Nicolson_10 Heuristics for Interdisciplinary Modeling Projects

7/29/2019 Nicolson_10 Heuristics for Interdisciplinary Modeling Projects

1/10

Ten Heuristics for Interdisciplinary Modeling ProjectsAuthor(s): Craig R. Nicolson, Anthony M. Starfield, Gary P. Kofinas and John A. KruseSource: Ecosystems, Vol. 5, No. 4 (Jun., 2002), pp. 376-384Published by: SpringerStable URL: http://www.jstor.org/stable/3658975 .

Accessed: 23/09/2013 01:42

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of

content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms

of scholarship. For more information about JSTOR, please contact [email protected].

.

Springeris collaborating with JSTOR to digitize, preserve and extend access to Ecosystems.

http://www.jstor.org

This content downloaded from 148.206.159.132 on Mon, 23 Sep 2013 01:42:59 AMAll use subject to JSTOR Terms and Conditions
http://www.jstor.org/action/showPublisher?publisherCode=springerhttp://www.jstor.org/stable/3658975?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/3658975?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=springer


2/10

Ecosystems (2002) 5: 376-384DOI: 10.1007/s10021-001-0081-5 ECOSYSTEMS 2002 Springer-Verlag

ORIGINAL ARTICLES

T e n Heuristics f o r InterdscipiModeling Projects

Craig R. Nicolson,l* Anthony M. Starfield,2 Gary P.John A. Kruse4Kofinas,3'4 and

'Department of Natural Resources Conservation, University of Massachusetts, Box 34210, Amherst, Massachusetts 01003-4210,USA; 2Department of Ecology, Evolution and Behavior, University of Minnesota, 1987 Upper Buford Circle, St Paul, Minnesota55018, USA; 3Institute of Arctic Studies, Dartmouth College, 6214 Fairchild, Hanover, New Hampshire 03755, USA; and 4Instituteof Social and Economic Research, University of Alaska-Anchorage, 3211 Providence Drive, Anchorage, Alaska 99508, USA

ABSTRACTComplex environmental and ecological problemsrequire collaborative, interdisciplinaryefforts. Acommon approachto integratingdisciplinaryper-spectives on these problemsis to develop simula-tion models in which the linkagesbetween systemcomponents are explicitly represented. There is,however, little guidance in the literatureon howsuch models should be developedthroughcollabo-rative teamwork. In this paper, we offer a set ofheuristics (rules of thumb) that addressa range ofchallengesassociatedwith this enterprise, ncludingthe selection of team members,negotiatinga con-sensus view of the researchproblem, prototyping

INTRODUCTIONIn the past 100 years, knowledge has become in-creasingly specialized. This specializationhas re-sulted in tremendous intellectualand technologicalgains,but it has also led to increasing ragmentationin the modern researchenterprise (Nissani 1997).Many of the important ssues in societysimplycan-not be addressedadequatelyby a singledisciplinaryperspective.This is particularlyapparentfor issueswith an environmentalcomponent, such as water-shed protection, sustainabledevelopment, and cli-mate change.These issues demand that we take anintegrated view; they are essentially systemsprob-lems. To addresssystems problems effectively re-

Received 27 April 2001; accepted 12 November 2001.*Correspondinguthor;e-mail: [email protected]

and refiningmodels, the role of sensitivityanalysis,and the importanceof team communication.Theseheuristics arose from a comparisonof our experi-ences with several interdisciplinary modelingprojects.We use one such experience-a project nwhich natural scientists,social scientists,and localresidentscame together to investigatethe sustain-abilityof smallindigenouscommunities n the Arc-tic-to illustratethe heuristics.Key Words: interdisciplinary;modeling; ecosys-tem; collaboration; ustainability;Arctic; ntegratedassessment;teamwork.

quires us to bridge perspectives and disciplines(Gundersonand others 1995; Parson 1995) anddeal with complex interactingprocessesthat oper-ate at differenttemporaland spatialscales (Holling1995;Likens1998). By integratingandsynthesizingknowledge from disparatedomains, the emergingfield of integratedassessment (IA) attemptsto ac-complishthis goal (Risbeyand others 1996).Within IA, simulation models are commonlyused for synthesizingdisciplinaryknowledge. Theyare by no means new tools for scientists (see, forexample,the work on modelingmarineecosystemsby Riley 1947), and since the early 1970s, the re-sults of such models have often been made acces-sible to the generalpublicas well (forexample, seethe much-publicized Limits to Growthstudy byMeadows and others forthe Clubof Rome in 1972).Integrated system models offer three extremely

