Trip Report Systems Laboratory IBM Thomas J. Watson Research Laboratory Yorktown Heights, New York November 19, 1980 through November 26, 1980 G. M. Engel E. C. Hendricks D. R. Johnson Background Andy Heller, IBM Fellow on Research Staff, and Floyd Van Scoy, staff to Dick Bergstresser, DP Marketing Group Director of Advanced Technology, visited IBM San Diego FSD on the third of November 1980 to discuss moving the ARPANET Internetwork Gateway project to the IBM Thomas J. Watson Research Center at Yorktown Heights, New York. After Andy and Floyd met in private with our manager Phil Stuver, discussions were opened with the technical staff, Mike Engel, Ed Hendricks and Dale Johnson. The objective of this project is to begin to bridge the gap between IBM computer systems and network technology predominant among government agencies, contractors and universities. More specifically, we are working to develop according to DOD standards the technical capacity to interconnect networks of IBM computers and systems to similar but different computer networks used by government agencies and their affiliates. We explained that past experience had demonstrated that projects of this kind are most successful in small business sites, relatively isolated from the routine distractions of large installations. This project can be carried out effectively by three people with limited hardware resources. There is little technical risk since the standard protocols involved have already been implemented and are operational across several government and university networks. Our hardware requirements include mainly a Series/1, access to VM/370 and access to a X.25 Level 2 (HDLC) ARPANET host interface. We do not require any secretarial support since all our documentation is done using VM/370 composition and VNET distribution facilities. IBM product acceptance in the ARPANET-related community would be greatly accelerated were it not necessary to modify existing (non-IBM) networks and systems to accommodate IBM-unique interfaces. The growth of VNET into the currently largest computer network in the world is a clear example of this principle. VNET has no central administrative control and requires no dedicated hardware or modifications to operating systems. VNET grew as individual problems gave rise to requirements for network access, and initial interconnection to solve such problems stimulated other application utilization. That growth pattern can be extended through development and maintenance of close contact with technical specialists in the government, universities and contractor organizations to establish software facilities that are functionally attractive and very easy to install and maintain. Andy Heller claimed that although Yorktown Research is a physically large facility, it can accommodate small projects and provide any needed degree of isolation. He went on to claim that our physical presence at Yorktown Research would yield a valuable exchange of ideas among other research staff and ourselves. He felt a one week visit to the Systems Laboratory at Yorktown Research would evoke the same perceptions among our group, and he asked us to let his secretary know when we would be available for such a visit so that the necessary arrangements could be made. Both Mike and Ed had prior commitments. Mike was scheduled to give a presentation on software development tools and approaches to a DPD sponsored engineering computing workshop at Monterey. At the request of DPD Scientific Computing, Ed had made arrangements to meet with the coordinator of the Computer Science Network development effort at the University of Wisconsin. Phil explained to Dale and Mike that it was important this trip be made as soon as possible, since he would not explore alternative job placement for us unless and until arrangements for work in Research fail to materialize. Phil went on to say that if we had not accepted placement by year-end we would be placed on the FSD overhead budget, which would mean that if we were not to accept the first subsequent placement offer our employment with IBM would be terminated as mutually agreeable separations. Phil's attempt to cancel or reschedule Mike's talk for DPD was not successful, so the trip could not be scheduled before November 17. Ed had previously scheduled a meeting with Jeff Barnett from Kingston SCD for November 17. (Jeff is working on a project to provide internetworking facilities for local nets of IBM equipment through a gateway to SBS.) This meeting had been postponed once previously, so rather than postpone it again we elected to leave for Yorktown Research on November 18 and return on November 26. Phil informed Andy Heller of the dates by telephone. Andy said he would not be present at Yorktown Research during most of that time, but he would ask Herb Schorr, Manager of the Systems Laboratory, to coordinate our visit. Tour of the Yorktown Research Systems Laboratory Wednesday, November 19 On the morning of November 19 we arrived at Yorktown Research and presented ourselves to Andy Heller's secretary. She informed us that Joel Birnbaum, Director of Computer Science at Yorktown Research, had resigned just that day to take a position with Hewlett-Packard on the west coast. She then located Herb Schorr who asked us why we were there. We did our best to explain our understanding of the reasons we had been invited to Yorktown Research, and we related our understanding that he, Herb Schorr, was to coordinate our visit. Herb replied it was all news to him, and it became evident that no itinerary or other arrangements had been made. We returned to Andy's secretary who contacted Andy on the west coast by telephone. Dale Johnson explained the situation to Andy, who insisted he had asked Herb to make the arrangements and Herb had agreed to do so. Andy suggested that in view of the circumstances, we should arrange to meet with all the senior managers in the entire Computer Science department. During this conversation, Herb Schorr arranged to have his Technical Assistant, M. Wayne