TALK GIVEN AT THE NIELS BOHR INSTITUTE, 8/11/2000

BY JOAN WARNOW-BLEWETT, ARCHIVIST EMERITUS

AIP CENTER FOR HISTORY OF PHYSICS

INTRODUCTION

Sometimes it feels as though the organizational structures for research in the physical sciences are changing on a daily basis. The latest thing now - as far as I've heard - is the visual laboratory or collaboratory.

Organizational change - along with technological change - present enormous challenges to archivists and others who must learn how to document the new phenomena. These changes and challenges - along with our ongoing need for individual scientists to help provide essential documentation - are the subject of my talk today

Before going further, I should say a word about the American Institute of Physics - the AIP as I shall call it. The AIP is best known as a publisher of scientific journals. The mission of my division at the AIP - the Center for History - is to preserve and make known the history of modern physics and allied fields. That means, among other things, that we work to see that records of archival value are identified and saved; it also means that the AIP must be in a position to provide guidance when it's needed to those responsible for records. The AIP Center, in fact, invented a revolutionary approach to documenting a subject: instead of collecting everything for its own Niels Bohr Library, it would work cooperatively with other institutions to save papers in appropriate repositories. That approach is known as a documentation strategy.

When I first started in 1965, the concern of the AIP Center was to document physics from about 1890 to the pre-World War II period. During that period, the most important science - at least in the physics community - was being done by individuals in academia. Accordingly, our first documentation strategy was to work in cooperation with academic archivists to save the papers of individual academic physicists

CHALLENGE OF NEW INSTITUTIONAL STRUCTURES

Serious challenges to our initial strategy came with the appearance of new organizational research structures that went beyond academia. In the early post-World War II period, we saw the increasing importance of nonacademic research laboratories - particularly government laboratories and government-contract laboratories. More recently we witnessed how large, multi-institutional collaborations have become the platform for much of the significant research in contemporary science. We at the AIP were ignorant about the ways these new kinds of institutions functioned. We didn't know what organizational structures were in place, how records were created and used, which records were of archival value, and how and where valuable records could be saved. We didn't begin to understand how to document scientific research in these new arenas.

To meet the challenge, the AIP Center developed a new kind of field research to resolve archival problems or - to put it another way - to give us enough understanding of the new organizational research structures to know how to document them. We call this field research documentation strategy research because - once we have achieved the necessary understanding of how to document the particular area - we can enlarge the scope of the AIP Center's ongoing documentation strategy.

Thus far, we have applied documentation strategy research in two projects: the AIP Study of Department of Energy (DOE) National Laboratories and the AIP Study of Multi-Institutional Collaborations. The goals of both studies were: first, to identify the most important organizational structures, functions, records creators, and events; next, to understand how and why records are created and used; and, finally, to identify the probable locations of valuable records.

I'll be happy to discuss the findings and achievements of the AIP Study of DOE National Laboratories after my talk. Suffice it to say that we did achieve our goals of understanding the way nonacademic research laboratories operate and how to document them. Our reports on this study are particularly useful to institutional archivists.

AIP STUDY OF MULTI-INSTITUTIONAL COLLABORATIONS

The second AIP documentation strategy research project - and our main topic today - is the AIP Study of Multi-Institutional Collaborations. It was in the late 1970s, while we were studying the DOE National Laboratories that we first encountered multi-institutional collaborations and realized that we did not know how to identify the evidence needed to document them. It was to us a new, awesome, and bewildering sight: teams of researchers from many institutions joining together to carry out an experiment and then disbanding. These characteristics seemed to present archival nightmares. We had many questions: for example, What kinds of records did these far-flung collaborations create? Who owned the records? What happened to them? Indeed, was it even possible to document multi-institutional collaborations?