376

http://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsp


3/10

Heuristics or InterdisciplinaryModelers 377useful advantagesfor interdisciplinary esearchers.First, systems models provide a way to codifyknowledge from differentdisciplines nto a unifiedand coherent framework.Second, they encouragefocused and disciplinedthinking about the causalrelationships in a system. Third, they allow re-searchers,ecosystemmanagers,and stakeholders oexplorehow their systemmay respondto a varietyof scenarios so that responses can be formulatedand management actions can be implemented.However, system models can only achieve theseadvantages f they are developedand used deliber-ately and thoughtfully.Developingsimulationmodelsis partscience andpartcraft; here are no general,infallible rules. Dif-ferentpractitionersprocesstheir experiencesin dif-ferentways. In thispaper,we offer 10 heuristicsforinterdisciplinarymodelingthat we have developedover a period of severalyears through our experi-ences in a variety of integratedresearchprojects.The primaryaudience we have in mind is peoplewho are not presentlyengagedin interdisciplinaryresearchbut are interested n moving in this direc-tion in the future.However,we also hope to stim-ulate thinking,discussing,andwritingaboutmeth-odologyamongcurrentmodeling practitioners,andwe believe that our emphasison rapidprototypingand sensitivityanalysiswill be of interest to them.

What do we mean by "heuristic"?Polya (1945)definedthis term as "the name of a certainbranchof study"whose aim is "to understand he methodsand rulesof discoveryand invention."However,inthis essay, the word is used in the sense defined byStarfieldand others (1994): "aheuristicis a plausi-ble or reasonableapproach hathas often provedtobe useful, a rule of thumb."In other words, this is a paperin which the find-ings have been generated nductivelyfromour col-lective experiences on a range of interdisciplinaryprojectsrather than a deductive literaturereviewthat investigates the success or failure of otherprojectsbased on whether they did or did not usethese heuristics.To illustrateour 10 heuristics,we give examplesof lessons we have learned from developing inte-gratedinterdisciplinarymodels for a recent projectinvestigating he Sustainability f ArcticCommuni-ties (SAC).Thisprojectinvolved a team of 25 sci-entists (representing eight different disciplines inboth the naturaland the social sciences) and resi-dents from four indigenousArcticcommunities inthe Yukon Territory,Northwest Territories,andAlaska.Research eam memberscame fromseveraluniversities and from government agencies. The

in climate, tourism, oil development, and govern-ment fundingcould affectthe sustainabilityof ourpartnercommunities.The communitiesthemselvesdefined their goals for sustainability n the earlypartof the project(G.P.Kofinasand othersunpub-lished). These goals included (a) maintaining astrongrelationshipwith the land and the animals,(b) developinghealthy mixed economies (thatis, asubsistenceharvestingeconomy in parallelwith acash economy), (c) exercising local control overland use and resourceuse in their homelands, (d)educating their young people in both traditionalknowledgeand Western science while also educat-ing outsidersabout their way of life; and (e) main-taining a thriving native culture (evidenced, forexample,by the use of indigenouslanguage,respectfor community elders, and spending time on theland). In other words, the communities saw sus-tainabilitynot simply in terms of sustainablere-sourceuse, but also in economic,political,and so-ciocultural terms. To address this holistic set ofcommunity goals, it was obviouslyessential to takean interdisciplinaryview of the system. An inte-grated approachwas in any case implicit in theframingof the originalprojectproposaland in therange of disciplinaryscientistsincluded in the re-searchteam. Their expertise covered the fields ofvegetation ecology,cariboubiology,cariboubehav-ior, household economies, culturalecology, socialanthropology,policy analysis,Arctictourism, andnatural resourcemodeling.The emphasis of this paper is not on the SACProject itself, although examples will be drawnfrom that projectto illustrateour heuristics.Also,the heuristicsgiven here relate primarily o scien-tists workingwith other scientistson interdiscipli-naryprojectsratherthan to scientistsworkingwithstakeholding.The SACProjectnot only served tobringscientists ogether,but alsoinvolvedresidentsof indigenous Arctic communities. A companionpaperto this one (G.P.Kofinas and others unpub-lished) offersheuristicsfor researcher-stakeholderinteractionsand for synthesizinglocal knowledgeand science. Finally, although we discuss variousaspectsof teamworkand collaboration,our focusisnot on collaborationgenerally (as in, for example,Gray1985, 1991 or Kofinas and Griggs1996) buton the processof the collaborativedevelopmentofsynthesismodels.Heuristic.Knowwhatskills o look orwhenrecruit-ing an interdisciplinaryeam. It is not a foregoneconclusion that any given team of specialistswillwork togethereffectivelyto producea tightly inte-gratedview of a system. Indeed, there are manygoal of the projectwas to investigatehow changes challenges and obstacles that must be addressed



