JOEL M. SMITHJARED L. COHON

Manaqing theDigital Ecosystem

Administrators must lead their universities intoa future in which every constituency has distinct needsand every decision has implications for all

Benjamin Disraeli's ironic comment, "Imust follow the people. Am I not theirleader?" aptly describes the feelings ofmany college and university administra-tors as they develop institutional plansfor information technology (IT) that willsupport research, teaching, and learning

in the coming decades. The context in which we are expectedto lead our institutions in IT decisions has changed dramat-ically. We are experiencing unprecedented technologicalchange emerging from a much greater diversity of sourcesthan ever before. Students, faculty, and staff arrive at the begin-ning of each school year with new ideas (and associatedhardware and software) for using information technolo-gies that will enable them to accomplish their diverse goals—educational, professional, and personal. Technology firmsalong with increasingly influential open-source softwaredevelopment efforts present us with a staggering numberof technologies that hold promise for enabling our funda-

mental missions of creating and transferring knowledge. Whatleadership strategies are appropriate in such a complex,dynamic, and unpredictable context?

We've gone through a qualitative rather than merely a quan-titative transition in the nature of IT. We have recentlybecome accustomed to thinking of IT as one of the manyinfrastructures we provide on campus and for the nationalacademic community. But the complexity and dynamismof IT, especially in academe, now warrants thinking aboutit in different terms: as an IT ecosystem. Computing inhigher education has evolved from islands of innovation,to activities that depend on campuswide and worldwideinfrastructures, to an ecosystem with many niches of exper-imentation and resulting innovation. These niches are filledby faculty and students who do what they do with IT fol-lowing whatever motivations they may have, undirected bya central authority. But, as in any ecosystem, they are con-nected to the whole and often depend on a set of core tech-nical, physical, and social services to survive. Many inno-

56 ISSUES IN SCIENCE ANDTECHNOLOGY

MANAGING CHANGE

vations depend on services—networking, authenticationand authorization mechanisms, directories, communica-tion protocols, domain-naming services, global time serv-ices, software licensing, etc.—becoming widely availablebeyond the niche in which they develop. A simple exampleis the dependence of almost any IT innovation that uses thenetwork to share information on the "domain naming sys-tem" to find the devices with which it needs to communi-cate. Managing ecosystems calls for a very different set ofstrategies than those used to administer islands of innova-tion or the more static, top-down services that character-ize many infrastructures.

Academic computing ecosystems, especially (but notexclusively) at research universities, are tremendously com-plex, and understanding this complexity is crucial for select-ing effective management strategies. These ecosystems evolveand change, often very quickly, as innovations developedby faculty and students are absorbed by the ecosystem.Unlike other campus infrastructural systems, IT systemshave important feedback loops. For example, successfulimplementations of "course management systems" (Web-basedapplications for posting syllabi, readings, assignments, etc.)require continuous communication between those provid-ing the central service and the faculty and students using it.The result is evolutionary modification, often at a rapidpace, of both the technological tool itself and the ways in whichthe faculty choose to use it. If this mutual modificationthrough feedback is ineffective, the tool will quickly becomeirrelevant and die. Many on-campus innovations rapidlybecome "nece.ssities" and redefine the nature of the coreservices needed to sustain them. Unlike natural ecosystems,the IT ecosystem cannot evolve without some external man-agement. Nevertheless, top-down planning, uninformed bythe diverse niches of use and innovation, will not work.

Leadership evolutionTwenty-five years ago computers were relatively large, rela-tively rare, and used for computation. The communities atthe university interested in computing were small, isolated,and largely self-sufficient. Institutional leadership and plan-ning entailed helping these communities get resources toacquire the next fastest mainframe or the newest micro- orminicomputer for word processing or data acquisition anddata processing.

By 1990, personal computers (PCs) were ubiquitous. Eachhad computing power that exceeded most of the mainframesof the previous decade, and some of them were networkedtogether and to the NSFnet. IT planning had become morecomplex, but the shape of leadership needed from the cen-

tral administrations and the associated strategies were fairlyclear: Find resources to build campus networks; invest in v̂ ade-area networking for research; fund opportunities for fac-ulty to experiment with using PCs for research, teaching, andlearning; and hire staff to look after the support needs of thoseusing computing for a Viide variety of purposes. In short, buildand sustain an IT infrastructure for our individual institu-tions and for higher education as a whole.

