can /e do outcomes assessments on it. Bet, /ithin those communities /ho chose to use the
$onstitution, it is probabl" the most po/erful &= )ilob"tes of information )no/n. This is an
e+ample of transformational information.
#magine a spectrum of information t"pes, /ith transactional information 8the billing
statement9 at one e+treme and transformational information 8the $onstitution9 at the other.
6here does health care information fit7 !t first glance, billing information /ould appear to be
the perfect candidate for pure transactional information. Billing transactions can be added up
from a beginning balance, ha-e precise amounts, and ha-e specific dates. But suppose that the"
are related to a specific episode of care. 6e ha-e no/ launched bac) into the fractal /orld of
transformational meaning. 6hose definition of episode are /e using7 6hose interpretation of
the definition7 Pa"ing for an umbrella /ith a credit card and pa"ing for an episode of care
through a third part" pa"or are t/o radicall" different concepts.
ealth information e+ists far into the transformational end of this spectrum, possibl" past
the $onstitution. There is much ambiguit" and imprecision in health care, and there is no
0bottom line.0 !n indi-idual5s health does not start the "ear /ith a 0beginning balance0 against
/hich health transactions can be added to arri-e at a 0closing balance0 at the end of the "ear. 2ne
cannot add up a tonsillectom", appendectom" and an annual ph"sical to calculate >ohn Smith5s
total health last "ear.
<or does health care fit into the traditional mar)et model. ealth is not a scarce resource,
sub.ect to the la/s of suppl" and demand. 2ne person5s becoming healthier does not deplete
some central reser-oir of health. #f an"thing, one person5s increase in health increases e-er"one
else5s. 6hen fa+ machines /ere introduced, each additional fa+ machine increased the -alue of
each e+isting fa+ machine. The increase in the -alue of the fa+ machines attracted more bu"ers,
/hich in turn fueled the -alue of e+isting machines. The increasing -alue of the /hole ma)es
each of the components more -aluable.
ealth can /or) this /a". The healthier the population, the greater the health of each
indi-idual. !s indi-iduals increase in health, the healthier the population. !s >onas Sal) said in
an inter-ie/ /ith Bill (o"ers, 0#t is possible to create an epidemic of health.0 The #nternet,
coupled /ith inno-ati-e /a"s of dealing /ith the transformational -alue of health care, can ser-e
as the foundation for this epidemic.
The Transactional 'allacy
! traditional method for dealing /ith comple+ s"stems it to reduce the s"stem into
components, /hose beha-ior can be measured in terms of transactions. These transactions are
aggregated and 0rolled up0 hierarchicall" from the lo/est le-el to the highest. Each le-el is
anchored at some le-el on the continuum of scaleA the top of one hierarch" is the bottom of
another, and onl" the aggregates of transactions get passed up/ard.
$onsider the transactional -ie/ of the national econom". #magine a teenage girl bu"ing a
pac)age of cigarettes. The transaction is recorded at the store, /hich then rolls up to the
compan" o/ning the store, to the manufacturer, to the tobacco industr", and finall", the gross
domestic product 8CDP9. Similarl", imagine that girl :% "ears later, a /oman /ho has de-eloped
lung cancer. The health care ser-ices supplies she consumes rolls up to contribute to the CDP.
4rom the transactional -ie/point, someone tr"ing to optimi,e the CDP /ould come to
the conclusion that /e should sell more cigarettes and treat more lung cancer @ as sho/n b" the
bottom lin. 4rom the indi-idual5s transformational perspecti-e, ho/e-er, things are different.
This e+ample illustrates the fundamental clash bet/een the transactional and
transformational perspecti-es, /hich can be called the transactional fallacy The beha-ior of the
s"stem as measured b" the transactional rollup procedures is radicall" different than the goals of
the s"stem from the perspecti-e of the ob.ect being acted upon.
Some ma" ta)e e+ception /ith this anal"sis, sa"ing that the transactional model did not
include e+ternalities, such as 0utilit" of smo)ing.0 01ualit" of life,0 or do a present -alue anal"sis
to account for the inter-al bet/een smo)ing and onset of cancer. The" /ould tin)er /ith the
model, to precisel" factor in all of the issues so that the manager of the CDP /ould ma)e
0rational0 decisions on the matter.