4/10

378 C. R. Nicolson and othersbefore a variety of scientists can work together ef-fectively in an interdisciplinary mode. Among theseobstacles is the problem of cross-discipline commu-nication, since specialists are used to interactingwith peers from within their fields who share acommon view of the issues and a common lan-guage for discussing them. The problem of commu-nication will be addressed in heuristic 9, but it isalso relevant here because it can be a stumblingblock when choosing and recruiting team members.To foster good communication among prospectiveteam members even at the recruitment stage, it isessential to develop a prototype conceptual modelof the system (see heuristic 3).A major obstacle to interdisciplinary work is thatscientists are trained and socialized from their grad-uate school days to focus on narrow, tractable prob-lems within clearly defined boundaries. They aretaught how to identify problems that lie on thecutting edge of their discipline, and they learn ap-propriate methods for solving these problems. Inother words, using Holling's (1996) distinction be-tween the science of parts and the science of theintegration of parts, scientific training is essentiallyan induction into the methods and norms of thescience of parts.In the science of parts, investigatorswithin a discipline focus on a narrowly definedquestion with the goal of reducing uncertainty tothe point of consensus. In contrast, the science ofthe integrationofpartscalls for people who are com-mitted to studying a complex system by focusingnot so much on the individual components of thesystem as on the interrelationships among its com-ponents. Although it is often true that outstandinginterdisciplinarians also have very high reputationswithin a specialist field, the best disciplinary mindsare not necessarily the best interdisciplinary teammembers. Interdisciplinary projects are intellectu-ally demanding in a different way from classic re-ductionist science, and they need at least some big-picture researchers who will creatively explore thelinkages and interfaces between their own disci-pline and other fields of inquiry in which they maythemselves have no special expertise.Another key attribute of a good interdisciplinaryteam member is the ability to simplify what isknown and, when necessary, guess at the unknown(see heuristic 8). These activities call for people witha deep grasp of their own disciplines: Weak or in-secure disciplinary minds can frustrate teamprogress by refusing to explore linkages, to simplifytheir field, or to guess at unknown factors. Not onlyare these activities necessary to make an interdisci-plinary study a success, but they are often not the

will recognize as valuable contributions to thescholarship of their field, and publishing their workmay not be easy. Young scientists are particularly atrisk because they have not yet established theirreputation and because the reward systems of aca-demia bend to favor disciplinary specialists.All team members who embark on interdiscipli-nary projects need to be made aware of these kindsof problems in advance so that they join the teamwith realistic expectations, an adventurous attitude,and a willingness to work at cross-disciplinary com-munication. For a project leader to know if some-one is right for the team, he or she should look forscholars who can see the big picture, whose trackrecord shows an ability to work with people outsidetheir own discipline, who are good listeners, andwhose interest in a problem outweighs their con-cern for career advancement! How can such peoplebe motivated to participate? One incentive maysimply be an appeal the intellectual satisfaction ofseeing how their disciplinary interests fit within alarger framework, thereby sharpening their under-standing of their own disciplines.Heuristic2. Investstrongly n problemdefinition earlyin the project. By their very nature, projects thatinvolve complex systems with many interactingcomponents lend themselves to multiple focuses.Different stakeholders often have different percep-tions of the problem. In addition, each disciplinaryexpert has a vested professional interest in definingthe problem so as to give his or her discipline aprominent role with a bias toward the researcher'sparticular expertise within the discipline. In theabsence of strong leadership, it is far easier to endup doing multidisciplinary research (where expertswork in parallel with each other without muchmeaningful integration) than it is to do truly inter-disciplinary research. To address this challenge, theproblem needs to be thoughtfully and clearly de-fined from the outset. This is never a straightfor-ward exercise, even when there are compelling rea-sons for the study. The parties involved in problemdefinition need to understand that choosing thefocus of an IA project is fundamentally a negotiatedprocess. For this reason, all the parties (disciplinaryresearchers, stakeholders, and funding agencies)must be given opportunities to exchange perspec-tives and must be aware of each other's priorities.Heuristics 1 and 2 are obviously interrelated. Un-til you define the problem, you cannot assemble ateam; and until you have a team, you cannot reallydefine the problem. (This is why we promoted theidea of developing a prototype conceptual model,even at the stage of recruiting team members.) Thetypes of activities that their own disciplinary peers ideal situation is one in which a small group has the



5/10

Heuristics or InterdisciplinaryModelers 379opportunity o make an initialattemptto define theproblem and then go on to recruitthe additionalexpertise necessary. This kind of opportunity re-quires either project development funding or aninfrastructure hat bringsthe smallgrouptogether.

In the Sustainability of Arctic Communitiesstudy, the High Latitude Ecosystems Directorate(HLED)of the US Man and the Biosphere (MAB)programprovided an opportunity for a group ofnatural and social scientists to interact with eachother across disciplinesand to formulate a roughandpreliminaryprojectdefinition.TheHLED roupof six individualsstarteddiscussions2 yearsbeforethe fundingopportunityarose and decidedto focuson the combined effects of future climate changeand oil development on barren ground caribou(Rangiferarandus)nd the indigenouscommunitiesthatdependon caribouasa subsistenceresource.Atthis initialstage,the stakeholdercommunitieswerenot directly involved. This was a mistake, eventhough several group scientists had worked withindigenouscommunitiesformany yearsand there-forehad a goodgraspof the issues involved. Onthebasis of the preliminary problem definition, thegroupobtainedpermissionfromthe US MABcom-mittee to advertiseposition descriptionsfor addi-tional HLEDmembers with expertise in culturalecology, modeling, and caribou biology. As newpeople were recruited,they broughtnew perspec-tives on the problem,and the processof negotiatinga common focus continued. The stakeholder com-munities joined the project in the 1st year andprovidedan importantrealitycheck on our under-standingof the issues (G.P.Kofinasand others un-published).It is extremelydifficult o anticipate he appropri-ate problemdefinition at the outset of a project.Infact, duringthe early part of the SACproject,theparticipating cientistsfelt that the targetwas for-ever shifting.In hindsight,the team ought to havebuiltmore structure nto the negotiationprocesstoensure convergenceon the problemdefinition. Topromote clarity of thought and allow the groupmembersto see whether the initialproblemdefini-tion is correct,we recommendthe following struc-tured procedures:1. Cooperateon the development of first proto-type "straw" ystem simulationmodels, and2. Submit the currentunderstandingof the sys-tem to a "peerreview" by stakeholders. Thediscipline of having to articulatethe problemdefinition to an audiencebeyondthe team itself

