Upload
others
View
5
Download
0
Embed Size (px)
Citation preview
1
A Complex-Network Perspective on Alexander’s Wholeness
Bin Jiang
Faculty of Engineering and Sustainable Development, Division of GIScience
University of Gävle, SE-801 76 Gävle, Sweden Email: [email protected]
(Draft: February 2016, Revision: April, July 2016)
“Nature, of course, has its own geometry. But this is not Euclid’s or Descartes’ geometry. Rather, this geometry follows the rules, constraints, and contingent conditions that are, inevitably, encountered in the real world.”
Christopher Alexander et al. (2012)
Abstract The wholeness, conceived and developed by Christopher Alexander, is what exists to some degree or other in space and matter, and can be described by precise mathematical language. However, it remains somehow mysterious and elusive, and therefore hard to grasp. This paper develops a complex network perspective on the wholeness to better understand the nature of order or beauty for sustainable design. I bring together a set of complexity-science subjects such as complex networks, fractal geometry, and in particular underlying scaling hierarchy derived by head/tail breaks – a classification scheme and a visualization tool for data with a heavy-tailed distribution, in order to make Alexander’s profound thoughts more accessible to design practitioners and complexity-science researchers. Through several case studies (some of which Alexander studied), I demonstrate that the complex-network perspective helps reduce the mystery of wholeness and brings new insights to Alexander’s thoughts on the concept of wholeness or objective beauty that exists in fine and deep structure. The complex-network perspective enables us to see things in their wholeness, and to better understand how the kind of structural beauty emerges from local actions guided by the 15 fundamental properties, and by differentiation and adaptation processes. The wholeness goes beyond current complex network theory towards design or creation of living structures. Keywords: Theory of centers, living geometry, Christopher Alexander, head/tail breaks, and beauty 1. Introduction Nature, or the real world, is governed by immense orderliness. The order in nature is essentially the same as that in what we build or make, and underlying order-creating processes of building or making of architecture and design are no less important than those of physics and biology. This is probably the single major statement made by Alexander (2002–2005) in his theory of centers, in which he addressed the fundamental phenomenon of order, the processes of creating order, and even a new cosmology – a new conception of how the physical universe is put together. In the theory of centers or living geometry (Alexander et al. 2012), the wholeness captures the meaning of order and is defined as a life-giving or living structure that appears to some degree in every part of space and matter; see Section 3 for an introduction to wholeness and wholeness-related terms. As the building blocks of wholeness, centers are identifiable coherent entities or sets that overlap and nest each other within a larger whole. Unlike the previous conception of wholeness focusing on the gestalt of things (Köhler 1947), the wholeness of Alexander (2002-2005) is not just about cognition and psychology, but something that exists in space and matter. Different from the wholeness in quantum physics mainly for understanding (Bohm 1980), Alexander’s wholeness aims not only to understand the phenomenon
2
of order, but also to create order in the built world or art. The wholeness is defined as a recursive structure. Based on this definition, Jiang (2015) developed a mathematical model of wholeness as a hierarchical graph with indices for measuring degrees of life or beauty for both individual centers and the whole. This model helps address not only why a design is beautiful, but also how much beauty it has. However, this previous study had some fundamental issues on the notions of centers and wholeness unaddressed. Specifically, what are the centers, how are they created, and how do they work together to contribute to the life of wholeness? In addition, the wholeness remains somehow mysterious, particularly within our current mechanistic worldview (Alexander 2002–2005). To address these fundamental issues, this paper develops a complex-network perspective on the wholeness. A complex network is a graph consisting of numerous nodes and links, with unique structures that differentiate it from its simple counterparts such as regular and random networks (Newman 2010). Simple networks have a simple structure. In a regular network, all nodes have a uniform degree of connectivity. In a random network, the degrees of connectivity only vary slightly from one node to another. As a consequence, a random network hardly contains any clusters, not to say overlapping or nested clusters. On the contrary, complex networks, such as small-world and scale-free networks (Watts and Strogatz 1998, Barabási and Albert 1999), tend to contain many overlapping and nested clusters that constitute a scaling hierarchy (Jiang and Ma 2015; see a working example in Section 2). The scaling hierarchy is a distinguishing feature of complex networks or complex systems in general. For example, a city is a complex system, and a set of cities is a complex system (Jacobs 1961, Alexander 1965, Salingaros 1998, Jiang 2015c), both having scaling hierarchy seen in many other biological, social, informational, and technological systems. This paper demonstrates that the wholeness bears the same scaling hierarchy as complex networks or complex systems in general. Relying on the complex network perspective, this paper aims to demonstrate that wholeness is not just in cognition and psychology, but something that exists in space and matter. It also aims to show that the concept of wholeness is important not just for understanding the phenomenon of order, but also for creating order with a high degree of wholeness through two major structure-preserving transformations: differentiation and adaptation. I argue that there are major differences between the whole as a vague term and wholeness as a recursive structure. The wholeness comprises recursively defined centers induced by itself, whereas the whole, as we commonly perceive, comprises pre-existing parts. The mantra that the whole is more than the sum of its parts should be more truly rephrased as the wholeness is more than the sum of its centers. This paper examines the notions of wholeness and centers from the perspective of complexity science. It also discusses two types of coherence respectively created by differentiation and adaptation processes, which are consistent with the spatial properties of heterogeneity and dependence for understanding the nature of geographic space. The remainder of this paper is structured as follows. Section 2 briefly introduces complex networks and scaling hierarchy using head/tail breaks – a classification scheme and a visualization tool for data with a heavy-tailed distribution (Jiang 2013a, Jiang 2015a). Section 3 compares related concepts, such as whole and parts versus wholeness and centers, and discusses the theory of centers using two examples of a cow and an IKEA desk. Section 4 presents three case studies to show how wholeness emerges from space and how it can be generated through the two major structure-preserving transformations of differentiation and adaptation. Section 5 further discusses implications of the complex-network perspective and wholeness. Finally, Section 6 draws a conclusion and points to future work. 2. Complex networks and the underlying scaling hierarchy Small-world and scale-free networks are two typical examples of complex networks, which fundamentally differ from their regular and random counterparts. A small-world network is a middle status between the regular and random networks, so it has some nice properties of its regular and