Wilson, present to us an overview of the Systems Laboratory. By mid-morning it had been announced that Herb Schorr would take the position of Director of Computer Science, vacated by the departure of Joel Birnbaum from IBM. This meant we had to vacate Joel's office that we had been using temporarily to allow Herb to move in, and that we could anticipate some difficulty in scheduling a meeting with Herb to discuss our situation. Wayne's overview substantiated much of our dubiety toward the proposed move of our project to Yorktown Research. Wayne began by explaining that in response to company concern over Research's business relevance, the Systems Laboratory was created to focus more directly on perceived business needs. Consequently, Systems Laboratory projects are evaluated and prioritized on the basis of the "leverage" they exert on the software product line, and are categorized into three areas: MVS development work, SNA development work and large system performance enhancement. Wayne also observed that because Research activity in connection with development efforts is not included in development costs, the Systems Laboratory produces the side effect of reducing accounted costs of a development project. This briefing continued for five hours and covered all aspects of the Systems Laboratory besides the communications activity, an area Wayne said was unfamiliar to him. He explained that by design, ninety-five percent of the work carried out in the Systems Laboratory is so closely related to strategic product development that it cannot be discussed outside IBM. (Daniel Abensour, Manager of Advanced Networking in the Systems Laboratory, and Irving Wladawsky-Berger, Manager of Systems Organization in the Systems Laboratory, confirmed that figure later in the day.) Wayne went on to say that technical progress is sometimes hampered by the requirement that the work be done on the IBM "strategic" product operating system, MVS, which presents serious difficulties due to its singular complexity. (He characterized MVS internals as resembling "spaghetti.") Of course, a project such as ours which is intended to establish internetwork communication compatible across differing systems in cooperation with government, university and contractor representatives could not be carried out under such guidelines. Our overall reaction to the presentation was that the ARPANET Internetwork Gateway project could not have been started within the Systems Laboratory, and that our expressed doubts concerning prospects for continuation of the project within that organization are well justified. Following Wayne Wilson's briefing, Irving Wladawsky-Berger and Jay Unger, a project manager in Systems Organization, met with us for an informal discussion. Having transferred to Yorktown Research from Poughkeepsie DSD a year ago, Jay expressed enthusiasm for the quality of the working environment he has found in the Systems Laboratory, and he remarked that availability for experimentation of large system equipment provides a very efficient environment for the sort of work he does. We agreed there is much important work that can be performed effectively in centralized organizations, particularly work that provides the "leverage" needed to justify acquisition of expensive current technology hardware. We pointed out, however, that projects such as ours which neither require expensive dedicated equipment nor produce extensive short-term "leverage" are most expeditiously pursued in small decentralized groups. There are many past examples of highly valuable work having been performed in such isolated environments, including at least VM/370, VNET, APL, and HASP. We went on to explain that the only major problem faced by our project is managerial in nature, and that it would make the most sense to address the problem through innovative management solutions. We said that as far as we can understand the situation, Gaithersburg FSD is insisting on terminating their involvement with our project, even though Al Babbitt, FSD Division Director of Technology and Operations, regards the project as offering more than adequate potential business value to justify its continuation. We reported that we had been informed by our management that if we fail to accept new placement by year-end, we would have to accept the first placement offer in 1981 or be dismissed from the company. At that, Jay recommended we contact Al Babbitt and request that the matter be dealt with as an Open Door. We responded that we are reluctant to do that because in doing so the focus would probably shift toward secondary personnel considerations and away from the primary problems which are of a managerial nature. Thursday, November 20 We met first with Daniel Abensour and Mike Conner, a project manager in Daniel's area, Advanced Networking. Daniel began by asking us why we were visiting Yorktown Research. When we explained the situation, he stated his belief that it would be inappropriate to discuss managerial issues with us, but that he would be happy to hold a technical exchange to review our respective projects. Daniel explained the overall layout of the Systems laboratory communications area, described its main purpose as being to improve SNA, and gave an overview of the individual projects in his area which were indeed entirely SNA-oriented. We agreed that rather than go through repeated descriptions of our project for individual members of the Systems Laboratory, we would hold a general presentation for them the following Monday. Mike Conner described in some detail the particular areas of his technical work, which included improvements to the SNA protocol specification language and a gateway function to operate between pairs of SNA networks. Mike explained that this latter project was brought about as a consequence of an SNA design provision that restricts a single SNA network to a size not to exceed 256 addressable entities, which is inadequate in cases now commonly arising. In order to avoid making major changes to SNA itself, the