Heuristic . Use rapidprototypingor all modelingefforts. Not only is it hard to define the problemcorrectlyon the first attempt, it is also extremelydifficultat the start of a new projectto discerntherelative importance of each of the components.Therefore,rapidprototypingof models is essential.Instead of trying to specify at the outset of theprojectpreciselywhat the finalmodel will look likeand what questionsit will address,the participantsshouldrecognizethat the firstyear will be devotedto the developmentof a prototypemodel aimed atclarifying he objectivesof the study. Moreover,insubsequentyears, the problemand the model willbe further refined through successive prototypes(see Schrage2000 fora number of casestudies fromthe businessworld in which prototypingand suc-cessiverefinementconsistently ed to superior inalproducts).It is only when projectparticipantsseeactualoutputfromthe model thatthey canbegintograspthe big pictureand gain an understandingofthe systemdynamicsas a whole. Thisunderstand-ing allowsthem to placetheir own contributions nperspective.Furthermore, t is only when a proto-type model is up and running that the relativeimportanceof the various components of the sys-tem or weaknesses in the framing of the originalhypothesesgraduallybegin to emerge.For example, the originalproposal for the SACProjectenvisioned a model time horizon of 100-200 years. The revised proposal (with added em-phasis on the Arcticcommunities) defined a timehorizon of 40 years.One of the originalhypotheseswas that climate change would lead to changes insummer vegetationbiomassand plant communitycomposition,that caribouherds would be affectedby these changes, and that Arcticcommunities, inturn, would be impactedby the caribou. On thebasis of this definition of the problem, a set ofsustainability ndicatorswas developed; these in-cluded plant biomass, caribou herd size, hunters'time-on-the-land, and seasonal caribou harvest.We designeda synthesismodel that would addressthe problemas it had been defined.However,in theprocess of developing and testing the model, wediscovered hat there aretime lags of 50-100 yearsbefore any substantial simulated effects of climatechange are apparentat a plant community or bio-mass level (Epsteinand others 2000); within a 40-year time horizon, climate-related vegetationchanges were therefore almost insignificant. Wealso learned from the initialmodeling exercisethatas long as the caribouherd size is above a certainthreshold(estimated o be about60% of its presentlevel), annual cariboumigrationpatternsaffecthar-vest successfar more than a decline or increase inelps to get the ideas clear.



6/10

380 C. R. Nicolson and others

herd size. This suggests that the initial emphasis onherd population dynamics may have been some-what misplaced. In both of these examples, ourproblem definition led us to believe that certainfactors were more important than they turned outto be.It is also important to make sure that the linkagesamong different parts of the system are strong andthat the system behavior is not dominated by asingle component (in which case the problem doesnot necessarily call for an interdisciplinary ap-proach). Our experience, and that of other practi-tioners (for example, Holling 1978; Walters 1986),has shown that it is more fruitful to begin with thesystem itself and to look outward to the compo-nents rather than to look piecemeal at the systemfrom within the perspective of the individual com-ponents.One danger in interdisciplinary modeling work isthat people who are not fluent in systems modelingmay not engage properly with the task. The solu-tion is not to recruit only model-oriented scientists(which would limit the scope and breadth of thesynthesis), but rather to work at drawing nonmod-elers into the process. Prototype models that aresimple enough to demonstrate and explain to allteam members are an essential step in the educa-tion of nonmodelers.

Heuristic 4. Allow the project'socus to evolveby notallocating allfunds upfront. This is a luxury seldomavailable to research scientists, given the currentpolicy of multiyear, multi-investigator projects.However, one of the inherent difficulties with in-terdisciplinary research is that defining the problemoften represents a major part of the project. Thus, achicken-and-egg situation arises. The problem can-not be defined until a working team is in place, butit is impossible to know how deeply to involvespecific team members until the problem has beendefined. Even when the problem is apparently welldefined, it is extremely hard to assess a priori whichcomponents determine the system dynamics moststrongly until a first prototype of the synthesis workhas been constructed. It is likely that the relativeimportance of the various components will onlyemerge during the study. We have already alludedto the initial hypothesis of climate change -> vege-tation change -> caribou herd dynamics -> caribouavailability to human communities. By the time wediscovered that this apparently central hypothesiswas not a main driver of change, the project's fundshad been allocated and could not easily be shifted toaddress newly evolving hypotheses.It might be better if funding agencies awarded

