lASSIST Newsletter, Vol. 2, No. 1 (Winter 1978) THE NETWORK-BASED SCIENTIFIC COHHUNITY ECONOMIC CLIMATE AND SOCIAL STRUCTURE Richard C. Roistacher Center for Advanced Computation University of Illinois, Urbana, IL 61801 ABSTRACT The effort is made in this paper to identify those elements of the scientific community requiring computer services; evaluate the economies of using computing network facilities to satisfy those needs; and to pro- vide some suggestions as to the substantive advantages of network use. ECONOMIC CLIMATE Several tors make way to o search.. T market has of talent universiti university trial sett tutions a Ph.D.s wit research, smaller in facilities ganizing search. S cient res institutio research f snaring is such shari by allowin research f basis. H costs wil expensive. social networki rganize he decli resulte from th es into , govern mgs. re now h first- Howe ve stitutio and a for the ince the ources t n with a acilitie necessa ng has g outsid acilitie owever, 1 make and econo n g an at scien ti ning acad d in a di e major a wide va ment, an Many smal staffed b class tra r, many ns lack tradition conduct re are no o provid complet s, some ry. In t been acco ers to u s on a increasin travel ev mic trac ic emic sper rese riet d m 1 in y y inin of rese of of t su e e e se for he p mpii se m visi g en f ac- tive re- job sion arch y of dus- sti- oung g in the arch or- re- ffi- very t of m of ast , shed a 3or ting erg V more At of en of c have and dropp years conti the e compu trave cilit acces small tions the ergy om pu been comp ing f a nue xten tati 1 an ies. s t and sa has tati dec utat by t ren inde t t on d t it o re ser me time increas on and lining, ion cos one hal d which finitely hat comra can su he dupli is poss search vice ori tnat ed, t commu Commu ts ha f eve is exp unicat fastitu cation ible t faciii ented the cost he costs nication nication ve been ry five ected to hus, to ion and te for of fa- o expand ties to institu- Communication networks and data bases are natural monopolies; they function best when a single source serves the largest possible clien- tele. Groups of scientific workers are probably best suited to an en- vironment of pure competition in which there are no barriers to the setting up of new groups. The nec- essary communication network is al- ready rapid and c been satis numbe task commu ers i them, them they in p iy. § ommerci estabii factory r of cl in est nit ies s to i f inan with t need. lace a everai al da shed a servi ients. ablish of sci nf orm ce th he sp nd is major ta arch nd are ces to Thus, ing net entif ic people, em , an ecific expanding university ives have providing a growing the major work-based research- organize d provide resources ing stil new the eval and and such tech ronm resu soci exte cien netw supp V e n t h incr ea 1 an scien dissera uation govern instit thing n o 1 o g y ental It in al SCI nt tha t way orking ort. ougn singl ever- tific inati find ment ut ion s as asse impa an i ence t net to CO wil reso y sc mcr kno on ings ope al prog ssme ct ncre res work nduc 1 re urces arce , easing wledge of res into r a t i on reguir ram ev nts. Stat em asin ea ing is t such ;ing trch. are th nee an earc comm s. emen alua and ents dema T an res fin grow- ere is d for d for h and ercial Legal ts ror tiop. s, envi- will nd for o the eff i- earch. ancial USERS OF SCIENIIIIC NETWOMS In this discussion, users of scientific networks will be classi- fied by their relation to the net- work, rather tnan by job title or institutional affiliation. Scientists A enti whos disc edge in u cies dust sibl of a prof but tion n ob fie e p over M nive and ry. e th mate essi with s as viou rese rima y of OS t rsit lab Net e s ur s onal out sci s cl arch r y 1 new scie ies, orat work uppo cien tr ins enti ientele net wor nterest scien ntists gove ories, ing als rt and tists , aining tit ut 10 sts. tor k is is tif ic are e rnraen and o mak inte peop in nal a a sci- people in the knowl- mployed t agen- in in- es pos- gration le with science ffilia- - 19 - lASSIST Newsletter, Vol. 2, No. 1 (Winter 1978) Scientist-professional SOCIAL STRUCTURES IN ~ ~ :^ET5DlK£lIiED~5R0UPS An increasing number of doctoral level graduates are going directly The physical and computational into social service and clinical resources required to support a positions. While many people in scientific network are organized in such positions are not active prod- a formal fashion, since they are ucers of research, they are often subject to a host of financial and avid consumers and critics of sci- legal constraints. However, the entific research. Scientist- pro- social structure of network-based fessionals have many interests scientific groups is still not which are not shared by pure and clear. We have had little experi- applied scientists, and would prob- ence with forming and operating ably form their own network, based such groups, and there is no re- groups.. However, it would be easy guirement for any strict division to maintain communication between into types. However, it is possi- groups of scientists and scientist- ble to guess at some types of professionals when dictated by com- groups which may evolve, mon interests. Practitioners Task Groups