internetwork gateway function has to simulate end-to-end sessions by joining pairs of actual sessions at the gateway, which in turn creates difficult problems with data buffering and pacing. It is often noted that systems in general and computer systems in particular seem to resemble the organizational structures from which they arise. The peculiar state of SNA arises naturally and inevitably from its basic assumptions and objectives, which in turn reflect IBM organizational considerations pertaining to computer communications. Our overall reaction to the Systems Laboratory's activities aimed at correcting the difficulties with SNA was that they tended to suggest symptomatic treatment rather than the reconceptualization needed to overcome the basic problems. It was later pointed out by Paul Green, Manager of Systems Communications in the Systems Laboratory, that in view of the extensive organizational commitment to the existing SNA structure, it is unreasonable to suggest that IBM now carry out any sweeping change to basic SNA concepts. Following our meeting with Daniel Abensour we met with Paul Green. Paul asked that we introduce ourselves and provide an overview of our project. Having listened carefully to our explanation, he observed that the Systems Laboratory has little to offer us as an environment in which to pursue further the goals of our project. He explained that they already had much more SNA-related work than could be accomplished by the existing staff. He speculated that perhaps the intent was that we would transfer into the Systems Laboratory to be reassigned to such work, and he made it clear that we would be welcome to join under those conditions. We explained that our only intent at that time was to seek arrangements under which we could continue our current project. Paul said he knew of no organization within IBM that would sponsor the work, and he suggested that we take the matter to the Corporate Technology Committee, an organization on which he had served in the past. Paul offered to help us schedule a meeting with Bill Bernstein, a CTC staff member and communications expert. Before we began our trip, Floyd Van Scoy had made arrangements with us to meet for lunch on Thursday at the Yorktown Research cafeteria. As the meeting with Paul Green had not quite concluded by the time we were to meet Floyd, Mike excused himself and left to find him. Upon meeting Floyd, Mike gave him an overview of the discussions up to that time. It was pointed out that the visit had not been coordinated and that most of the people we had met were unaware of our purpose for being there. Dale and Ed restated this observation when they arrived for lunch, and related that we had found little interest or enthusiasm toward our project in the Systems Laboratory. Upon further discussion of the purpose of our visit to Yorktown Research, it became apparent that Floyd firmly believed we had been offered positions in the Systems Laboratory. Floyd was evidently disturbed when we told him we were aware of no such job offers, and he said he could not understand how we could possibly believe we hadn't been offered jobs. Ed responded by asking Floyd to name the manager we would work for were we to accept the jobs which supposedly had been offered. Unable to identify such a manager, Floyd then agreed he could understand our reasons for doubt. We spent the remainder of the day trying to schedule a meeting with Herb Schorr, using the library and visiting colleagues at Yorktown Research. Peter Capek, a project manager in Systems Analysis and editor of the IBM VM/370 Newsletter, listened to a description of our circumstances and suggested that we speak with Marc Auslander, a member of the Corporate Technology Committee who happened to be visiting Yorktown Research that day. Peter brought Marc to his office where we met to discuss the situation. We explained to Marc the major theme of our efforts is to plant one foot in IBM and the other in the university computing world, and to work from there to enrich the exchange between the two. We went on to say that while we have found a strong spirit of open-minded cooperation among colleagues from the government and universities, we had been unable to establish any stable working situation in IBM. Marc concurred that our major problem would be with the IBM organization, and he confirmed the remark made by Paul Green that there is no IBM organization in a position to sponsor an effort such as ours. Peter Capek commented that although technical innovation may be cited as a prime objective, our situation illustrates the actual need for management innovation to reach that goal. Marc closed by saying that he couldn't do anything before Thanksgiving, but that he would try to draw the CTC's attention to our problem during December. Mike then telephoned Phil Stuver in San Diego to summarize the status of our visit. Phil voiced concern that the visit had not gone more smoothly, and he asked whether Andy Heller was aware of the situation. Mike replied that we had spoken with Andy on the telephone, that Andy was aware of the situation, and that he planned to be at Yorktown Research on Monday. Mike explained to Phil that we still had been unable to arrange a meeting with Herb Schorr to discuss our project, but that we would continue to try to do so. Mike added that the mutual feeling of our group was that we were seeing an accurate perspective of the attitudes of the Systems Laboratory people toward our work. Friday, November 21 Before we left San Diego we had scheduled Friday morning meetings with Dick Bergstresser at Harrison, and with Sam Matsa, DPD Scientific Computing, at One Corporate Park. Dick was somewhat late for our meeting, having been unavoidably detained by personal business. When Dick arrived, Ed presented an overview of the project progress that had transpired since Dick's visit to San Diego in July. The overview highlighted significant developments surrounding our involvement with CSNET (the NSF-funded effort to develop a network of computer