through the development of a first prototypemodel) and then funded the remainder of theproject only when it was demonstrated that thecorrect mix of scientists was working together ef-fectively and attacking a well-defined problem. If allthe funds are committed up front for the full dura-tion of the study, the project leadership has noflexibility to add new people as their expertise be-comes necessary or to reallocate funds from a com-ponent of the work that offers little to the inte-grated effort.Heuristic5. Ban all models or model componentshatare inscrutable. An "inscrutable" model is a blackbox in which the inner workings are inaccessible toall but the original developers. The user is requiredto take the output on faith. The problem with in-scrutable models is that people have no incentive toengage with them intellectually. If the model pro-duces any counterintuitive results, people cannotaccess the logic that led to those results. It is notsurprising then that their usual reaction is to losetrust in the model rather than ask about the inter-mediate relationships that led to those final results.In the SAC Project, a complex model of caribouenergetics (Hovey and others 1989; Kremsater1991; Daniel 1993) was initially thought to be es-sential at the interface between vegetation changeand caribou population dynamics. We realized laterthat what we really needed were models of herddistribution and movement, but a commitment hadalready been made to this energetics model. Untilwe developed a much simpler caribou populationmodel, the project depended on output from a blackbox model that only a few people understood andused. A top-down, rapid-prototyping approachcould have helped avoid this situation. Graphical"box and arrow" representations of the system (Jor-gensen 1986; Walters 1986) combined with thesimplest possible component models, programmedusing software that is easily accessible to all teammembers (such as spreadsheets), allow a team ofscientists from different disciplinary backgrounds tounderstand and engage with the key relationshipsof the model.

Heuristic6. Insteadofconcentratingn oneall-purposesynthesismodel, investin a suite of models,each with awell-definedobjective. This heuristic applies particu-larly to the collaborative development stage of aproject. It allows participants from a subset of dis-ciplines to engage with models that focus on theinterfaces between those subsets.The idea of building a suite of models may seemto go against the very idea of interdisciplinary syn-thesis modeling, but meshing existing submodelspreliminary planning funds (say, for the 1st year or together can be a difficult and time-consuming ex-



7/10

Heuristics or InterdisciplinaryModelers 381ercise. Submodels may operate at different timescales because of the nature of the underlyingpro-cesses. Similarvariables n two submodelsmay berepresentedat different levels of detail from oneanother. The probabilisticoutcomes produced byone stochastic submodel may not translate easilyinto hard-and-fastnput values for other determin-istic submodelsin the system.Although it is essential to represent adequatelythe logicandthe resultsof each submodelin all theother relevantsubmodelsto which it links, if this isdone properly, t may not be necessaryto have one"supersynthesis"model that runs each submodelwithin the same overallprogramming ramework.In fact, for quality-controlpurposes, it is probablygood to have some kind of human interface be-tween submodels.This allows the results of eachsubmodel to be assessed and the qualityof its con-clusions evaluated, so it can be determined howbest to includethe insightsgainedfrom each modelin the next submodel. Thisprocesshelps to deter-mine which details are not essentialand allows thedevelopment of a "boiled down" whole systemmodel at a level of abstractism hat may initiallyhave been unacceptable o some team members.A furtherargument for a suite of models is thatfew users of the overall synthesis model will beinterested in all of its components. Most peoplehave an interest in only three or four of the out-comes of the model. A suite of models allows usersto examine the components with which they arefamiliar and to see how these results fit with theoutcomes they expect, based on their knowledgeandexperienceof thatpartof the system.However,for nonscientificusers, it is helpful to have a seam-less interface hatallows them to explorewhicheverpartof the systemthey are most interestedin. Ifnosuch interfaceexists, users will not readily recog-nize that they are seeing an integratedview of thesystem,and much of the benefit of the exercise willbe lost.Heuristic . Maintaina healthybalancebetween hewell-understoodnd thepoorlyunderstoodomponentsfthe system. All system models are balancing actsbetween what one knows and understands andwhat one does not know. The temptationis to puttoo much emphasis on those parts of the systemwhere understandingand data are good and toignoreorglossoverthe areas where little is known.This is not surprising,given the way in which thescientific enterprisetends to favor specialists.Forexample, even though there may be a clear andobvious linkbetween cariboumigrationand house-hold economic production, a caribou biologist