The AIP Study of Multi-Institutional Collaborations was launched in 1989 and completed this year. Initiated by the AIP Center because of the increasing importance of large-scale research projects and the many unknowns and complexities of documenting them - the AIP Study was the first systematic examination of the organizational structures and functions of multi-institutional collaborations. We covered research projects involving three or more institutions in physics and related fields: high-energy physics (Phase I), space science and geophysics (Phase II), and ground-based astronomy, heavy-ion and nuclear physics, materials science, and medical physics (Phase III). For each discipline under study we had a Working Group of historians, archivists, sociologists, and - most important of all - distinguished scientists and science administrators.

Throughout the AIP Study, our field work consisted - on the one hand - of structured interviews with scientists who participated in collaborations selected to serve as our case studies and - on the other hand - of site visits to numerous archival and records management programs. The interviews provided data on organizational patterns, records creation and use, and the likely locations of valuable documentation. The archival site visits - to academia, federal science agencies, the National Archives, and elsewhere - provided data on existing records policies and practices and the likelihood of collaborations being documented under current conditions. Reports published at the end of each phase of the study include our findings, appraisal guidelines, and recommendations to agencies, research laboratories, university archives, and others. They are available on request from the AIP Center; summary reports are on the Center's Web site. The final reports - covering the whole, decade-long study - are in preparation.

SOME HISTORICAL-SOCIOLOGICAL FINDINGS

The stories of collaborations in the contemporary physical sciences constitute a fascinating tapestry of patterned diversity. Within each scientific specialty covered by the AIP Study, the scientists' quests for effective, feasible, and soul-satisfying organizational frameworks for research have produced variations on classic themes. In order to obtain insights into processes that must be understood to begin framing a historical investigation - or developing a new documentation strategy - we designed our program of interviews to cover a range of characteristics :

How did the collaboration form and who made it form;
Who provided the collaboration with funding and what obligations did the collaboration owe to its patron(s);
How was the collaboration organized and managed and who took responsibility for the collaboration's administrative needs;
How did the collaboration structure communication among its individual and institutional members;
How did the collaboration divide labor and what was the role of the participating institutions in the collaboration;
Who determined the timing, placement, and content of dissemination of scientific results stemming from the collaboration's activities;
What were the opportunities, challenges, and obstacles to international participation in the collaboration; and
What significance did the collaboration have for the course of scientific research and the careers of its individual participants?

The interviews thus provided at least skeletal information on the origins, organization, and legacy of each collaboration. The historical and sociological analyses of this information can help archivists, science administrators, and policy analysts to assess how collaborations generate and use records, why collaborations organize themselves in the ways they do, and why they seem more or less successful in the eyes of their participants.

TYPOLOGY OF THE AIP STUDY

One of the most fascinating products of the AIP Study's program of interviews is the classification scheme or typology developed by the project historian and sociologists for the organization and management of collaborations. The area of organization and management is the aspect of collaborations most closely connected to the generation and accumulation of records.

The basis for the typology is "cluster analysis" - a statistical technique that groups objects on the basis of how closely they resemble each other across a range of variables. The project team performed cluster analysis on the organization-and-management variables for the 46 collaborations for which we had complete information. They found variables that were sufficiently inter-related to justify reducing them to four factors:

formalization (which combines presence of written contracts, presence of an administrative leader, division of administrative and scientific authority, self-evaluation of the project, and outside formal evaluation),
hierarchy (which combines levels of authority, system of rules and regulations, style of decision-making, and degree to which leadership subgroups made decisions),
presence of scientific leadership, and
style of division of labor.

The result of the cluster analysis is that collaborations can be reasonably divided into four organizational types. With one notable exception, organizational types are not field specific - meaning that the particular disciplinary specialty of a collaboration (e.g., materials science or geophysics) is not a clue to its organizational type. The exception is particle physics. More on this later.

The first organizational type is comprised of collaborations with a high degree of formalization, high degree of hierarchy, high scientific leadership, and specialized division of labor. We designate this type "highly structured." The second and third types differ from the first in that they are comprised of collaborations that are either less formal or less hierarchical than the highly structured. They are distinguished from each other by their needs for scientific leadership and by their method of dividing labor. The second type - "semi-structured with no scientific leader" - never has a designated scientific leader and usually has a specialized division of labor; the third type - "semi-structured unspecialized" - usually has a designated scientific leader and always has an unspecialized division of labor. The collaborations in the fourth type register the lowest amounts of formalization and hierarchy, while still possessing scientific leadership and a specialized division of labor. We designate them "low-structured."