random clusterinlength (Wtheir degnodes thhave thehighest that invcompon Complex2004). Asub-struscaling hwe adopcomplexcommunfar morecommunof sizes organs tMore gehuman bas cohernotion oharmony3 for mo
(Norelatio
commu
The undin generbiologiccomplexthrough Tree. Hecomplex
counterpartsng coefficienWatts and Sgree of connehan well-cone highest degand the loweolves a largents are thos
x networks A cluster has cture that adhierarchy of pt the Karatx networks. nities of differe small commnity of size 286, 5, 2, 1, an
that functionenerally, a cobody growinrent parts areof adaptationy or coherenore details) a
Figure 1: (Coote: Panel a shonships of theunities than la
derlying scalral (Simon 19cal, technologx network wnaturally ev
e used the tex networks w
s. These propnt, and globaStrogatz 1998ectivity demnected ones
gree of conneest. Complex
ge number ose built from
tend to conmany inside
dapts to its cof far more smte Club netwThis networrent sizes: 28
munities than 8 consists of nd 1. The scan as independomplex netwng or unfoldine integrated n or fit, whice at individs a building
olor online) Thhows the sociae 14 communiarge ones. The
ing hierarchy962). It appegical, and in
with many redvolved cities,erm semi-latwas not know
perties are loal efficiency8). Scale-free
monstrates a p(Barabási an
ectivity, manx networks af nodes andthe basic un
ntain many ce links and a ontext. Clust
mall communwork (Zacharrk, consisting8, 15, 10, 6, 5
large ones. Ta community
aling hierarchdent coherenork is not asng from a seinto their coch recurs be
dual local scablock of the
he Karate Clual network, coties of sizes 2
ese communiti(Jiang 2
y is an impoars in a varie
nformationaldundant and, as Alexandttice to referwn then. In
3
ocal efficienof the randoe networks a
power-law dind Albert 19ny nodes havare so calledd componentnits.
components,a few outsideters within anities than lary 1977), wig of 34 node5, 5, 3, 2, 2, 2There are alsy of size 15, why is formed nt parts, but ssembled fromeed or an emontexts as a wetween adjacales. Clusterswholeness.
ub network bonsisting of 3428, 15, 10, 6, 5ies were detec2013a, Jiang 2
ortant properety of naturaones. This s
d overlappingder (1965) nir to the scalithis respect,
cy of regulaom networksare a specialistribution, in99). In other
ve the lowest precisely be
ts. The node
, termed come links, so coa complex nearge ones (Jiidely studiedes and 78 lin2, 1, 1, 1 andso many nestwhich is furthfrom the clustill fit into
m mechanicambryo. From
whole (Alexcent elements resemble th
roken into mu4 nodes and 75, 5, 3, 2, 2, 2,cted by an algo2015a).)
rty of complal and societascaling hierag links. Thisicely articulaing hierarchy, Alexander
r networks cs measured bl type of comndicating far r words, verydegree, and
ecause of thees are the ba
mmunities onstitutes a co
etwork nest eiang and Mad in the litenks, can be b
1 (Figure 1)ted relationshher broken dosters, which the contextal parts, but the design p
xander 1964)ts and systemhe concept of
utually nested8 links. Pane 1, 1, 1 and 1 orithm inspire
ex networksal complex syarchy is not s insight wasated in the cly simply becwas far ahe
characterizedby short avermplex networ more less-cy few nodes,some in bet
eir complex asic units, w
or clusters (Noherent sub-each other, fa 2015). To ierature of sobroken down
). Apparently hips. For examown into comare similar t
t of the humis more like
point of view). This introdms, and helpf centers (see
d communitiesel b depicts newith far more
ed by head/tail
s or complexystems such tree-like buts originally lassic A Citycause the co
ead of the tim
d by high rage path orks, and onnected , or hubs, ween the structure
while the
Newman -whole or forming a illustrate, ocial and n into 14 there are
mple, the mmunities to human
man body. a tree or
w, clusters duces the ps create e Section
s ested e small l breaks
x systems as social,
t rather a observed
y is Not a oncept of me when
4
complex networks were understood. More importantly, his thoughts are not just limited to understanding surrounding things, just as what current complex network theory does, but aim for creating things with living structure (i.e., buildings, communities, cities or artifacts). The scaling hierarchy can be further seen from the relevance or importance of individual nodes within a complex network. Through Google’s PageRank (Page and Brin 1998), nodes’ relevance or importance can be computed, as there are far more less-important nodes than more-important ones. PageRank is recursively defined and resembles the wholeness as a recursive structure. This will be further discussed in the following section. 3. The wholeness and the theory of centers The general idea of wholeness, or seeing things holistically, can be traced back to the Chinese philosopher Chuang Tzu (360 BC), who saw the structure of a cow as a complex whole, in which some parts were more connected (or coherent) than others. The butcher who understands the cow’s structure always cuts the meat from the soft spots and the crevices of the meat, so makes the meat fall apart according to its own structure. The butcher therefore can keep his knife sharp for a hundred years. In the 20th century, wholeness was extensively discussed by many writers prominent in Gestalt psychology (Köhler 1947), quantum physics (Bohm 1980), and many other sciences such as biology, neurophysiology, medicine, cosmology and ecology. However, none of these writers prior to Alexander (2002–2005) showed how to represent or formulate wholeness in precise mathematical language. According to the Merriam-Webster dictionary, whole is something complete, without any missing parts. On the other hand, wholeness is the quality of something considered as a whole. However, both whole and wholeness have deeper meanings and implications in the theory of centers (Alexander 2002–2005). In particular, the notion of wholeness is so subtle and profound that Alexander (1979) previously referred to it as ‘the quality without a name’. He struggled with different names, such as alive, comfortable, exact, egoless and eternal, but none of these captured the true meaning of the quality. In this paper, as in Alexander (2002-2005), the three terms wholeness, life, and beauty are interchangeably used when appropriate to indicate order or coherence. Things with a high degree of wholeness are called living structure.