and energetics,but relativelylittle about herd mi-gration patterns.An economist may have a goodunderstandingof the factors hat influence people'sdecisionsto take wage employment,but know rel-atively little about the factorsthat account for thesuccessfulharvestand productionof cariboumeaton the land itself.Furthermore,people like to concentrate on thedetails they know about and understand (Likens1998). In particular, cientistsare socialized nto anepistemological ramework hatplacesa high valueon detailedquantitativehard facts and tends to takea dim view of uncertainty(even when the uncer-tainty involves an educatedguess in an importantarea where little else is known). They are oftenskepticalof simplificationand even more uncom-fortablewith the idea of makingeducatedguesses.Forexample,the village economy model we devel-oped for the SACProjectcontainedover 90 differ-ent job categories,each characterizedn termsof itsrequirededucation level, its seasonal availability,andwhethermen or women were more likelyto befound in that position. These definitions werefirmlygrounded n surveydata.However,the econ-omists on our team were reluctant to speculateonhow these definitionsmight evolve over the next40 years;as a consequence, the model containedthe implicitassumptionthat social norms such asgender preferences orjob types would not changeduringtwo generations.Maintaining he balancebetween the known andthe unknown requiresstrongprojectleadership.Ina review of IA projects,Parson (1995) observed:"Since researchersworking within their fields donot normally attend to borders of other fields,achievingthis attentionshiftrequiressome form ofauthority in an assessment project, or at least acoordinatingmechanismand a common languageforcommunicatingacrossboundaries."One way toachieve such coordinationwould be to bringin anoutside modeling consultant who could facilitatekey workshops. In addition to providing a freshviewpoint, an outsider who has the trust of theteam could also providethe kind of authoritythatParsonrefers o. Modelersarecertainlynot the onlypeople who could fulfill this role. The two mostimportantqualificationsare an abilityto see the bigpicture and the earned trust and respect of otherteammembers.Thesequalitiesmay well be presentin one of the disciplinary pecialists f he or she isalso a good big-picturescholar.However, throughrapidprototypingandsensitivity analysis,modelingcan be particularlyuseful for ranking the relativeimportanceof the partsandprocesses n the model,might know a great deal about caribou activities as well as making a rapid assessment of the value



8/10

382 C. R. Nicolson and othersand differences between alternative conjecturesabout the unknowns. The synthesis modelersshould be encouragedand empowered to use theirskills to help resolve the tensions between simplic-ity and detail that are inherent to any modelingproject(Costanzaand Sklar1985;StarfieldandBle-loch 1991).Heuristic . Sensitivity nalysis s vitalat all stagesofthe modelingeffort. Thorough sensitivity analysisinvolves testingnot only differentparametervaluesbut also the assumptionsand the effect of alterna-tive educated guesses at the underlying processes(see, for example, Starfieldand others 1995; Star-field and Bleloch 1991). Sensitivity analysis is theonly availablemeans of determiningwhat goes intothe model and what level of detail is necessary.It isan essential tool for estimatingthe likely effects ofalternativehypotheses for system processes. Sensi-tivity analysisshould not simplybe thoughtof as anautomated process that tests all parameters,butrather an importantpartof the culture of modelingthat is used for the thoughtful explorationof alter-native assumptions.It follows that the work of sen-sitivity analysisshould be done by most (ideally,all)of the projectteam, not just the modeler. Becauseeach personon the teambringsa differentperspec-tive to the problem,he or she is thus likely to rundifferentexperimentsand uncover differentprob-lems. In fact, team efforts are essential both foridentifying implicit assumptions (social norms donot change during two generations, for example)and fordeveloping plausiblealternativescenariosaspartof the sensitivityanalysis.Sensitivitytests are essentiallymini-experiments.To be effective in shapingthe prototype modelingprocess, the models supporting the mini-experi-ments need to run virtuallyin real time. Waitingdays, weeks, or months for model results is toolong. We found that the abilityto work as a groupto set up a model simulation, and then view theresults within a minute or two, was principallyresponsiblefor most advances in developingmodelrelationshipsthat crosseddisciplinaryboundaries.Heuristic . Workhardat communicationndbudgetforface-to-facemeetings. ffectivecommunicationliesat the heartof interdisciplinaryesearch.Not only isit necessary for scientists to engage with one an-other to produce an integratedview of a system,their findings must also be explained clearly tostakeholdersand to the public.In the Internet era,communication can take many forms, includinglist-servermemos, e-mails, phone calls, small face-to-face work groups, plenary team meetings, andpublic meetings. Each communication medium