Task groups are network-based Practioners, as used here, are research projects. Several people people whose interest is primarily would use the network to coordinatem the use, rather than in the dis- their joint work on a scientific covery or analysis ot specific problem. Although task groups knowledge. A network of profes- would consist of people at differ- sionals would obviously wish to ent locations and perhaps different make use of scientific research re- institutions, their research would suits. Occasionally, groups of probably be funded and operated as professionals might wish to obtain a joint project, consulting help trom scientists. Contacts between networK-based groups of scientists and groups of professionals woula probably tend Consortia to be mediated by institutional fa- cilities such as librarians and Groups of workers might form transfer agents, more tban by per- consortia for the purpose of con- sonal acquaintance. tracting for data. The members of the consortium would pay for the data collection and for the estab- lishment of a common data base, but Librarians and transfer agents would pursue the independent analy- sis of their own data. These are people whose primary interest is in providing services to network clients. Reference li- brarians would be able to act much Invisible colleges as they do now, but would have bet- ter access to potential clients and The most obvious form of organi- better knowledge of current issues. zation, and the one which is pres- The term "transfer agent" applies ently most common, is the "invisi- to someone wno would be part organ- ble college." Groups of people izationax development specialist with common interests use the net- and part extension agent. The work to exchange messages, manu- transrer agent would act as refer- scripts, screeds, broadsides, and ral service, social director, and gossip. People pay their own ex- spreader of information. Transfer penses and pursue their own inter- agents will become increasingly ests, using whatever public facili- necessary as the demand for inform- ties they need. The invisible ation and the complexity of ir:torm- college forms a ground for recruit- ation sources both increase. ing members into more formally or- ganized groups such as task groups It can be expected that librari- and consortia, ans and transfer agents, having is- sues of their own, will form their own network-based groups. Network journals By combining the document proc- essing, information retrieval, and communications facilities of a com- puter network with a system of edi- 20 - lASSIST Newsletter, Vol, 2, No. 1 (Winter 1978) tors and referees, it is possible Archival analysts to produce a network-based journal. Several schemes have been proposed A second clientele for a re- for the establishment and operation search network is social scientists of such a journal. One of the most who analyze machine-readable archi- attractive schemes would allow the val data. Such clients would find journal to "publish" all submis- access to a network highly reward- sions, but with the addition of a ing for both research and communi- referee's score. A network journal cation activities. Since the net- would allow formal and impersonal work provides data archive and contact between network clients on computational facilities which are the basis of evaluated research equivalent to those at a major uni- findings, rather than personal ac- versity, archival analysts will quaintance. find tne network's facilities equal or superior to those of their own institutions. Archival analysts shoula find affiliation with a net- Brain trust work both professionally and so- cially rewarding. The ability to send messages to large groups of people makes possi- ble the use of an invisible college as a source of otherwise unobtama- Local firoducers o_f data ble information. Someone with a question may begin by asking col- A third category of scientists leagues or the reference librarian. consists of those who can meet If all else fails, it is possible their own instrumentation needs to broadcast a message asking for locally. Instrumentation as used help with the problem. Not only here includes all those facilities may someone out there have an an- and activities required to produce swer, but other people with the data either in machine-readable same problem may now share the an- form or in form suitable for key swer. One of the chief tasks of entry. Scientists in this category transfer agents may be to bid out are able to raise their own rats, and pass on questions and answers. use their own instruments, or ad- minister and code their own ques- tionnaires. Such scientists can use the network solely as a commu- OSERS OF A SCIENTIFIC NETWORK nication medium, to analyze their own data, or to collaborate and Membership in network-based re- share locally produced data with search communities, while far bet- others. ter than isolation and inactivity, will provide neither for all needs of scientists nor for the needs of all scientists. The most obvious Data contractors shortcoming of scientific networks is that there is no way that they can provide direct access to labo- __ ratories and instrumentation. Sci- tists who can contract for data, entists