Table 1: Comparison of whole and wholeness using two examples of a cow and an IKEA desk
Whole (noun + adjective) Wholeness (structure + measure) on the surface deeper below the surfaceconsists of parts (see below subsection) consists of centers (see below subsection) easy to see, the whole of a cow is the cow itself
hard to see as a recursive structure, the wholeness of a cow
hard to sense, one thing is more whole than another
hard to sense as the degree of coherence somehow like temperature
easy to sense, a cow is more whole than a desk
easy to sense, a cow has a higher wholeness than a desk
a whole assembled from parts like a desk a desk has low wholeness or low coherence a whole unfolded from an embryo like a cow a cow has high wholeness or high coherence Parts Centerspre‐existing in the whole induced by the wholenesson the surface deeper below the surfaceeasy to see hard to seemechanical organicnon‐recursive recursivesimple complex
Wholeness is defined as a recursive life-giving structure that exists in space and matter, and it can be described by precise mathematical language (Alexander 2002-2005). Although whole and wholeness seem different, they are closely related, and sometimes refer to the same thing. The whole is on the surface and is referred to informally, while wholeness is below the surface and is referred to formally (Table 1). The term whole can be both a noun and an adjective. For example, a cow is a whole, and a
cow is mstructurerelativelthe cow atoms, mthat oneassemblhigher dof whichto mechasimple. because the partsand the o
(Not
To furthwhole, othe left wholeneobjects opicture oprefer, wpick oneour feelexists bphenomcenters, than theexperimcannot pwholenewholene
more whole te, and as a mly coherent sitself. The w
molecules, cee thing has aed from part
degree of whh the whole anical piecesCenters are they exist d
s and centersorganic, whi
Figute: The left is
Levels ofStrong ceThick bouAlternatiPositive s
her elaborateor has a highone has a h
ess can be cor their imagof your own
which is likele that reflectslings, is sharboth physicaenon of whoboundaries,
e right, or mment is some
provide a press of two coess (Jiang 2
than a desk. measure for spatial set, bewholeness asells, tissues, a higher degts are often leoleness thanand wholenes that are easmainly creat
deeper belows show two ich underlies
ure 2: Two carmore whole th
Table 2:
f scaleenters undaries ng repetition space
e on the whoher degree ofhigher degreecaptured throges are put sin deepest or ly to be idioss your own inred among p
ally in the woleness comand local sy
more of theehow like rerecise measuomplex thing2015b), rely
On the otherdegree of wecause it apps a recursiveand organs f
gree of wholess whole th
n a desk. We ess respectivsy to see becated by the w
w the surface,different woAlexander’s
rpets with diffhan the right,
: The 15 fund
Good shapLocal symmDeep interContrastGradients
oleness, let uf wholeness. e, or is moreough the miride-by-side itruest self assyncratic, accnner self. Thpeople and world and p
mprises the 1ymmetries (se mirror-of-telying on huurement of tgs such as a cing on a h
5
r hand, wholwholeness or
pears on the e structure is forming a scleness, or is
han things grocan also com
vely consist. ause they ap
wholeness an, and are recu
orld views: ths radical thou
fferent degree or the left has
damental pro
pemetriesrlock and amb
us examine wBoth the car
e whole, tharror-of-the-sein front, and s a whole. Tcounting for
his part of huacross cultupsychologica5 fundamenee Table 2). the-self quauman sense temperature.cow and a tr
hierarchical
eness has twcoherence. Asurface. Forhard to see,aling hierarc
s more wholown from emmpare the twParts usuallypear on the s
nd refer to orursive and cohe mechanistught.
