Now More Than Ever : Computational thinking and a science of design(<Special Issue>What is "What's the Design"?)

NII-Electronic Library Service

Now

More

Than

Ever

-Computational

_thinking

_{and ascience ot}

design-MarkDGROSS

CarnegieMellonUniversity

L

50

Ab$tract

Revolutions in desktop rnanufacturing and embedded

computing are changing theway we make things. The$e

changes willenable citizenstoengineer and manufacture their own _goods.The role ofthedesignerisal$ochanging, deciding on themanufacture of specific artifacts, tosetting thebounds and rulesfordecislonsthatend-users make, Materialsare changing tool: Programmable matter made of ensembles of modular robots demands new and dynamic ways of describjng

designs.A science of design isan essential element tor

beneMing fromthese revolutions, and it

[s

likely

thata science ofdesignwll1beexpressedcomputationally,

Keywords

Computational

think'ng;

ptogrammablo matter; end-user design

1.Introduction

FortyyearshavepassedsinceHerbSimonwrote hisinfluential

The Sclencesof theArtiticialinwhich hecoined theterm,the Scienceof Design

_[1].

Atthattime,inthe late1960s, _people

hada _growingsense thattheworld we make and liveinwas

growjngso complex thatthetraditiona[ways otdesigningwere no iongetadequate tothe task,lfitwas truethen, itistruer today,

Yetforthe_pastseveral decades,designresearch hasbeen viewed with seme skepticism.

Many

look

down

on

design

research, believingthat

_(in

the words of Alexander's1971 prefacetohisNoteson the Synthesisof Form),'`People

who study designmethods without also _practicingthem are almo$t always frustrateddesignerswho have no sap inthem, who

have lost,or never had,theurge toshape thjngs.Such a personwillnever beabletosay anything sensible about [`homt' toshape thingseitheil'

_[2].

Withinthe engineering community, too,theideathattheremight be a

['science

ofdesign" has met with some skepticism. Many excellent engineers believethat research on design isa

`soft'

fieidof study. They believethat

i'V.r;.7iAsckMe-Whaiis'WhaVstheDesign'? SpeciallssueoflapaneseSacietyfottheS[lenceofDesign

voL16-2na622009

engineering designisdrivenbythe_propertiesand behaviors of specjficdomains.They thinkthatother thana need for"proper thinking"there islittleinterestingthatwe can say in_general aboutdesign.

Withdue respect tothese doubts, inlightoftherevolutions in manufacturing and techno[ogy thatwe are now experiencing, we can no longeraffordtoview ascience ofdesignas$oftoras an irrelevantintellectualgame.Rather,ascience of designisa necessaty foundation for the changes thatare already

beginningtopervadeoureverydayjive$.

Firstletusclearlydrawaiinebetweenthe''scienceofdesign"

and the[`science _{ofdesigners'i.Both}

{but

especially thelatter)

havebeenthesubjectsofagreatdea1ofresearchoverthepast several decades,The science ofdesignisthestudy ofdesign

processes,

regardless ofwho, or what, isdoingthedesign.For example, researchers may investigatehow a space ofdesigns can beefficiently searched, or how a notation or languagecan compactly express a class of designs,The science of

designersisthe study of human designers,how they think, what they do, and hew they communicate, Forexamp[e, researchers analyze designers'drawings,ask designer$to thinkaloud as theydesign, and videotape designers working together in_groups

_[3],

Bothfieldsof study are interesting,but theyarequitedifferententerprises.

2.Radical changes

in

how we make things

Today we are inthe earlystages ofa _prefoundchange inthe way we designand construct our _physicaiworld. Itisnot the firsttime thishashappened.ChristopherAlexandertellsthis story inNoteson theSynthesisofForm

_[2].

Atthebeginningof the industrialage inthelate18thcentury oursociety moved

from individualcraftsmen making artifacts one by one, to

assembling

(by

hand>oblects fromstandard cemponents, and thence tomass _productioninthe20thcentury

[4].