assume that simplybecause we have these tools atour disposal, people from different disciplinarybackgroundswill automaticallycommunicateeffec-tivelywith each other. In the SACProject,scientistsoften appearedto have reached a point of under-standingin their discussions,only to find out laterthat in fact they had two ratherdifferentthings inmind (sometimes as the result of using the samewords but meaningdifferentthings by them). Teammembers need to make an effort to become morefamiliarwith each other's mental frameworksandto be cognizantof what specific people mean whenthey use certain words or concepts. This is onereasonwhy rapidprototyping s so valuable. Itleadsquicklyto a productthat providesa common lan-guageandenablesparticipantso say "No, hat'snotreallywhat I have in mind."One way to foster better communication is bydeveloping simulation models in easily accessiblemodeling environments,such as spreadsheets.Thegoal is to work continually toward a culture oftransparentand accessiblemodels, so as to ensurethat the models areunderstandable o everyone onthe team. In this regard,we have found thatspread-sheetshave severaladvantagesover traditionalpro-gramming anguagessuch as FORTRAN, ASIC,orC++. Most scientists are familiarwith the spread-sheet environment and its basic concepts. Also,spreadsheetsallow us to quicklyand easily developstraw models as part of the dialogue among theparticipants,so we can constantly point to some-thing tangible and ask, "Isthis what you mean?"The built-in graphing unctionsof spreadsheetsen-able graphicalmodel output with very little pro-grammingeffort.Finally,becausespreadsheetsper-form calculationseach time a cell is changed,theyare powerful tools for sensitivity analysis.In addition to communicationwithin the team, asecond arearequiringcarefulconsideration s howthe integratedwork of the team will be communi-cated to the stakeholdersand the public,a taskthatis vastly underrepresented in many scientificprojects.Interdisciplinaryesearchthat affectspeo-ple's lives directly must be explained to them inaccessiblelanguage, strippedof its technical scien-tific terminology. The results need to be put intoeveryday erms,and it is crucial o spellout boththepractical mplicationsof the findingsand the areasof uncertainty.Funding agenciesneed to be willingto supportthe outreach and extension part of in-terdisciplinaryesearch,andscientistsneed the helpof communicationspecialists o get theirresults ntopublicdiscourse n a formthat can be digestedanddiscussed.The SACProject'seffortsat outreach in-serves a differentpurpose, and it is dangerousto clude the development of a simplified interactive



9/10

Heuristics or InterdisciplinaryModelers 383Web-basedmodel interfacethat we call the "Possi-ble Futures Model" (G.P.Kofinas and others un-published). The model demonstratesour attemptsat ongoing innovation in all areas of the interfacebetween model and user: ease of use, hypertextdocumentation,graphicaloutput, and built-in fea-turesforexplainingmodelresults and documentingusers' feedbackcomments.An important esson from the SACProject s thatthe budget for face-to-face meetings was inade-quate (both for meetings among researchersandmeetings between researchers and communitypartners).The proposaldid include funds for an-nualprojectmeetingsof the entireteam;these wereof some value, but we found it far more profitableto hold work sessions involving small numbers ofresearchers from the various components to de-velop specificcomponentlinkages.E-mail is not aneffective medium for planningand for the creativegeneration of ideas. Writtenexchanges work bestonce there is a commonunderstandingof the prob-lem, common assumptions,and a negotiatedset oftask assignments;face-to-face meetings are indis-pensable for these groundworkdecisions. Becauseface-to-face meetings are so much richer in com-municativecontentand becausethey allow trusttobe built more easilythan can be done in a series ofwritten messages (Daftand Huber1987), meetingswith componentresearchersare also criticalto thework of the synthesismodeler.When we began tohold these meetings, considerablemomentum wasgained. Face-to-face contact should be a nonnego-tiablepartof anyIAbudget, particularlywhen teammembers are geographicallydispersed.Heuristic10. Approach he projectwith humility.Even though the scientists on the team may beworld-classexperts in their respective componentfields, they are all likely to be amateurs when itcomes to the systemas a whole. It is worth remem-bering that a distinguishedgroup of componentexperts does not guaranteea distinguished systemteam. In fact,since laypeopleoften have a deep andholistic understandingof their local environment,we scientists may be no more "expert" than theyare,even though theirknowledgeis not necessarilyscientific.All team membersmust take the time toprobe and queryeach other'sapproaches,assump-tions, and methods.Moreimportant,they must bewilling to have their own assumptionsand state-ments probed by others. This requireshumility, awillingnessto be challengedby team membersout-side one's own area, and an openness to learningfrom such transactions.The excitement and chal-

ing together the unknown-namely, the behaviorof the system.Humility and caution are especially importantwhen scientistswork on projectsthat are intendedto inform policy, thereby affecting people's lives.Synthesismodelersbear the brunt of the responsi-bilityof ensuring,first,that the assumptionsbehindthe models are carefully spelled out and, second,thatrobustconclusionsare shown to be robustevenin the face of uncertainty.In the spiritof humility,we acknowledgethatthe10 heuristicspresentedhere are obviously not ex-haustive. Integratedassessment is a very compli-catedbusiness,and as a form of inquiry,it is stillinthe early stages of development. Not only are thedynamicsof largecomplex systemshard to under-stand,but the challenge of bringingdisparateper-spectives together is a formidable one. We offerthese principlessimplybecausewe believe it is im-portantforsynthesismodelersand interdisciplinar-ians alike to reflecton what they have done, in thehope of doing it better the next time around.ACKNOWLEDGMENTSFinancial upportwas providedby the NationalSci-ence Foundation (NSF OPP-95-21459). We ac-knowledgethe role playedby members of the Sus-tainability of Arctic Communities research teamand the communities of Old Crow, Aklavik, FortMcPherson,and ArcticVillagein our collaborativeeffort to builda sharedunderstandingof a complexsystem. Ann Kinzig and an anonymous reviewerofferedhelpful commentson a previous version ofthe manuscript.REFERENCESCostanzaR,SklarFH. 1985.Articulation, ccuracyandeffective-ness of mathematicalmodels: a review of freshwaterwetland

applications.EcolModel27:45-68.DaftRL,HuberGP. 1987.Howorganizationsearn: a communi-cation framework.SociolOrg5:1-36.DanielCJ,1993.Computer imulationmodelsforthe PorcupineCaribouHerd:user'sguide. Richmond Hill: (Ontario):ESSALtd., for CanadianWildlife Service, Environment Canada,Whitehorse,Yukon,17 p.