of wholenesss a higher deg
operties of w
RouEcho
biguity The SimNot
which of therpets possess
an the one onelf experimeyou are aske
The experimer 10 percent ouman experieures (Alexanally inside ttal propertieThe left carplities than tto compare
. Equally, itree. In this regraph for r
wo different mA whole is er example, thsince it lies
chy. It is usuae than anoth
mbryos. For ewo related termy pre-exist insurface and arganic piecesomplex. Thetic, to which
(Alexander 1ree of wholen
holeness
ghnessoesvoidplicity and innseparateness
e two carpets a high degrn the right. Tent (Alexanded to choose ent is not to of our feelingence, accountder 2002–20the human
es such as lepet has morethe right. Th
two temperis usually h
egard, the maepresenting
meanings: a reasy to see –he whole of
s deeper consually difficulther. Howeveexample, a crms parts andn the whole are non-recurs that are hae differences h we are accu
993) ness than the r
ner calm s
ts (Figure 2)ree of wholeThis phenomder 2002-200
one that repchoose one
gs. Instead, yting for 90 p005). The wself. Physic
evels of scale of the 15 phe mirror-ofratures. Our hard to comathematical the wholen
recursive – being a a cow is
sisting of t to sense er, things cow has a d centers, and refer rsive and
ard to see between
ustomed,
ight.)
) is more eness, but menon of 05). Two presents a
that you you must
percent of wholeness cally, the e, strong
properties f-the-self
feelings mpare the
model of ness, can
accurateperspect 4. The wThis secof its cepaper wicase stuwhole ostructure 4.1 A paThe whopresentethe sameplaced ois not hcenter oremains least 20 corners, only thewholenealso sup(Figure as indica
(Note: Tpaper
rectanglthe bogoing r
diagona
The pappaper co
ely measure tive on whole
wholeness frction presententers, and hoith a tiny dotdies, we dem
or wholenesse-preserving
aper with a toleness is veed a simple ee example to
on the paper. hard to imagf the paper. without mu
latent centeand differen
e paper and ess. Seen in pporting rela4a). The whated by the d
Figure 3: (CThe dot in the s
itself, (2) the le, (7) the top
ottom right corright, (16) the
al ray to the bo
per with the donstitutes a
the degree eness.
rom a compls three case ow centers at, followed bmonstrate thas can be cretransformati
tiny dot ery subtle, buexample invoo examine hoA blank papine because However, af
uch change oers, such as nt rays fromdot pre-exisFigure 3, th
ationships amhole or the ovdot sizes.
Color online) Asheet of paperdot, (3) the harectangle, (8)
rner, (12) the re white cross bottom left corn
dot is more wsimple netw
of wholenes
lex-networkstudies to sh
are created bby studies of at wholenesseated step bions.
ut also very olving a blanow the wholper is a whol
it is pretty fter a tiny doon the surfaca halo arou
m the dot to tst. All other
hese centers among the ceverall configu
A simple whor created at leaalo, (4) the bo) the top left cray going up, by these four rner, (19) the d
the
whole than thwork contain
6
ss or life. T
k perspectivehow how whby wholeness
the Alhambrs exists in spby step by f
concrete. Tonk paper withleness emergle, which is esimple, con
ot (no more tce, but its w
und the dot, the corners (rs are inducare real, notenters, whichguration of th
oleness emergeast 20 entities
ottom rectanglcorner, (9) the
(13) the ray grays, (17) the
diagonal ray toe top left corn
he blank papning only fiv
The model is
e holeness is res. We begin ra plan and tpace. More imfollowing th
o discuss whh a tiny dot (ges and chaneasy to see. Tnsisting of fothan 0.0001 o
wholeness chthe four rec
(Panels c-h oed by the o
t just in cognh constitute he space is th
ed from a blan, listed accordle, (5) the left-top right corn
going down, (diagonal ray t
o the top right ner.)
per itself. Thve nodes (Fi
s essentially
epresented awith the simhe Sierpinskmportantly, w
he 15 fundam
holeness, AlePanels a and
nges before aThe wholeneour corners aof the sheet)
hanges dramactangles arouof Figure 3).overall confignition and psa complex
he source of
nk paper with ding to their re-hand rectanglner, (10) the bo14) the ray goto the bottom corner, and (2
e wholeness gure 4b). Ho
y a complex
as a complexmplest case stki carpet. Thrwe elaboratemental prop
exander (200d b of Figureand after a tiess of the blaand the grav) is placed, thatically. Theund the dot, . Among theguration of sychology. Tnetwork as the centers’
a dot placed
elative strengtle, (6) the righottom left cor
oing left, (15) right corner, (20) the diagon
of the originHowever, the
-network
network tudy of a rough the e on how perties or
02–2005) e 3). I use ny dot is
ank paper vitational he whole ere are at
the four e centers, space or
There are a whole strength,
th: (1) the ht-hand rner, (11) the ray (18) the nal ray to
nal blank network
represenblank pawhole ohierarchdifferent4a, the ssuch as diagonal
Figure 4
(Note: Thwith do
ComparAs a rurelationsare furthit is wea18, 19 ais very s 4.2 The The Alhcenters. salient pFigure 5space. Fmany disub-whoasymmeThe left sub-wholevel canThe whorelationseventual The plan
nting the whaper (Figure
of the paper hy, i.e., far mt communitiestrongest cencenters 8, 9
l rays. These
4: (Color onli
he first netwoot sizes indica
red to the comule, weak cenship is cumuher enhancedaker than cenand 20 are thsteep, while t
Alhambra phambra plan Thick bound
properties. Th5). There areFor example,ifferent levelole, the midetrical, conta
sub-whole coles comprisn be further ole and its nships, usuallylly formed, a
n lacks globa
oleness of the 4a). First, twith the dotore less-connes, such as th
nter number 19, 10, and 11e sub-wholes
ine) The netwo
rk consists of ating their stre
higher
mplex netwonters tend to
ulative or iterd by some in-nter 1 because weakest, bthe one in Fi
plan possesses m
daries, local here are at lee more other, the whole ols. It consistsddle sub-whaining furthercomprises thes three subbroken dow
numerous suby with smalland it is the s
al symmetry
he paper witthe whole oft contain thenected nodeshe four corne1 is followed1 as corners constitute so
orks of the whand (
f the 20 nodes,ngth. It is obv
r degree of w
ork presentedo support strrative. This i-links in an ie of the itera
because they gure 4b is ve
many of the symmetries,
east 720 posicenters crea
of the plan cs of three subhole, and thr three sub-whree sub-who-wholes: top
wn into sub-wb-wholes (orsurrounding
source of life
y, but it is ful
7
th the dot is f the blank
e 20 nodes. Ss than well-cers, the four
d by centers 2; 12, 13, 14ome implicit
holeness respe(b) the blank p, while the secvious that th
wholeness, tha
d in Section 2trong ones, ais why centeiterative manative nature oonly contain
ery flat.