Witheach

shM in_productionha$ come acorresponding shiftindesigning.

Inthe age ot craft, designwas implicit-basedon a $hared understanding of common _goods.When you needed a new

NII-Electronic Library Service gateora hammer you went totheblacksmithand explained

what _youneeded and hemade _you one. The industrialage

broughtthe need ferexp]icitdesigningtoconsjder thefunction

of the artifact and toplan the materials and methods of

producing

jt.The designeror engjneer made paperdrawings

and models toplantheartifact.

When

the

designing

was

done

thedrawingswere used formanufacture. Mass

_production

made designeven more necessary, as highcosts of tooling

and setting up manufacturing lines _{demanded that}we

thoroughly thinkthrough an adifact beforebeginningtomake it. Today,designersno [ongeruse _paperdrawingstoconceive, consider, and convey theirdesigns.Instead,tilesare stored and transferred electronically fromdesignertomanufacturer.

Stiliwe are mostly inthemode ofmaking drawingsand mode[s todesigntheartifactswedesire.

Each shift inthe designand _productionof our _physical environment has resulted inbroad and _protoundimpactson our society

in

myriad ways

(health,

education, socialand economic

order) thatwou[d have been almost impossibletopredict.

Thereisevery reason tobelievethatthechanges ofour time willhave even broadersocietal]mpactthanthosethathave

come before.The fundamentalchange inmanufacturlng and

production

thatwe are inthemidst otnow ha$ thecapacity to enable and empower ordinary citizensinways thathavenever beforebeen_possibie.Byleveragingthescience of design,the engineering of desktop manufacturing, and thesoftware to

bring the two together,we can make the vision of "democratizjng _innovation"

[51

come true.[wil[argue thatitis the software, and _particularlycomputationai way$ of expressing designknowledgeand expertise thatwillbringthis dream toreality.

_.

The shiftinthe_productionof the_physicalworld derivesnot

fromaslngle technological advance butfromdevelopmentsin

severai arenas: computer controlled _tools

(e.g,,

de$ktop

manufacturing),embeddedcomputing,andscienceofdesign.

3.Desktop desjgn & manutacturing revolution

Already underway isthe shiftto desktop manufacturing

-peoplecan attord todesignand manufacture one-off artifacts

forthemselves. An early example was desktop _publishing.

Beforewe inventedlaser_printersit

_you

wanted abrochure,a newsletter, or a _po$teryou worked witha _graphicartistto designa layout,select

typetaces,

_paper,and so on _;and then wlth a _printerwho would execute the design and _producethe

L

fina[inked_{paperproduct,}Today laserand jnkietprintersare

practicallyfree,and using desktopsoftware anyone can design and _printtheirown newsletters, calendars, wedding invitations, and even

books,

The

desktop

_publishingrevolutionwas driven

firstby the development of laser_printingteehnology.

Application

software enabled _{professjonals}at first,and

eventually end users, to _{produce graphic}work, A key

component was the underlying Postscrjptlanguage that

applicatlonsusetoproducepagedescription$forlaserprinters,

Laserand water

jet

cutters,three-dimensionaiprinters,and computer-numerically controlled milling machinery are now extending thisshiftfromthe mostly flatworld ot paperand graphicarts tothericher threedimensiona[world of physical objects. The first

Cor

second) _generationofhardwaretosupport this revolution in "desktop

manufacturing" is already commercjalized and capabilitjes continue toadvance a$ costs drop

[6,

7].The specific technologies vary fromfasersintering

tofused

deposition

mode1ing,

but

we are c[ear[y moving along a trajectoryfromsingiematerjal

{e,g,,

_plasticor metal) tomultiple materials, tothe abilitytomanufacture-in small quantities-unique

_physical

ebject$ with embedded electroniccircuitry, printeddlsplaysand other actuators

(http:Mabathome.org

_; http:"www,2obiet,com). As with desktop _publish]ng,software

playsa key role: Computer-aideddesignand engineering app[ications todescribephysical obiects and simulate their

behavior,and the underlying representations

_<analogou$

to PostScripOenable designerstodotheirwork,

4. Embedded computing revolution

A second revolution-this one incomputing-is also underway :