Today - with limited time - I will focus on this last type, because it may be of greatest interest to this audience.

The low-structured type of collaboration is, as the label suggests, the absence of the classic features associated with Weberian bureaucracy. The membership of this type is dominated by particle physics collaborations. Among all the specialties in physical research we covered, particle physics alone has a distinct style of collaboration. Occasionally, particle physics collaborations fall outside the main category for particle physics and occasionally collaborations in other specialties most closely resemble a typical particle physics collaboration, but it seems justified to speak of "particle physics exceptionalism."

Particle physics collaborations are exceptional in their combination of two characteristics. First, the participants find that their collaborations are highly egalitarian. Compared to what we heard from collaborators in other disciplines, particle physics collaborators describe decision-making as participatory and consensual, define their organizational structure through verbally shared understandings rather than formal contracts, and institute fewer levels of internal authority. At the same time, in contrast to collaborations that did not publish scientific findings collectively, the scope of particle physics collaborations encompasses nearly all the activities needed to produce scientific knowledge, including those activities most sensitive to building a scientific career. The collaborations always collectivize the data streams from the individual detector components built by the participating organizations, they frequently track who within the collaboration is addressing particular topics with the data, and they routinely regulate external communication of results to the scientific community.

Particle physics collaborations minimize the powers that collaboration managers can exercise in order to make their members comfortable with the large breadth of activities that the collaboration as a whole regulates. In all other research specialties we examined, participants in collaborations were more autonomous than particle physicists in the generation and dissemination of scientific results; and the participants (more or less happily) allowed collaboration managers to exercise discretionary powers to secure what the collaboration as a whole needed.

The prevalence of high-breadth, egalitarian collaborations in particle physics is due to: (1) the dispersal of particle physicists among many universities, (2) the specialty's centralized institutional politics, and (3) competitive pressures. Because particle physicists in the United States and Europe are dispersed among many universities and because they crave integrated, multi-component detectors, they need to be in high-breadth collaborations in order to conduct publishable research. Because collaborations must submit proposals to central authorities for access to an accelerator, participants are behooved to commit to an organizational structure that convinces the accelerator laboratory's administration that they are properly organized to produce what they promise. With respect for internal structure thus secured before any commitment of resources to the collaboration is made, collaboration administrators have not required formalized powers to maintain order and could afford to grant broad rights of participation to all members of the collaboration, from graduate students to senior faculty. Such Athenian-style democracy has produced publications rather than cacophony because competition for discoveries - and for career-advancing recognition - limit the collective tolerance for intra-collaboration dissent.

TRENDS IN MULTI-INSTITUTIONAL COLLABORATIONS

Collaborations have changed - even over the decade since the AIP Study started. It is more than just the impact of the Web. Many factors have changed the nature of collaborations. There is, for example, increasing scrutiny by the funding agencies and regulations imposed on the grantees and the laboratories; these, in turn, have increased the accountability collaborations have to allow for and the levels of reporting they are required to make. The general trend is for multi-institutional collaborations to become larger, more formal, and more international. In the field of particle physics, the trend is toward very, very large collaborations.

Another striking change is that collaborations in one field may take on characteristics of those in another field. The point was made clear to us at the last meeting of the AIP Study's Working Group. The subject was the role of the builders and the users of detectors / instruments in the fields of particle physics and ground-based astronomy. A decade or more ago, most particle detectors were built and used by the same, single collaboration and most telescopes were built by a collaboration (and then maintained by the facility) for other scientists to use. The current situations are quite different because of the increasing sophistication of the instruments / detectors and the need for more sophisticated processing of much larger amounts of data. New multi-purpose detectors in particle physics have practical lifetimes that may equal those of the accelerators; this means the detectors are used by more than one collaboration and that maintenance has shifted to new permanent, technical staff at the accelerator facility; thus, detectors are moving toward the model of astronomy in terms of builders and users of instrumentation. Meanwhile, in the case of ground-based astronomy, the instruments - the equivalent of particle detectors - are increasing in cost faster than the telescopes; the huge increases in costs for instruments and data processing have inspired ground-based astronomers to begin looking into management practices in particle physics collaborations.