15 fundamen, levels of scitive spaces oated by the wconsists of mb-wholes at the right subwholes, so tholes: left, mip, middle andwholes or cer centers) cog centers poie or beauty o
ll of local sy
far more copaper contaiSecond, the connected on
rays, and th2, 3, and so o
4, and 15 as t relationship
ectively emergpaper cond network
he paper withan the blank
2, the networas shown in
er 1 is the strnner. Althougof support ren out-links. T
ntal properticale, and stroor centers, anwholeness or
many nested the first leveb-whole. Eahere are nineiddle and rigd bottom. Th
enters at the onstitute mutnting to big f the building
ymmetries at
omplex or wins only five20 nodes for
nes. Third, the top three non. There arerays; and 1
ps within the
ged from (a) th
consists of onh the dot is m
paper.)
rks for the wn Figure 4. Trongest, withgh center 2 hlationships. T
The hierarchy
es that help ong centers and 880 relatior the overall sub-wholes l, which can
ach of the e sub-wholesght. Each of he nine sub-wthird or fourtually reinforcentral onesg complex.
different lev
whole than the nodes, wherm a striking
he 20 nodes cnodes. Seen ie several sub7, 18, 19, a complex ne
he paper with
nly the four nomore whole, o
wholeness areThis kind ofh nine in-linkhas also nineThe centers y shown in F
identify maare probably onships (Jianl configuratioor centers d
n be named athree sub-ws at the secothe middle awholes at thrth level, recrcing and su. Scaling hie
vels of scale
hat of the ereas the g scaling constitute in Figure
b-wholes, nd 20 as twork.
h the dot,
odes, both or has a
e directed. f support ks, which e in-links, of 16, 17, Figure 4a
any latent the most
ng 2015b, on of the
defined at as the left wholes is ond level. and right
he second cursively. upporting erarchy is
e, such as
the thresymmetrEach ofOverall,in an itethe spacrecurringThe planpresenceiterativedifferentwide ranand eachunderlie
Figure (Note:
aliveadaptedcontinu
4.3 The The Siepossesseassessedmeets a to far moAt each Section 1/9, anddimensioconstituttwo kindconsecutcenters oalso nesrelationseight mi
e sub-wholeries are mainf the local s the plan has
erative or recce. Eventualg in the plann was likelye of the 15 fuely applyingtiation and ange of scalesh wholeness
es the structu
5: (Color onliIt is essentia
e and beautifd to its surrouuous or iterat
Sierpinski cerpinski carpes a high ded from two a
power-law rore small thiscale, all pie5 for further
d 1/27. Eachon of log(8)te a complexds of relatiotive scales thones are enhsted relationships, the ceniddle squares
es at the firsnly created
symmetries as a good shapursive mannly the large
n contribute ay to be desigundamental pg the 15 pdaptation in s from the smdevelops, o
re-preserving
ine) The compally the densful, forming unding, and ttive different
carpet pet, althoughegree of whospects: acrosrelationship wings than largeces are the sr discussion oh of these sc/log(1/3) = 1x network asonships: thoshat are direchanced by nenships amongntral square s and the 64
st level and by numerouadapts its inpe, because i
ner. Some of number of
a great deal togned unselfcproperties. Inproperties, na step-by-stemallest to thr more trulyg or wholene
plex network oity of the censo called liv
there are far mtiation procethan the rigid
h being oneoleness. Thess all scales, with the levege ones (seesame size, wof spatial depales respecti1.89 (Mandes a whole (Pase among a cted. The sureighboring cg the 73 indhas the highsmallest squ
8
the nine suus walls or thndividual locit comprises the shapes elocal symm
o the life of consciouslyn this regardnamely tranep fashion. The largest. Eay unfolds, froess-extending
of the 720 cennters, and the
ving structuremore small c
esses. The cod tree (Alexa
e of many se wholenessand at one sels of scale (Section 5 fo
which should pendence). Tively containelbrot 1982).anel b of Figsame scale
rrounding cecenters. In adduced centehest degree ouares. Event
ub-wholes athick boundacal context omany good
echo each othmetries and m
the individua(Alexander , the plan counsformation
The space conach step creaom the previg transforma
nters and 880 eir complex e or structuracenters than mplex netwoander 1965).