4rom the transformational -ie/, ho/e-er, this approach is li)e tr"ing to get out of a hole
b" digging it deeper. The basic problem is the inappropriate le-el of precision, and the
aggregationIhierarchical model /hich attempts to optimi,e a s"stem based on a single -ariable.
The smo)ingIlung cancer issue is a -er" comple+, multidimensional problem.
The bul) of /hat happens in health care is transformational, "et the bul) of our
information processing s"stems is transactional. 4e/ hospitals ha-e ignored the transactional
concept of charging for e-er" bo+ of tissue issued to a patient, "et fe/ ha-e stepped up to the
transformational challenge of using information s"stems to reduce pre-entable ad-erse drug
reactions.
Thus, /e ha-e a health care s"stem /hich is largel" dri-en b" the transactional fallac"
and per-erse incenti-es.
Some design issues !or a Chaordic model !or health care
6hat /ould a $haordic model for health care loo) li)e7 Some of Dee oc)s ideas are3
5 It must be equitably owned by all participants . ,o member should have intrinsic preferential
position %ll advantage must result from individual ability and initiative
8 Power and function must be distributive to the maximum degree . ,o function should be
performed by any part of the whole which could reasonably be done by any more peripheral part,
and no power vested in any part which might reasonably be exercised by any lesser part
= Governance must be distributive . ,o individual, institution, and no combination of either or both,
should be able to dominate deliberations or control decisions
6 It must be infinitely malleable yet extremely durable . It should be capable of constant, self$
generated, modification of form or function without sacrificing its essential nature or embodied
principle
> It must embrace diversity and change . It must attract people and institutions comfortable with
such conditions and provide an environment in which they could flourish

Possi"le Principles !or a Chaordic Health Care approach
"2nless we can define a purpose for this organi#ation that we can all believe in, we might as well
go home" &hat!s "purpose" as in, "*e the people of the 2nited States of %merica, in order to form a
more perfect union " &he purpose has to be an authentic statement of what the organi#ation is
about, not some platitude cooked up by a consultant
These are some candidate principles for a $haordic model of health care3
1. ealth care is a collaborati-e process bet/een indi-iduals and their health care
communit".
&. #ndi-iduals control access to their health care information outside the institution in /hich
the information /as generated. The" ha-e the right to )no/ /ho has accessed /hat
information, and /hen.
:. The institutions and health care pro-iders for the indi-idual are obligated to present their
information to the indi-idual5s electronic repositor".
;. ealth is a sub.ecti-e matter, based on the indi-idual5s perception of their en-ironment
and their functionalit".
=. #ndi-iduals are free to associate their genetic information to their ancestors or
descendents as the" see fit.
'. The same d"namics /hich create an epidemic of disease can be applied to create an
epidemic of health.
H. Do no harm. Nuote !DE stuff
8. (a+imi,e functional capabilities of the population
9. De-elop and maintain the trust of the indi-idual
1%. ealth $are as a collaborati-e process
11. !ssociated Deli-er" <et/or)

Bi"liography
Some selected boo)s on the science of comple+it" are3
1. $asti, >ohn, Complexification" !xplaining a Paradoxical World through the #cience
of #urprise" arper Perennial, 199=
&. Cleic), >ame, $haos, $haos 3 (a)ing a <e/ Science, Penguin, 1988
Cood/in, Brian o/ the Feopard $hanged #ts Spots 3
:. >ohnson, Ceorge, Fire in the $ind % #cience" Faith" and the #earch for rder
;. Mauffman, Stuart At &ome in the 'niverse % The #earch for (a)s of #elf*
rgani+ation and Complexit,
=. Mell", Me-in ut of Control % The -e) .iolog, of $achines" #ocial #,stems and the
!conomic World
'. (c(aster, (ichael, The Intelligence Advantage" rgani+ing for Complexit,,
Butter/orth?eineman, 199;
H. Pagels, ein,, The Dreams of /eason" the Computer and the /ise of the #ciences of
Complexit," Bantam, 1988
8. Dothschild, (ichael, .ionomics, !conom, as !cos,stem
9. 6aldrop, (. (itchell Complexit, % The !merging #cience at the !dge of rder and
ChaosA