Over the past 15 years, computers have become useful forcommunication, new forms of knowledge representation,knowledge management, visualizing complex data, cus-tomer relations management, music storage and transfer, videoediting, gaming of all kinds, grid computing, and some-thing else new seemingly every day. The user community nowincludes everyone on campus. IT tools have become essen-tial utilities without which we cannot function. At the sametime, they are a defining force in shaping the future of ourcore missions of knowledge creation and transfer. Further-more, the cheap microprocessor and the Internet created atipping point sometime during this period, when the hubof innovation moved from a small core of experts to a vastnumber of users.

By the turn of the 21st century, the flow of IT innovationswas no longer largely from the university to the students. Exceptperhaps for supplying the bandwidth for Internet connec-tions and educational discounts on expensive software, manycolleges and universities have little to offer to their studentswho arrive with computers (often more than one), cellphones, personal digital assistants, iPods, digital cameras,Sony PlayStations, Xboxes, LCD TVs, Bluetooth-enableddevices, e-mail accounts, personal Web pages, blogs, andhigh expectations for the role IT will play in their education.

There is a similar story regarding faculty, not just in sci-ence and engineering but in all disciplines. They depend ona wide variety of software, operating systems, computer con-figurations (including the emerging small supercomputersknown as "compute clusters"), multiple servers under theircontrol, and access to all of their resources at any time fromany place in the world. They are increasingly involved incross-institutional research groups that require everythingfrom sharing large data sets to remote digital access to super-computers and state-of-the-art research instruments.

For students and faculty, the number and kinds of infor-mation technologies they expect the university to supportis more diverse than ever before and more rapidly chang-ing than most would have imagined when universities begantheir commitment to IT as a fundamental infrastructure. Doc-uments such as the National Science Foundation reportRevolutionizing Science and Engineering through Cyberin-

FALL2005 57

frastructure highlight some of the issues. For example, fac-ulty routinely work with colleagues at distant institutions,depending on e mail, instant messaging, wikis, videocon-ferencing, file-sharing protocols, cybertrust mechanisms,and many other IT tools. Grid computing (the use of manycomputers working together to solve large-scale problems)will require not only high-bandwidth network connectionsbut data architectures and authentication and authorizationprotocols that ensure the coordination and validity of thecalculations and data.

The management challengeAlthough many have noted the need to adapt to the chang-ing IT environment, few have offered clear suggestionsabout how university leadership must change to deal withthe complexity of the situation. We believe that some unusualmanagement strategies are needed, and some very difficultdecisions will have to be made in response to a few impor-tant trends, including:

ii \\ \

• The proliferation of niches of IT applications has madeproviding the core services for sustaining the system as a wholemore complex to manage.

• The costs of providing the levels ot IT services expectedby both faculty and students workijig in these niches are increas-ing because of this complexity.

• Information technologies emerging from niche inno-vation are being rapidly identified as essential tools forresearch, teaching, and learning—tools without which suc-cess cannot be achieved—and the rate at which this is hap-pening is accelerating.

A core business of higher education is innovation inresearch and teaching, which is one of the ways in which ITecology at the university differs from that in the commer-cial sector. In the latter, some efficiencies in IT are producedby the standardization of hardware and software. Experimen-tation with new hardware and software is centrally plannedand approved. In contrast, a guiding principle for IT sup-port strategies at universities has been to encourage exper-

t I'/

c

^^fi

AKIHIKO MIYOSHI, NO Title, Pen marks on paper, 27 x 44 inches, 2004.

A note from the artist: "This is actually a circuit. I took a schematic ofa functioning circuit and traced it.The connection is turned into

curves instead of straight lines. My initial plan was to use conductiveink so the final drawing would still be 'functional'. I was not able to

use conductive ink, hence, the circuit/drawing is broken."

5B ISSUES IN SCIENCE ANDTECHNOLOGY

MANAGING CHANGE

imentation by both faculty and students. We keep the net-works open and provide at least best-effort support for a widerange of software and hardware. Not surprisingly, the resulthas been a proliferation of technologies and high expecta-tions that universities will create core services that supportwhatever niche users evolve. Diversity is both a value forour ecosystem and a drain on scarce resources.