strictly definof the carp
cale. Across(Salingaros
or further discbe extended
To be more spns one, eigh. The 73 squgure 6). Withthat are und
enters usuallddition to thrs (Panel cof wholenesstually, the un
t the secondaries as one or need to mshapes at dif
her in the ovemany of theal centers an1964), but nuld be thoug
properties, ntinues to beates the contous one. Thi
ations (Alexa
relationships relationships
al beauty. Evelarge ones, work as a who)
ned fractals et or the coall scales, th
1999), whichcussion on spto more or l
pecific, thereht, and 64 squares of the th the compledirected, andy support the support relof Figure 6
s, which is anderlying sca
d levels. Theof the 15 pr
make a bettefferent levelserall configu
e other 15 pnd the plan asno one can ght of as gene
to the we differentiatetext for the nis generationander 2002–2
of the Alhambs that make tery center is
which arises ole is no less
(Mandelbrooherence canhe number oh should be patial heteroless similar se are three scquares, with three differeex network, d those betwhe center onelationships, 6). Because accumulated aling hierarch
ese local roperties. er space. s of scale uration of properties s a whole. deny the erated by
whole by ed with a next one, n process 2005).
bra plan the plan well from the ordered
ot 1982), n also be f squares extended
ogeneity). sizes (see cales: 1/3, a fractal
ent scales there are
ween two es, or the there are of these from the hy of the
Sierpinsnot a tre Despite for in prmajor diis by stAlhambrSierpinsand all additionincrementhe two local ancreation divided to fit to 15 propstructure
(Note: (arepresent
of the
Throughconcreterecursivsimplicitfurther dliving gethe Earth 5. ImpliThe comfragmenthe kindlike thosof wholJiang 20living stfrom onmost int
ski carpet is see, but a com
its high degractical desigifferences betep-by-step ra lacks glo
ski carpet areshapes are
n, the Sierpinnt of the numpatterns are
nd global scaor unfoldingand differentheir local c
perties (Alexe-preserving
Figurea) the carpet wt degrees of in complex netw
h the case ste and visibleely defined cty, we adoptdiscusses theeometry is poh’s surface m
ications of thmplex-networnts. Complexd of problemse in biologyeness, is the
013b). Seen tructure that
ne to anotherteresting part
shown in a trmplex network
gree of wholegns. Instead,etween the twgeneration;
obal symmete the same sisquares in tnski carpet mber of squae the same aales. The whg: differentia
ntiated to meecontexts. Desxander 2002transformati
e 6: (Color onlwith three scaln-links, (c) neswork, in which
tudies, the se, helping rcenters, and ted these sime implicationowerful and
more whole o
he complex-rk perspectiv
x networks pm a city is (Jy. The notion e same as onfrom the whdetermines t
r, but there it for urban d
ree structure k.
eness, the Si we seek thewo patterns. the Sierpins
try, but withize in one scthe Sierpinskis too rigid
ares across twat the fundamholeness of tation and adaet the scalingsign or makin2–2005, Salions toward a
line) A compleles, (b) its comsted relationshh the dot size
d
eemingly abreduce the m
the centers mple cases to ns of the com
unique in teror more beau
-network peve enables uprovide a powacobs 1961)that a city is
ne city beingholeness or ththe image ofs a shared imesign theory
9
(Panel d of F
ierpinski carpe kind of patThe Sierpin
ski carpet hh only local cale, while thki carpet, win terms of
wo consecutimental level the Alhambraptation. In tg hierarchy, ang is much m
lingaros 201a whole with
ex network pe
mplex networkhips among threpresents the
degrees of life
bstract concemystery of ware induced illustrate th
mplex networrms of plann
utiful.
erspective anus to see thinwerful mean). A city is es more wholeg more imaghe theory off the city. Thmage of the
y, which has b
Figure 6). It
pet is not exttern like theski carpet is
has local ansymmetries
hey are morewhile the Alf the scalingive scales is in terms of
ra can only bthe course oand newly admore challen13, Salingaroh a higher de
erspective on k of 73 pre-exhe 73 created ce strength of te.)
epts of wholwholeness. Tor created be subtlety ofrk perspectivning and repa
nd wholenesngs in their wns to further essentially pre than anotheginable or lef centers, it ihe image of city for all
been persiste
should be no
xactly the type Alhambra p
by creation,d global sy; all centers
e or less simihambra has g ratio at prby exactly e
f wholeness obe obtained f design, a wdded centersnging than unos 2015) prgree of whol
the Sierpinskiisting centers,centers, and (dhe 73 pre-exis
leness and cThe wholene
by the wholenf the wholenve to argue wairing our env
ss wholeness, rastudy whole
roblems of oer, or that a c
egible than as the city itsthe city for ipeople. Thisently criticize
oted that the
pe of patternplan. There , while the A
ymmetries, ws (or squaresilar in the Adifferent sh
recisely 1/3,eight times. Hor coherencethrough step
whole is conts or scales arnderstandingrovide guidleness.