We are embedding microcontrollers, actuators, and sensors

intoour _physjcalenvironment, and the communication and control of these devices

_[9],

Advances inmicro-

{and

nano-) electronjcs ieadingto

_low

cost sensors and actuators, micro-controllersthatareaspowerfulasyesterdayismajnframes,and

newwirelesseommunication$protQcolsiue1thisrevolution,We

see itsimpactineveryday Iivesas our clothing,our turniture, ourbuUdings,automobiles, and citiesbeceme computational{y enhanced. Applicationsare simple so tarbutalready we have

the capacity to make things that exhjbit computationa[ly complexbehaviors,

The desktopmanufacturing revolution applies hereas well. As recently as a decade ago only an experienced engineer

could designand manufacture a _printedcircuit board

(PCB).

j f,tl).#ffst/wte-whatjs'What'stheDesjgn''?

Spe[lalFsHueofJapHmeseSecietyfortheScienceofDesign

vol.16-2 mo.6?, 2009

NII-Electronic Library Service Today even a highschool student ean easily acquire theskills

todesigna boardusing off-the-shelf software and send thefile toafabricatorforlow-costovernightmanufacture.Inkjetprinter

companies are now envisioning afferdable

desktop

PCB

manufacture

[8],

5.Design methods & science ofdesign

In addition to advances in desktop manufacturing and

ubiquitous computing, another relativelyrecentdevelopmentis relevant:The recognition thatthe compiexity ofthe thingswe

make and theirinteractioninthe world demands thatwe understand designingbetter.Althoughitsroots _gobackfurther, inthe mid-1960s researchers-in what became known as ''design

methods movement"-began to recognize that increasing complexjty demanded a comprehensive understanding of designlng-in Simon's memorable _phrase,a

"science

_{of design"}

[1].

A(though thefocused intensityefthe

design

methods movement

faded,

theagenda

did

not.

Today

the software engineering and human-computer interaction

communities have embraced Alexander's"Pattern _Language" approach

_[9].

HorstRitte['s"issue _{basedinformationsystems"}

[1

O]ledtodesignrationaleand knowledgemanagement,

The recent US NationalScience Foundation"Science _ot Design: SortwarelntensjveSystems''initiative

[1

1]isfurther evjdence thatthe need$ thatdroveSimon,Ritteland other$ in the 1960 sand 1970s-to understand designjnginthetaceot

increasingcomplexity-remain relevant today. We stilllacka coherent fundamentalscience of design

(an

understanding of

the structure of design decision making, abstracted trom

specific domains). StHl,we have seen steady _progressin

modeling design processes and developingcomputational

designmethods and tools.As computer hardwareadvanced and more _{powerfulprogramming} environments became the

norm, theearly insightsofthe designmethods movement took form inincreasingly

_poweriul

computer-ajded design

_(CAD)

toolsforarchitectural,mechanical,electrical,civi1,andsoftware engineering,

6.Code as thecarrier fordesign expertise

The move, starting inthe 1960 s,fromdesignbyhandtodesign

w[th computer toolsenabled us tobegintoautomate some of the reasoning and decisionmaking thatisat theheart of designing.One of theear[iest examples, of course, was lvan

Sutherland'sSketchpad program. Sketchpad i$ known for

52'J'Vf).t'nvsc#"sc'/tt

Spec/lul)ssL/eofJfipaneseSoc[eVvfDrtheScienc/eofDesign

'-

VJhatis''What'stheDesLg//''?

vol,1fi ? mos2 E-eg

many things,_buttorthescience otdesign,Sutherland'smost important contribution inSket¢hpad was todescribea design as a set ofconstraints _thatthe_program could manage as the

human

designer

made change$. Later,duringthei980s and

1990s,researchers inexpertsystems, case basedreasoning, and other fieldsot artjficialintelljgence,followedthis_general approach and applied the$eideastodesigninmany different

domains-from buildlngsto circuits to software to machines,

Advances incomputer hardwareand software duringthe 1980

s and 1990 smade it_possibletoimplementtheideas thatthe deslgnmethods researcher$ hadworked on inthe 1960sand

1970s.