science departments at U.S. universities), and contacts with Jeff Barnett's SBS-related internetworking ad-tech effort in Kingston SCD. We described to Dick our impressions of the Systems Laboratory at Yorktown Research; Dick appeared disappointed that we had not found a commonality of purpose there, and he attributed the confusion over arrangements for our visit to the recent departure of Joel Birnbaum. Following our meeting, Dick asked Mike to stay for a review of a proposed joint effort between the Cambridge Scientific Center and Queen's University in Kingston, Ontario, Canada, to explore applications for array of array data structures. Mike agreed to stay while Dale and Ed left to meet with Sam Matsa as had been previously arranged. During his work with the Navy, Mike had applied some of the techniques discussed at the project review and had found them to be quite useful. The proposed joint effort would further develop notations which provide the data base designer with a concise set of primitives which can be used in non-procedural formal expressions of algorithms. Execution of such primitives can then be optimized on a pipeline processor in much the same way as inner product, the array theoretic operator, has been optimized. Thus the proposed work promises benefits across a broad spectrum of computation, from design languages to hardware implementation. Following the review, the group met for lunch at the IBM Harrison cafeteria where there was some discussion between Dick Bergstresser and Mike of organizational matters surrounding our project. We had been told that Paul Knaplund, IBM Corporate Vice President and Assistant Group Executive, had recently been given the mission to coordinate and enhance IBM's posture in the scientific and engineering computing marketplace, another area of rapid application growth. Mike suggested to Dick that since our group has regularly been involved in consultation to DPD on scientific and engineering computing, and since our internetworking project falls naturally into that area, it would appear to make sense to organize us formally within Paul Knaplund's domain. Were that done, it would help solve the many chronic problems with organizational harmony we've experienced in connection with DPD-related consultation work, and at the same time it would provide organizational sponsorship for the continuation of our project. Mike then spoke with Marty Schatzoff, a manager at the Cambridge Scientific Center representing that organization at the project review. Four years ago Mike had presented his work to a corporate task force examining IBM's weakness in advanced technology activity, and Marty had served on that task force. At that time Mike reported that the single factor limiting his ability to carry out needed advanced technology work was frequent distraction by a variety of short-term tactical business chores. Recalling that exchange, Mike pointed out to Marty that our present situation amounts to a classic example of the same organizational phenomenon. Project funding, head count, competent staffing, strategic importance to the business, technical approach, and progress to date are not at issue by general agreement. The only apparent problem is that the project is located at a remote site away from typical day-to-day distractions, and the project is to be cancelled unless the participants relocate to a large plant site where they would be more available to be distracted. Dick Bergstresser observed that IBM had experienced unusually great difficulty in moving people from the San Francisco Bay area to Dallas, that such increasing difficulties seem to be a developing trend, and that techniques must be found to move work to people rather than people to the work. Mike responded that ours would be an ideal pilot project for such an effort. It would be particularly appropriate for the purpose because it seeks to develop precisely the tools that will be needed to manage physically distributed efforts. Our actual use of those tools as the work proceeds would provide direct feedback regarding their effectiveness, and those experiences would constitute an invaluable aid in choosing further design and development options. Dick then expressed some concern for our future career prospects were our group to remain in San Diego. He wondered what would happen after the project is completed. Mike explained that such concerns might best be addressed at a more timely moment, and that the only area of concern right now is to see the project continued as expeditiously as possible. Any physical relocation at this stage would seriously delay the work and could even cause its demise, and it would be unwise to accept that degree of cost and risk on the basis of speculative concern. Upon completion of the project, we all will be in a far better position to decide what to do next, and complicated constraints arising from the need to continue the project would not have to enter into those decisions. The meeting at One Corporate Park among Sam Matsa, Dale and Ed was the first time Sam and Dale had met. Sam discussed Dale's background and experiences with him, and reviewed the personal disadvantages that a relocation would impose on each member of our group. Having done so, he acknowledged that a permanent move would seem not to be advisable at this time, and that there is no substantial reason for not continuing the work in San Diego. Sam went on to question what the three of us could expect to be doing in San Diego after several years, when the current project activity may be winding down. After considering all the aspects of the problem, Sam observed that the only suggestion he could offer would be a temporary attachment to the Research Division for six months or so, to be followed by a reevaluation of the situation in mid-1981. The meeting closed with a review by Ed of the status of his CSNET activity and his upcoming trip to the University of Wisconsin to meet with Prof. Larry Landweber, CSNET coordinator, and hold a colloquium on VM/370 and VNET. Following