EpsteinHE,WalkerM, ChapinFS,StarfieldAM,2000. A tran-sient, nutrient-basedmodel of arctic plant community re-sponseto climaticwarming.EcolAppl 10:824-41.GrayB. 1991. Collaborating:indingcommon groundfor mul-tipartyproblems.SanFrancisco: ossey-Bass.GrayB. 1985.Conditionsacilitatingnter-organizationalollab-oration. HumRel 38(10):911-36.GundersonLH,HollingCS,LightSS,editors.1995. Barriers nd

bridgesto the renewal of ecosystemsand institutions. NewYork:ColumbiaUniversityPress. 593 p.HollingCS.1978.Adaptiveenvironmentalassessmentandman-lenge of interdisciplinary research lies in uncover- agement. London: Wiley.



10/10

384 C. R. Nicolson and others

HollingCS. 1996. Surprise orscience,resilienceforecosystemsand incentives forpeople.EcolAppl 6(3):733-5.HollingCS. 1998.Two culturesof ecology.ConservEcol[online]2(2):4. Availableon the Internet:URL:http://www.consecol-.org/vol2/iss2/art4.HollingCS. 1995. Whatbarriers?Whatbridges? n: Gunderson

LH,HollingCS,LightSS, editors. Barriersand bridgesto therenewalof ecosystemsand institutions. New York:ColumbiaUniversityPress.p 3-34.Hovey FW, KremsaterLL,White RG,RussellDE, Bunnell FL,1989. Computer imulationmodels of the PorcupineCaribouHerd.II. Growth.Technicalreportno. 54. CanadianWildlifeService,PacificandYukonRegion,VancouverB.C.J0rgensenSE.1986.Fundamentalsof ecologicalmodeling.Am-sterdam:Elsevier.389 p.KofinasGP,BraundSR,ArchieB, CharlieJ Sr,EamerJ, JamesS, KruseK,NicolsonC,TetlichiJ. Integratingocalknowledgein an integratedassessment:essonsfromthe Sustainability fArcticCommunitiesProjectexperience.Unpublished.KofinasGP,GriggsJR.1996. Collaboration nd the B.C.RoundTableon the Environmentandthe Economy:an analysisof a"betterway"of deciding.Environments23(2):17-40.KremsaterLL.1991. Briefdescriptionof computersimulationmodels of the PorcupineCaribouHerd.In: Butler CE,Ma-honey ST, editors.Proceedingsof the 4th North AmericanCaribouWorkshop.St. Johns (Newfoundland):p 299-314.LikensGE.1998. Limitationso intellectualprogress n ecosys-tem science.In:PaceM, GroffmanP, editors.Successes, imi-tationsand frontiers n ecosystemscience.New York:Spring-er-Verlag.p 247-71.

MeadowsDH, MeadowsDL, RandersJ, BehremsWW. 1972.Limits o growth,New York:Universe Books. 205 p.NissaniM. 1997. Ten cheers for interdisciplinarity:he case forinterdisciplinary nowledge and research. Soc Sci J 34(2):210-6.ParsonE. 1995.Integrated ssessmentandenvironmentalpolicy

making. EnergyPolicy23(4/5):463-75.PolyaG. 1945. How to solve it. Princeton(NJ):PrincetonUni-versityPress.253 p.RileyGA.1947. A theoreticalanalysisof the zooplanktonpop-ulation of GeorgesBank. J Mar Res6(2):104-13. (Quoted n:SwartzmanGL, Kaluzny SP. 1987. Ecological simulationprimer.New York:Macmillan.370 p).RisbeyJ, KandlikarM, Dowlatabadi H. 1996. Learningfromintegratedassessmentof climate change. Clim Change 34:369-95.SchrageM. 2000. Seriousplay:how the world's best companiessimulateto innovate.Boston: HarvardBusinessSchool Press.244 p.StarfieldAM,BlelochAL. 1991. Buildingmodels for conserva-tion and wildlifemanagement.2nd ed. Edina(MN):Interac-tion. 252 p.StarfieldAM, Roth JD, RailsK. 1995. "Mobbing"n Hawaiianmonk seals (Monachuschauinslani):he value of simulationmodelingin the absence of apparentlycrucialdata. ConservBiol 9(1):166-74.StarfieldAM, Smith KA, Bleloch AL. 1994. How to model it:problemsolvingfor the computerage. 2nd ed. Edina (MN):Interaction.206 p.Walters CJ. 1986. Adaptive management of renewable re-sources.New York:Macmillan.374 p.

Thi t t d l d d f 148 206 159 132 M 23 S 2013 01 42 59 AM
http://www.jstor.org/page/info/about/policies/terms.jsp

Nicolson_10 Heuristics for Interdisciplinary Modeling Projects

Documents