What isimportantabout this_pieceofhistoryisthatsoftware became the medium foFcarrying themethods and techniques that the early design researchers developed.In a kind of chicken-and-egg _process,as the software became more

sophisticated, designersin _practicebegan toadopt jtand dependon

it,

lnsome fields,notably integratedcircuit design,

thesoftware began toembed automated designmethods that

human designerscould not_performinreasonable amounts of

time.Designknowledgeand expertise begantotakethe form

ofcode. Designersbegantoadopt computationa[ thinking

_[1

2].

Sti11,duringtheshiftfromdesign-by-handtocomputer-aided

design,thedominantmodelhasbeenthecomputerprogramas toolor assistant tothedesigner.Thedesignerisincontrol and makes aH the decisjons.The computer has served mostly to record and displaythedecisionsthe designermakes, and to ca[culate, Iook"p, and render informationabout the design. AIthough adopting computer aided design tooishas affected

design_practice,so farwe have experienced on[y a small departure fromthetraditionalway of making design decisions. Thatisabout tochange.

7.End-user design and computational thinking The revolutionsindesktop manutacturing and In embedded

computing push us towards computational ways ofthinking

about design.One example isend-user designing,now

becoming _popularas [co-creation'

[1

3].End-userdesignisthe ideathatas we move awayfrom ma$s-production and embrace

theideaofindividualizedor `mass-customized' _{manufacturing,} ordinary citizens will be able to designand make things for themselves, We are seeing the firstwave of co¢ reation, in which citizens

(sometimes

called "consumers"}

participatein

making decisions about a design, The examples are many,

NII-Electronic Library Service

from

shoes tocars totoys,

(AIthough

itisnow

becoming

popular,enabling end users todirectlymake design decisions

is

an old

idea

: Beginning

in

the earlyt960s Dutch

design

methodologist N. John Habraken developed a theoryand

method forengaging citizens inthedesignottheirhousing

[14],)End-userdesignrequiresprofessionaldesignerstosetup

a design$pace thatcitizenscan work within,They specitythe

rules that_governtheend-userdesigns.

(This

too,ofcourse, isa

de$ignact.)Today thedesign and

_production

_process

is

usually computationally mediated, so thebounds ofthe space and the ru[es that _govern designs are a[so expressed computationally.

The

advances inpersonaldesktopmanufacturing thatare empowering end-users todesign and manufacture theirown

goods demand advances in software. We need repre$entations to describe designs and applications to

manage and manipulate those representations.

The

representationsaredesign1anguagesthatmachinescanparse, recognize, and process,The applications are compilers and other development tools. Instructions in a high-level programming languageIikeRubyorLispdescribethebehavior

we want a computer to_perform.Instructionsina high-level

designlanguagedescribewhat we want ofour designartifact.

A

_.design

compiler takes high-leveldescriptionsofthebehavior

and

generates

implementationintheformot an object.For

example, acompiler might _generatecode thata 3D printer,or

other desktop manufacturing machine can execute to

physjcallyproduceadesign.

Itmight seem thatthisway ot desjgnjngwill Iimitcreativity,

The opposite istrue.Computationaldescriptionsofdesign will enrich, not impoverishopportunities toreveryday creativity. It should beclear, then,thattheway thatthe computational tools tordesignare contjgured willstrongly color the ways inwhich citizens can becreative. Nakakojihas outlined an interesting and valuable framework forunderstanding-and designing-computationaltool$tosupportend-usercreativityindesign[151,

8,

The

_programmable

world

Another,perhap$ even more profound, change ison the horizon: a phy$icai werld whose behaviorthatwe can

program,We already see mjcroprocessors embedded into

many of our everyday thjngs-from clothing to transportation-and withthatcomes theabi1ityto_programtheirbehavior.As our

thingsand our world become enhanced with computation, we

must findways forcitizenstoprogram and repregram their

behavior.As with our end-user designstory,citizensbecome de$ignersofthe

dynamic

behavjor

of thjngsand _placesinthe

world.