That computing is a sine qua non of contemporary sci-entific research requires no additional arguments from us.What is less well known is how widely the dependence onIT has spread through all disciplines. It would have heen hardto imagine 30 years ago that academic philosophers wouldbe deeply involved in fields such as computational logicand compittational epistemology or that the theater depart-ment would join with computer scientists to create anentirely new academic enterprise called "entertainmenttechnology." And what will be upsetting to some is that ITis becoming equally essential in teaching and learning. Thehistory of technology-based or technology-enhanced learn-ing is not one of substantial demonstrable improvement inlearning outcomes. But things are changing. As those devel-oping e-learning tools draw increasingly from the body ofknowledge and techniques of cognitive science, effectivetechnology-enhanced learning is becoming a reality.

Carnegie Mellon faculty have had measurable success inusing cognitively informed computer-based instruction inseveral areas. The earliest results came from intelligenttutoring systems developed using lohn Anderson's theoryot cognition. Researchers used painstaking talk-aloud ses-sions to understand how novices and experts solve a prob-lem. They then developed software that compares stepstaken hy the user with these cognitive models of how oth-ers solve problems to provide intelligent individualizedfeedback to students working their way through problems.1 he result was Cognitive Tutors for middle- and high-school students that have produced documented improve-ments in the learning of algebra, geometry, and other sub-jects. Expanding the application of cognitive science (andCognitive Tutors) to online courses for postsecondary edu-cation, we have had considerable success with courses devel-oped as part of Carnegie Mellon's Open Learning Initiative(OLI), which is devoted to developing high-quality, openlyavailable, online courses. The OLl's goal is to embed in thecourse offering all of the instruction and instructional mate-rial that an individual novice learner needs to learn a topicat the introductory college level. These efforts are part of aworldwide movement to develop IT tools to provide effec-tive transfer of knowledge.

An additional development at universities is the pace at

which successful IT experiments are expected to becomepart of the IT utility. Consider the difference between e-mail and wireless networking. E-mail was invented as partof the ARPANET in 1971. Yet it was not really an expectedpart of universities' IT infrastructure until the late 1980sand early 1990s. Very early wireless networking was firstdeployed as an experiment at Carnegie Mellon between 1994and 1997. By 2001, not only was the entire campus coveredwith a commercial 802.11b network, but students and fac-ulty expected that network to be ubiquitous and alwaysavailable. The RIM (Corporation introduced the first Black-berry wireless handheld device in 1998. By 2002, many fac-ulty and staff on campuses started to use Blackberries. Today,many expect and depend on Blackberry (or Treo or othercellular technolog)') connectivity to keep up with their e-mailand calendars as they travel. Students, faculty, and staff willarrive this fall with a range of wireless devices that theyexpect to connect to their university e-mail, course manage-ment systems, central calendars, the university portal, andother core information services provided hy the university.

We should anticipate even more challenges. The expecta-tion of our constituents is that if a technology is available andit can help them accomplish their goals, the university shouldprovide whatever core services are required to support it.

Management strategiesWe propose the following as critical strategies for addressingnew challenges posed hy IT ecosystems in higher education.

Creating more robust feedback loops between the mem-bers of the academic community who are generating ITinnovations and those responsible for supplying a sustain-able environment. We have described how IT over the past23 years has changed from innovation and use by the few toinnovation and use by the many on campus. Without gath-ering data from the many, no central organization can effec-tively predict what the university must do to provide thecore services that that will both sustain the everyday uses andenable the evolution of innovative uses. The diversity ofsources of use and innovation requires new and more aggres-sive techniques for gathering data. Central IT organizationsmust engage in information-gathering outreach unheard ofin earlier times. University leadership must create the con-duits of communication and encourage active participationby faculty, students, and staff in that communication todevelop services to best support their needs and expectations.