i carpet , in which thed) the scaling sting centers o
centers becomness compriseness. For theness. The nexwhy the whonvironment o
ather than aseness and unorganized cocity has a higanother (Lynself or the unindividuals ms shared imaed for lackin
whole is
n we look are some
Alhambra while the s) of the
Alhambra; hapes. In and the
However, e at both p-by-step tinuously re created g, but the dance for
dot sizes hierarchy
or their
me more ses many e sake of xt section leness or r making
s parts or nderstand omplexity gh degree nch 1960, nderlying may vary age is the ng a solid
10
and robust scientific underpinning (Jacobs 1961, Marshall 2012). The complex network perspective or more truly the wholeness itself will inject scientific elements into urban planning and design. The Sierpinski carpet is an image of the Earth’s surface metaphorically. This is because there are far more small things than large ones across all scales – the spatial property of heterogeneity, and related things are more or less similar in terms of magnitude at each scale – the spatial property of dependence. Both heterogeneity and dependence are commonly referred to as spatial properties about geographic space or the Earth’s surface (Anselin 1989, Goodchild 2004). The carpet is also an ideal metaphoric image toward which our built environment should be made. It implies that any space ought to be continuously differentiated to retain the scaling hierarchy across all scales ranging from the smallest to the largest, and any building or city ought to be adapted to its natural and built surroundings at each scale. The two processes of differentiation and adaptation enable us to create a whole with a high degree of wholeness. In this regard, the wholeness or the theory of centers would have enormous effects on geography, not only for better understanding geographic forms and processes, but also for planning and repairing geographic space or the Earth’s surface (Mehaffy and Salingaros 2015, Mehaffy 2007). Built environments must adapt to nature, and new buildings must adapt to their surroundings. The two spatial properties of heterogeneity and dependence constitute a true image of the Earth’s surface at both global and local scales. Globally, spatial phenomena vary dramatically with far more small things than large ones across all scales, while locally they tend to be dependent or auto-correlated with more or less similar things nearby. These two properties are the source of the two kinds of harmony or coherence respectively across all scales and at every scale. For example, there are far more small cities than large ones across all scales globally (Zipf 1949), whereas nearby cities tend to be more or less similar in terms of the central place theory (Christaller 1933, 1966). However, the geography literature focuses too much on dependence, formulated as the first law of geography (Tobler 1970), but very little on heterogeneity. More critically, spatial heterogeneity is mainly defined for spatial regression with limited variation, governed by Gaussian thinking (Jiang 2015d). This understanding of spatial heterogeneity is flawed, given the fractal nature of geographic space or features. Spatial heterogeneity should be formulated as a scaling law because it is universal and global. The wholeness or the theory of centers in general brings a new perspective to the science of complex networks. For example, the detection of communities of complex networks can benefit from the wholeness as a recursive structure. Instead of a flat hierarchy of community structures, a complex network contains numerous nested communities of different sizes, or far more small communities than large ones (Tatti and Gionis 2013, Jiang and Ma 2015). This insight about nested communities can be extended to classification and clustering. Current classification methods can be applied to mechanical assembly, so mechanical parts can be obtained. For a complex entity such as complex networks, the parts are often overlapping and nested inside each other. This resembles both wholeness and centers as a recursive structure. From a dynamic point of view, a complex network is self-evolved with differentiation and adaptation. A complex network has the power to evolve toward more coherence at both global and local scales, and is self-organized from the bottom up, rather than imposed from the top down. This insight about self-organization reinforces Alexander’s piecemeal design approach through step-by-step unfolding or transformations for a living structure with a high degree of wholeness. 6. Conclusion This paper develops a complex network perspective on the wholeness to make it more accessible to both designers and scientists. I discussed and compared related concepts such as whole and parts versus wholeness and centers to make them explicitly clear. A whole is a relatively coherent spatial set, while wholeness is a life-giving structure – not something about the way they are seen, but something about the way they are (Alexander 2003, p. 14). The wholeness is made of centers rather than
11
arbitrarily identified parts. The centers are created or induced by the wholeness and made of other centers, rather than just those pre-existing in the whole. Given these differences, the mantra that the whole is more than the sum of its parts should be more correctly rephrased as the wholeness is more than the sum of its centers. I demonstrated that the complex-network perspective enables us to see things in their wholeness. More importantly, I elaborated on design or making living structure through wholeness-extending transformation or unfolding in step-by-step differentiation and adaptation. Although understanding the nature of wholeness is essential, the ultimate goal of Alexander's concept of wholeness or living geometry is to allow us to make artifacts and built environments with the same order and beauty of nature itself. The complex-network perspective enables us to develop new insights into planning and repairing geographic space. For example, differentiation and adaptation are two major processes for making living structures or geographic space in particular. This study showed that these processes underlie the two unique properties of geographic space: spatial heterogeneity and spatial dependence. These are respectively formulated as the scaling law and the first law of geography. Globally across all scales, there are far more small things than large ones. In contrast, locally at every scale, things tend to depend on each other with more or less similar sizes. These two spatial properties are the source of the two types of coherence respectively at global and local scales. In this connection, the theory of centers or living geometry would significantly contribute to understanding and making geographic space. Our future work points to this direction on how to rely on wholeness-extending transformations to create geographic features with a high degree of wholeness. Acknowledgement XXXXXXXXXXX References: Alexander C. (1964), Notes on the Synthesis of Form, Harvard University Press: Cambridge,
Massachusetts. Alexander C. (1965), A city is not a tree, Architectural Forum, 122(1+2), 58-62. Alexander C. (1979), The Timeless Way of Building, Oxford University Press: Oxford. Alexander C. (1993), A Foreshadowing of 21st Century Art: The color and geometry of very early
Turkish carpets, Oxford University Press: New York. Alexander C. (2002-2005), The Nature of Order: An essay on the art of building and the nature of the
universe, Center for Environmental Structure: Berkeley, CA. Alexander C. (2003), New Concepts in Complexity Theory: Arising from studies in the field of
architecture, an overview of the four books of the Nature of Order with emphasis on the scientific problems which are raised, http://natureoforder.com/library/scientific-introduction.pdf
Alexander C., Neis H. and Alexander M. M. (2012), The Battle for the Life and Beauty of the Earth, Oxford University Press: Oxford.