CIRCUMSTANCES AFFECTING RECORDS CREATION

Another interesting finding of the study was the circumstances that affected records creation.

Of all the disciplines studied, we found that collaborations in particle and nuclear physics generate the most documentation. In this field, just about everything individuals and teams do for an experiment is subject to review or use by others in the collaboration. In every other discipline we studied, individuals and teams have more autonomy. Consequently, collaborations in particle and nuclear physics are burdened with creating a rich collective record, while collaborations in other fields are less burdened.

We found that particular circumstances pressed for the creation of valuable documentation:

Size (especially in terms of numbers of institutions) and geographical dispersal of institutions tended to foster records of intra-collaboration communications on technical progress.
The emergence of fax and e-mail, with correspondingly less reliance on the telephone, resulted in additional documentation of collaborations.
Collaborations made up of multi-disciplinary components or that required extensive communication with engineers found it necessary to write out more of their ideas than they would otherwise.
Collaborations that had what physicists call "interesting results" tended to create more documentation of potential value to future scholars, e.g., by stimulating more written arguments over analysis and by producing more publications.
Collaborations that contract out for construction of instrumentation they have designed in detail provide more written documentation for their contractors.

MOST VALUABLE COLLABORATION RECORDS

Now that the AIP Study of Multi-Institutional Collaborations is completed - we have a good idea of which collaboration records are most valuable for scholarly research and where these valuable records are likely to be located in active files.

In virtually all fields, multi-institutional collaborations need the financial support of government funding agencies. The records of successful applications, consisting of proposals, referee and panel records, and narrative and fiscal reports provide good documentation. They are to be found in the files of program officers at funding agencies.
Collaborations in most disciplines also require access to major national or multi-national research facilities such as accelerators of all sorts, ocean-going vessels, and supercomputers. Panels of distinguished scientists review applications for access to these facilities. Their work generates files of proposals, comments of panelists, and, frequently, impact statements and periodic progress reviews; these can offer the single best core record for collaborations. Records of these advisory committees are located at the research facilities or - in such cases as oceanography - in the office managing the particular facility.
Some of the projects AIP investigated were expensive enough or relevant enough to public policy (e.g., NASA's Voyager Mission) that the process of funding was far longer and far more political. The likely products of this process are records in parliaments, national academies, funding agencies, and some ad hoc bodies - all of high interest to scholars in a broad range of fields.
Papers of individual scientists are critical to the documentation of collaborations. The great majority of scientists participating in the collaborations come from the academic sector. When academic scientists play key roles (as spokesperson, chief scientist, chair of a policy committee, and so forth) their collaboration records will probably be located among their professional papers. Whenever the collaboration in question is of scientific significance, those records will be sought after by future scholars - even in cases where the remainder of the scientist's papers are unimportant.

AIP STUDY APPRAISAL GUIDELINES

Our final reports include guidelines for appraising records. These guidelines reflect two of the purposes of the AIP Study: (1) to identify a small set of core - or basic - records that should be permanently preserved for all collaborations in a given disciplinary field and (2) to distinguish the wider array of documentation that should be preserved for selected experiments - those that are of major scientific significance - and that, therefore, will be of high interest to future historians, sociologists, and other users.

Among the appraisal guidelines is a discussion of the papers of individual scientists. We say that archivists and records officers should place the highest value on the papers of PIs and other leaders of multi-institutional collaborations. Papers of these scientific leaders are prime locations for documentation of a number of topics, including details of staffing, plans for data gathering and analysis, and use of the data by collaboration members. The papers will typically contain proposals, personal notebooks, and correspondence with other collaboration leaders and with funding agencies. In cases where the scientific leader was also an instigator of the collaboration, the files may provide especially unique documentation of the initial thinking and early plans of the project. When individual scientists have been leaders of significant collaborations or have regularly played a leading role in important research, the records of their participation should be saved (whether or not the full range of papers documenting their careers merits archival preservation).