Collaborating with other universities to develop sharedsolutions for both intra- and interinstitutional supportfor academic IT. There are many examples of collaborativeefforts among universities in IT. These range from regional

FALL 2005 59

education and research networks to consortia for softwarelicensing to joint software development Recent examples includethe development of the National Lambda Rail, a project bya consortium of research universities, along with Internet2,to create an all-optical, extremely high-bandwidth networkthat will serve the bandwidth and network research needsof higher education, and "Sakai," a project led by MIT, Indi-ana University, the University ot Michigan, and Stanford Uni-versity to develop an open-source course management sys-tem (technology to support Web posting of course materialsand collaborative work) for higher education. Universitiescould do much more, but resources are limited. As IT organ-izations at our various universities are called on to providemore and more services without additional resource input,collaborative work does and will suffer. Leaders are faced withthe difficuh decisions entailed in choosing between keep-ing up with increasing daily pressures on basic IT servicesand supporting collaborative projects with other institutionsto build sustainable and evolving IT core services for the future.

Selecting for adaptive and nonadaptive technologiesfor resource allocation based on their contributions to ourfundamental missions. Under resource limitations, not allIT applications can be equally supported. Thus, difficuhdecisions are required to select which expectations to meetand which to disappoint. Creating the robust feedback loopsmentioned above will help universities better adapt to chang-ing needs and expectations. But more than understandinguse and user expectations is required. We also need to makechoices based on solid data about the contribution of themany IT applications to our central research and teachingmissions. Although faculty are in the best position to iden-tify what IT effectively supports their research, even here thereare questions about whether the methods they use are glob-ally effective and efficient. For example, several universitiesare now encouraging faculty to forego having their computeclusters near their offices in favor of locating them in cen-tral machine rooms. The theory is that the university willspend less overhead and be able to provide better profes-sional IT support for central farms of clusters than for clus-ters distributed all over campus. If this is true, it is reason-able for university leaders to find ways to encourageshared-resource strategies at the cost of some individualconvenience. The same principle applies to many core serv-ices that are currently replicated across our institutions forthe sake of convenience at the potential cost of global inef-ficiencies and lack of interoperability.

Judgments about the relative contributions of educa-tional technologies to fulfilling the core mission of knowl-edge transfer are also essential. Despite being burned repeat-

edly by claims that technology will transform learning out-comes, central leadership is nevertheless reluctant to denyrequests for potentially promising new technologies forteaching and learning. Although finding effective technolo-gies has often been done through trial-and-error experimen-tation, there is increasing information from cognitive andlearning sciences about what is likely to help and what is likelyto hurt that can give us guidance about where to place ourbets. We have not yet really broken the pattern of deploy-ing new technology with only the hope that we will find eftec-tive pedagogical applications for it.

Take the fairly old notion of a laptop requirement oncampus. Many vendors have sold K-12 districts and univer-sities on the notion that equipping all students with a lap-top will "obviously" improve learning outcomes. Yet it isnearly impossible to find a study that reports anything morethan anecdotes about use or user satisfaction at "laptop uni-versities" or that controls for other educational innovationsintroduced at the same time as the laptops. In a recent studyconducted by the Office of Technology for Education andthe Eberly Center for Teaching Excellence at Carnegie Mel-lon, we learned that distributing laptops to students in acase study actually discouraged some collaborative learn-ing behaviors, which learning sciences have shown improvelearning outcomes. The study also indicated positive conse-quences of laptop ownership. The point is not that univer-sal laptop ownership is not a good thing; rather, that beforewidely deploying a technology, we should understand bothwhat problems we are trying to address and what we alreadyknow about how that technology might solve the problems.

In the absence of rigorous data, we cannot afford toinvest in proliferating devices and software that merelyappear to hold some promise to improve learning. Anotherexample is the increasingly vociferous claim that educa-tional software needs to exploit the fact that the currentgeneration of learners are "natural multitaskers." This seemsan increasingly dubious or at least complicated claim inthe light of developing evidence that multitasking is accom-panied by reduced cognitive attention. Leadership is requiredto say "no" in the face of unsubstantiated claims that a tech-nology will transform teaching and learning.

Revitalizing commitment to open standards to ensurethe sustainable and evolvable development of IT in academe.Finally, education must do more to return some sanity tothe IT standards movements. Our current IT ecosystem's (some-what fragile) stability depends now on far-sighted work in"open standards" that have allowed software and hardwareto interoperate and have structured, shared services. Mostof these standards originated at individual universities.