Anselin L. (1989), What is special about spatial data: Alternative perspectives on spatial data analysis, National Center for Geographic Information and Analysis, Santa Barbara, CA.
Barabási A. and Albert R. (1999), Emergence of scaling in random networks, Science, 286, 509–512. Bohm D. (1980), Wholeness and the Implicate Order, Routledge: London and New York. Christaller W. (1933, 1966), Central Places in Southern Germany, Prentice Hall: Englewood Cliffs,
N.J. Goodchild M. (2004), The validity and usefulness of laws in geographic information science and
geography, Annals of the Association of American Geographers, 94.2, 300–303. Jacobs J. (1961), The Death and Life of Great American Cities, Random House: New York. Jiang B. (2013a), Head/tail breaks: A new classification scheme for data with a heavy-tailed
distribution, The Professional Geographer, 65 (3), 482 – 494. Jiang B. (2013b), The image of the city out of the underlying scaling of city artifacts or locations,
12
Annals of the Association of American Geographers, 103(6), 1552-1566. Jiang B. (2015a), Head/tail breaks for visualization of city structure and dynamics, Cities, 43, 69-77.
Reprinted in Capineri C. and Huang H. (editors, 2016), European Handbook on Crowdsourced Geographic Information, Ubiquity Press: London.
Jiang B. (2015b), Wholeness as a hierarchical graph to capture the nature of space, International Journal of Geographical Information Science, 29(9), 1632–1648.
Jiang B. (2015c), A city is a complex network, in: M. W. Mehaffy (editor, 2015), Christopher Alexander A City is Not a Tree: 50th Anniversary Edition, Sustasis Press: Portland, OR, 89-98.
Jiang B. (2015d), Geospatial analysis requires a different way of thinking: The problem of spatial heterogeneity, GeoJournal, 80(1), 1-13.
Jiang B. and Ma D. (2015), Defining least community as a homogeneous group in complex networks, Physica A: Statistical Mechanics and its Applications, 428, 154-160.
Köhler W. (1947), Gestalt Psychology: An Introduction to New Concepts in Modern Psychology, LIVERIGHT: New York.
Lynch K. (1960), The Image of the City, The MIT Press: Cambridge, Massachusetts. Mandelbrot B. (1982), The Fractal Geometry of Nature, W. H. Freeman and Co.: New York. Marshall S. (2012), Science, pseudo-science and urban design, URBAN DESIGN International, 17,
256-271. Mehaffy M. W. (2007), Notes on the genesis of wholes: Christopher Alexander and his continuing
influence, URBAN DESIGN International, 12, 41–49. Mehaffy M. W. and Salingaros N. A. (2015), Design for a Living Planet: Settlement, science, and the
human future, Sustasis Press: Portland, Oregon. Newman M. E. J. (2004), Detecting community structure in networks, European Physical Journal B,
38, 321-330. Newman M. E. J. (2010), Networks: An introduction, Oxford University Press: Oxford. Page L. and Brin S. (1998), The anatomy of a large-scale hypertextual Web search engine,
Proceedings of the Seventh International Conference on World Wide Web, 107–117. Salingaros N. A. (1998), Theory of the urban web, Journal of Urban Design, 3, 53-71. Salingaros N. A. and West B. J. (1999), A universal rule for the distribution of sizes, Environment and
Planning B: Planning and Design, 26(6), 909-923. Salingaros N. A. (2013), Algorithmic Sustainable Design: Twelve lectures on architecture, Sustasis
Press: Portland, Oregon. Salingaros N. A. (2015), United Architectural Theory: Form, language and complexity, Vajra Books:
Kathmandu, Nepal. Simon H. A. (1962), The architecture of complexity, Proceedings of the American Philosophical
Society, 106, 468-482. Tatti N. and Gionis A. (2013), Discovering nested communities, Blockeel H., Kersting K., Nijssen S.,
Železný F. (editors), Machine Learning and Knowledge Discovery in Databases, 32-47. Tobler W. R. (1970), A computer movie simulating urban growth in the Detroit region, Economic
Geography, 46(2), 234-240. Watts D. J. and Strogatz S. H. (1998), Collective dynamics of `small-world' networks, Nature, 393,
440-442. Zachary W. W. (1977), An information flow model for conflict and fission in small groups, Journal of
Anthropological Research, 33(4), 452-473. Zipf G. K. (1949), Human Behaviour and the Principles of Least Effort, Addison Wesley: Cambridge,
MA.