Alogicalextension otthecomputationally embedded things

we havetodayisawor[d bu[ltfromensembles ofthousandsof

modular robots. Each robot would be able to sense its

immediateenvjronment, move itselfand _perhaps itsrobot neighbors, and communicate with otherrobotsintheensemble. The robotscould be programmed torespond automatically to

changes intheirenvironment, or tochange configurations on command. Forexample, a buildingmade of robot building

blocks

_[1

6]coulci recontigure itselftoadapt toditterentweather conditions, differentuses, or torespond toemergencies such as earthquakes, tires,or floods,Althoughmaking thisideaa realitymay seem farinthefuture,several research _groupsare

developjng

the core technologies for''programmable

mattef' today[17-19],

lfprogrammable matter becomes an everyday reality,hew

wi11we designforit?Aswe saw withend-userdesign, therole of theprofe$sionaldesignerwillchange.NolongerwHTthejobofa designerbe tomake informeddecislon$about a specific artifact,lnstead,the

_job

ofthe

_professional

designerwillbeto program the artifact's dynamic and responsive behavior.Or rather,to

_program

thedynamicand responsive behaviorofthe

ensemble ofmodu]ar robots of whjch the artifact ismade. To thedesignersoftoday,thismay seem a _quitedifferentkindof

job

thanwhat we usually thjnkof as design. Really,though, the designeristask will stjllbe-as italway$ has

been-to

create thingsthatmeet certain needs. The differencei$thatinsteadof creating thethingsdirectlyina"one-off"

fashion,thedesigner

willprogramthematerialstorespondtodifferentconditions.

9.Discussion

We beganwith areference toSimon'slectureon theScienceof

Deslgn.

Simon

made

his

remarks atatjmeof_greatsocial and technologicalchange around theworld.We aretodayagain at atjmeof

_great

change :enormous chal[enges tace humanity-climate change and itseffects, theneed toteeda

_growing

world population,mass urbanization, and so on. More thanever we need a scjence of design-a [igorous and systematic understand]ng ofhowtodesign.

Ascience ofdesign _{promisestobedomain-agnostic.Thatis,} the ideaofa science ofdesignisthat,apart fromthe domain

rvft'1''-"rmg--Whatis'Whai'stheDeslgn"?

Spt,ciallssued･lupaneseSocietyfoFtheSdenceofDesign

vol,16-2 no.6? 2009

NII-Electronic Mbrar

NII-Electronic Library Service

54

specific expertise of rockets, hearing-aids,anti-retroviral drugs, or _{publicpolicy,}there isalso knowledgeand expertise inthow todesign]thatwe can bringtobearon each domain.As we

mentioned above, thisideahasmet with some skepticism inthe design,engineering, and scientific communities. And to be honest,so farithas not bornethe fruitthatwe optimistically

hopedforintheearlydaysofthefield,

1havearguedthatthewaytoascienceofdesign-athorough

and systematic understanding of the _processes to reach desiredoutcomes-lies in the approach of computationa[ thinking.Over the_pastdecades,computer-aided de$ignhas become widely _practicedin every design domain-from architecture,to industria[design,toelectronics engineering. The most importantcontributions of computer-aided design

have not been inmore realistic renderings or _performance simulations. Certainly,these havebeen valuable. Butthe real contribution ha$ been tooffer computation as a way of conceiving design,as a med[um forexpressing and exploring

designideas.Computationalrepresentations

-

not onlyofthe form ot things, butarso of theirinteractivebehavior-area

powerfulway torepresent designsand design_processes.That iswhy 1believethatifwe are tohaveascience of

design,

itwill tikelybecomputationailyexpressed,

1argued also that_{thetechnological}changes inour world

today are already moving us toward a

_protoundly

computational view of designing.Inthisworld, the designer's

role willnot be merely tomake obiects for

_people,

butto

describedesignspaces and the rules thatboundthem, inways

thatwillenable citizens todesigntheirown things,and thatwill

provide''programmable

mattef' with dynamic and responsive

behavior,

References

1.Simon,H.