60 ISSUES IN SCIENCE ANDTECHNOLOGY

MANAGING CHANGE

iMembers of broader academic communities helped guidethem through sanctioning bodies and lobbied for vendoracceptance. Enabling diverse IT ecosystem configurationsdepends on the existence of open standards that allow manydifferent devices and pieces of software to coexist, commu-nicate, and use common features of the environment. Ithas always been a challenge to convince vendors to adhereto open standards if there is any commercial advantage tobe gained by creating features that step outside those pro-

AKIHIKO MIYOSHI, NO Title (detail), Pen marks on paper,27x44 inches, 2004.

FALL 2005 61

UNIVERSITY LEADERS

MUST INSIST THAT

STANDARDS BE USABLE,

THAT THEY BE DEVELOPED

AND DOCUMENTED IN A

TIMELY MANNER, AND

THAT THEY CAN BE EASILY

ADOPTED BY COMMERCIAL

VENDORS AND

INDEPENDENT OPEN-

SOURCE DEVELOPERS.

tocols. The situatioti is not getting better. Indeed, there is asubtle deception about claims of adherence to standards. Ifa product mostly follows a standard, the vendor will say it"adheres." But the hard reality faced by those implement-ing the products is that anything short of complete adher-ence often results in complete failure of the service to workproperly or requires time-consuming customization. Verylarge vendors can often succeed by stepping outside thestandards and encouraging the adoption of their proprietarysystems as a "top-to-bottom" solution for an institution.But even smaller vendors will often opt to ignore open-standard options if they deem them too great an obstacleto getting their product to market in a timely fashion.

Indeed, more than the vendors must bear the blame foropen standards not playing the role they should in sustain-ing academia's diverse and evolving array of applications.Standards definitions have often become too cumbersome,and the ratification process too slow. The communitiesdeveloping standards seem to lose touch with both ven-dors and users, open source movements that should beusing their standards in the creation of products. The eftortto develop the IMS and SCORM standards to allow inter-operability of course management systems and repositoriesof related materials has become a multiyear marathon thatis producing standards so detailed that vendors and univer-sity software developers are reluctant to use them. This

could result in each university opting for its own closedstandard, which would make it impossible for universitiesto take advantage of shared core services and much moredifficult for successful innovations to spread.

University leaders must insist that standards be usable,that they be developed and documented in a timely man-ner, and that they can be easily adopted by commercial ven-dors and independent open-source developers. Collectively,colleges and universities constitute a large market and arethus in a position to play a powerful role in the standardsgame. We can use our intellectual strengths to lobby researchersin academia and the corporate sector who are engaged increating standards, as well as organizations such as the Insti-tute of Electrical and Electronics Engineers, who are respon-sible for ratifying them. Finally, we can use our buyingpower to reward vendors that build features that reallyadhere to open standards.

The implications of tbe transition from IT as infrastruc-ture to IT as ecosystem are profound for leaders in highereducation. We are already making tough choices about howto structure IT organizations and allocate resources to IT.Our arguments here suggest that IT organizations shouldstart to look and act differently than they do today: Theyshould be part of robust feedback loops with the dynami-cally changing niches of innovation throughout the univer-sity community; they should be looking beyond their ownwalls to collaboration with other institutions; and theyshould belp revitalize the open-standards movements thatare so critical to sustaining diversity in our ecosystems. Butthis means hard choices for university leadership outside ofIT. University leaders must partner with faculty, students,and IT leadership to make some hard choices about bow tosustain those niches in the ecosystem that are most valuablefor our core missions. It means some paths of innovationand some core services will be starved. It means divertingresources from services we could provide now to fuel thecollaboration and development of standards that will sus-tain us into the future. In a context in which IT gives us allinstant gratification—we want to use it now!—these will notbe popular decisions. Few leadership decisions to sustain anecosystem for tbe future at the cost of current expectationsand needs are.

Joel M. Smith (joelms@an(hew.cmu.edH) is vice provost andchief information officer and fared L Cohon is president ofCarnegie Mellon University.

62 ISSUES IN SCIENCE ANDTECHNOLOGY