can be classifiea into five Work of this type requires the ex- categories with respect to their act specification of data collec- dependence on instrumentation. tion procedures, but does not nec- essarily require that the investigator actually operate the eguipment. Many physical scien- Theoret icians tists essentially contract for data, buying time on such facili- Some scientists, either mathema- ties as telescopes and nuclar reac- ticians or theoreticians, require tors. Survey researchers often buy access only to libraries and col- interview time and questions on na- leagues. Participation in a net- tional surveys, work-based community provides them with a peer group and an audience. Most of today's data contracting A network-based clientele of theo- facilities are national resources, reticians would be relatively easy so huge and costly as to be beyond to support. In fact, social sup- the means of most universities, port is perhaps the only unique Examples of such facilities are the service a network would provide to Mt. Palomar telescooe, the Stanford the unaffiliated theoretician. linear accelerator^ and the Na- tional Election Survey. It might be feasible to establish data con- tracting facilities equivalent to those at a major university. Such facilities would provide "contract A fourth category of potential network clients consists of scien- lASSIST Newsletter, Vol. 2, No. 1 (Winter 1978) data on a scale and at a price Inte rna l funding affordable by the individual or by small groups of researchers. Con- Access to networks will be dif- tract laboratories staffed by re- ficult for those who operate en- search assistants, technicians, and tirely within institutional budg- a supervisor would perform experi- ets, without control of their own ments according to protocols pro- funds. Most universities greet the vided by remote researchers. While very mention of purchasing external it is not clear which, if any, computing services with fear and sorts of research could be pursued loathing. Deans and department in such a fashion, the feasibility heads do not see why "real" money of such contract facilities must be should be used to buy computing investigated. when "free" computing is available locally. There are no easy answers ft second way in which scientists to such questions, and the individ- might successfully contract for ual scientist or faculty member is data is through the formation of in a poor position to change insti- consortia. A network-affiliated tutional constraints, scientist could propose that a group of researchers jointly sup- port the costs of contracting for data, which they could then analyze Extern al funding jointly or as individuals. One strategy suitable to exter- nal funding is to have the sponsor write a special condition to the Hands-on experimentalists grant requiring the principal in- vestigator to nave discretion over A fifth type of scientist re- where computing money may be spent. ?uires personal access to equipment Another alternative is to have the ound only in the research labora- funding agency execute a separate tories of a major university. It is contract with the network facility, difficult to see how such scien- setting up an account for the re- tists could use network facilities searcher. The researcher's univer- for anything other than scientific sity thus never exports any comput- and social communication. ing' funds because it never imported any computing funds. FINANCI NG Terminal costs Computer conferencing it has a policy of denying faculty members access to professional The on line conference not only None of these things will happen allows for consulting over a net- unless there is money available to work and for scientific communica- pay for them. People's major co- tion of a new and powerful kind, nern is with the acquisition of but also provides a point of access terminals and with paying for net- to networking. By definition, a work access. The price of termi- local computing facility cannot nals has declined precipt iously provide access to a conference tak- over the last five years. A termi- ing place on a remote computer, nal which cost about $3000 five Thus, the would-be conference user years ago can now be bought for cannot be told that the remote site less than $1000. It is unlikely has no unique facility. Also it that the decline in the prices of should be difficult for a univer- terminais will continue to be as sity administration to state that steep; however, the price of excel- lent printing terminals is fast ap- proaching that of ordinary office meetings, electric typewriters, and will probably soon be below the cost of There are no easy answers or present typewriters. Thus, the of- easy predictions as to when net- rice typewriter of the near future works will grow large enough and will be nothing more than a com- powerful enough to attract a major puter terminal lacking communica- rraction of scientists as users, tions equipment. Turning the of- However, the example of Telenet, fice typewriter into a terminal growing from five to 85 cities in will require nothing more than the two years, gives some cause for op- installation of a one- or two-hun- timism. dred-dollar communications circuit board. - 22 -