₍₁₉₆₉₎

Sciencesof theAM'ficia4Cambridge, MA :MIT Press.

2,Alexander,

C.

(1966}

Notes on the

Synthesis

of Fban

Cambridge [HarvardUniversityPress.

3.Suwa,M,,Tversky,B.

(1

997)What architects and students

perceiveintheirsketches: A protocolanalysis. Design

Studea

(18),385-403.

4.Giedion, S.

₍₁₉₄₈₎

MechanZzation 7lates Command: A

Conmbufion

toAnorymous MStoty / OxtordUniversity

Press.

5.von Hippel, E.

₍₂₀₀₅₎

Democrama'ng binovation

j'V-r;.7bl."",vaEe-Whatis''VVhat'stheDesign'7 SpecLallssueoflapaneseSocletyfertheScianceofDesign yoL16-2no.622009

Cambridge,MA :MITPress.

6.Arabe,K.C,

_(2004}

3D PrintingMakes [tsMark.Industrial

MarketTrends,

http

:

"news,thomasnet.comtlM17archivesf

2004f09I3dntprinting-mak.html

7. Hansell,

S.

(2007}

Beam ltDown From theWeb, Scotty,

AiewYOrkimes.

8.Epson, I.

₍₂₀₀₄₎

Epson InkjetTechnology U$ed to

FabricateWorld'sFir$tUltra-ThinMultj)ayerCircujtBoard. 1111t2004, http//fwww.epson.co,jpfe/newsroom/newSm 2004 11Ol,htm

9.Alexander,C.,lshikawaS.,Siiverstein,M.

₍₁

977),4 pattem

language : tovvns bu"dfrTgqconstnyction, New York:

OxtordUniversityPress.

10,Rittel,W., Kunz, W.

₍₁₉₇₀₎

lssuesas elements of

information systems.

Center

for

Planning

and

DevelopmentResearch,UniversityofCalifornia,Berkeley.

1ri.Freeman, P.,

Hart,

D.

(2004)

A

science of

design

for

software-intensjve$ystems. _{Commtm. ACM 47(8),}19-21. 12.Wing, _J,M,

₍₂₀₀₆₎

_{Computationalthinking.Communicatibns}

oftheACM,49(3),33-35.

13,Trendwatchlng.CustomerMade. 2006

http:Mrendwatching,comftrendstCUSTOMER-MADE.htm.

14.Habraken, _N.J.

_<1972>

_Sappons :an AltemativatoAvlass

Hbusing London: The ArchitecturalPress,and New

York/ Praeger.

(Originally

published

inDutchunder the

title:De Dragersen deMensen.Am$terdam :Scheltema

en Holkema,1962,),1962,

15,Nakakoji,K,Meanings of Tools,Support,and Uses for

CreativeDesign Processes. in Proceect'ngsof the

lntemattonalDesign Reseatvh

$JtmposiumOa

Seoul

2006i156-165.

16,We[jer,M.P.,Do EY-L.ArchitecturalRobotics: A new

paradjgmforthe bujltenvironment. inEuropM.11

.'

11th

lhtemationaiConierence on Design Sciences &

lechnologJtMontreal2007, 353-362.

17.

Goldstein,

S.C.,Campbell,J,D.,Mowry, T.C.

_{2005)

ProgrammableMatter.IEIIECompute4 38<6),99-101.

18. Murata,S. et al.

(2002)

M-TRANz selt-reconfigurable

modular robotic system, Mechatonics tEEenSME

Ttansachans on ;7{4),431

-

441.

19,Yim, M. et al,

(2001)

DistributedControl tor3D

Metamorphosi$.AutonomousRobots,10(1>,41-56.