I wish there were time to say more about our appraisal guidelines.

CURRENT ARCHIVAL PRACTICES

We must now turn to the troublesome subject of what documentation may not be left in a decade or two. That means we need to pay serious attention to the situation in archives and pray that current practices improve.

Here is a glimpse of current practices in the key sectors: archives in academia, government-funded research laboratories, and national agencies.

University archival programs in the US have for decades provided the most secure location for papers of individual scientists. Academic archives are also strong in some other countries such as England and Australia (primarily in English-speaking countries, strangely enough).
But, academic programs in most other countries - with a few exceptions - lack the tradition of saving papers of key faculty, including scientists.
Archival programs have been initiated in a number of government-funded scientific research laboratories; notable examples are CERN in Europe and several DOE national labs and oceanographic institutes in the US.
But, for each happy story, there are 10 or 20 major scientific institutions that lack archival programs or even professional records management programs.
A few national science agencies have archival programs. Four come to mind immediately: the Atomic Energy Authority in England, CNRS in France, the DOE in the US, and the European Space Agency.

Scientists and historians should realize that archival programs - where they exist - are of low priority to their home institutions. As such, archival programs are often short on funding, space, and staff - all of which are needed to identify and preserve records of value to future scholars.

AIP RECOMMENDATIONS

All chapters of our report - whether they be findings, analyses, or assessments - lead to our project recommendations. A number of them pertain to changes in records programs at research institutions and federal agencies that must be made to secure an adequate record of multi-institutional collaborations and their contributions to science and our society.

The project recommendations of the AIP Study are aimed at preserving only a small fraction of the records created by multi-institutional collaborations. As shown in our appraisal guidelines, records of archival value will consist of a small set of core records plus, in a few cases, a wider range of records for very significant collaborations. Our experience indicates that records of this quality will meet the needs of scientist/administrators and future scholars.

The most important recommendation urges a new approach to securing the documentation for future collaboration projects. We suggest that - once a project has been approved by a research laboratory - the collaboration be required to designate a member to be responsible for its collaboration-wide records. In addition - where historical significance warrants - individuals should be named to be responsible for group- (or team-) level documentation of innovative components or techniques. This information should be incorporated into any contractual agreement with the collaboration. Use of this simple mechanism would assist archivists by assuring that records will be available for appraisal and by providing information on their location.

CONCLUSION

I hope my simplified and fragmentary report has given you some idea of the work of the AIP and, especially, its Study of Multi-Institutional Collaborations. The long-term study has been funded by the AIP and by public and private foundations, including the NSF, the Mellon Foundation, the National Archives, and the DOE.

Has the experience been worth the investment? One conclusion of the AIP Study is that new actions will be required to secure the wide range of records scholars will need. Are the AIP and the archival community in a better position now to overcome the obstacles? I say yes. Yes, it was worth it, and yes we can (with patience and perseverance) achieve substantial improvements in securing the documentation.

The main reason for optimism is that we have gained some intimacy with the process of collaborative research and some understanding of the value of the records created by the process. These new understandings have given the AIP a voice of authority. It is already making a difference in our efforts to establish procedures that will protect records of research and funding created by key science agencies. In the scientific community as well, we are confident we can get help from boards of the National Academy of Sciences (and, perhaps the National Science Board of the NSF) to identify the most significant scientific research collaborations. Achievements like these will help the AIP and others to save the broad spectrum of documentation future scholars will need. The papers of individual scientists will be critical to that effort.

It may be difficult for scientists - even those who direct collaborative work - to recognize the importance of saving documentary source materials. It may seem to them that their personal recollections and those of their colleagues are sufficient. This is unfortunate from the standpoint of present needs. From the standpoint of the future it is disastrous, for even the imperfect personal recollections will die with the scientists, and later generations will never know how some of the important scientific work of our times was done.