lunch, Dale and Ed returned from their meeting with Sam Matsa and we left Dick Bergstresser's area for our cars. We were joined by Marty Schatzoff and Mike Jenkins, Chairman of the Information Science department at Queen's University, who represented his department at the project review. Marty introduced Mike and Ed, mentioning that Ed had been a major developer of VNET. There ensued a brief discussion of networking, in which Mike remarked that some form of networking facilities in operation between Queen's University and IBM would be greatly useful to his joint project. Ed briefly described his emerging role as IBM technical liaison to CSNET, and Mike responded with great interest in establishing Queen's University participation in that effort. Ed promised to relate Mike's interest to Prof. Landweber during Ed's visit to the University of Wisconsin on his return trip to the west coast. We returned to Yorktown Research that afternoon, where Ed had an appointment to meet with Jerry Anderson, Research Division Director of Personnel, concerning the organizational situation. Mike worked on a load balancing algorithm, and telephoned Phil to report on our visit to that point. Phil voiced concern that the visit had not been going more smoothly, and requested that we ask Andy Heller to telephone him Monday morning. Monday, November 24 The following Monday we repeated for personnel in the communications area of the Systems Laboratory the presentation of our internetworking project that had been given as a Research Seminar at Yorktown Research several weeks earlier. We were told that they had been unaware of our earlier presentation, as they do not regularly read the bulletin boards. Questions from the audience following the presentation revealed some significant differences in viewpoint between their group and ours. One such line of discussion questioned our approach to internetworking that would require some functional extensions to the participating systems. The major area of concern seemed to focus on the point that SNA would have to be modified in order to participate in internetwork communication, rather than a suggested alternate approach that would build extensive function into the gateways so that such modification could be avoided. The audience appeared to have missed the basic point that our project is intended to develop an IBM capacity to adhere to a major internetworking standard emerging outside IBM, rather than to establish an independent and incompatible approach to the problem. This session reconfirmed our strong impression that our approach to technical work would not harmonize with the thinking characteristic of the Systems Laboratory. The internetworking technology evolving under the auspices of DARPA is the product of a great deal of joint work and experience among many universities, companies and government agencies. While we might eventually move into a position to present functional design of our own, the only way we could realistically expect to reach such a position would be to begin by accepting the work that has already been done. Any attempt to assert functional requirements or constraints prior to demonstrating IBM's ability and willingness to accept commonly agreed-upon interfaces would certainly be perceived as confirmation of IBM's poor reputation as a cooperative participant in professional computer science activities. Paul Green, Floyd Van Scoy and a number of others who had attended our presentation joined us for lunch at the Yorktown cafeteria. It was during this lunch that Paul observed that it would not be reasonable from an organizational point of view to expect IBM to be able to make any major shift away from established SAN concepts. After lunch Mike again called Phil to summarize our activity, and Phil again asked Mike to try to get Andy Heller to call him. We had been told that Andy was expected to be at Yorktown Research Monday morning, so we went to see his secretary to try to locate him. Andy had not yet arrived, having spent the preceding night traveling to the east coast, so we left a message with his secretary asking him to call Phil when he arrived. As we were unable to arrange meetings with either Herb Schorr or Andy Heller, we attended a talk given by Mike Blasgen entitled, "What is a Researcher Doing in Washington D.C.?" Mike broadly described his current assignment in IBM's government liaison office. Of particular interest to us was Mike's discussion of joint IBM and government data transmission policies and agreements. Questions in this area often arise in the context of our work, and Mike might be able to offer us valuable advice regarding them in the future. Following the presentation, Mike Engel happened to encounter Andy Heller in the hall by Andy's office. Andy said he had some time to talk, and the two were joined in his office by Ed who was also in the vicinity. In all our discussions with Andy we have experienced unusual difficulty in maintaining focus on a particular matter long enough to achieve any substantial exchange of ideas. During this meeting both of us tried repeatedly to offer Andy our thoughts and impressions regarding our visit, but each time the topic would spontaneously shift to matters that were not germane to our purpose for being there. Andy exhibited a great deal of enthusiasm in relating a range of issues that dealt with his activities and interests, from encoding a certain SNA protocol on a single chip to the purpose of the lounge chair in his office. After several attempts by Ed and Mike to explain to Andy that our visit had served mainly to confirm our doubts regarding the proposed transfer, and that we still could not understand any reason why we should not simply continue our work in San Diego, there was some response. Andy repeated his assertion that he had gotten Herb Schorr's agreement to arrange our visit, and so the confusion surrounding the situation was not Andy's fault. He went on to explain that we couldn't continue our work in San Diego because management would be unable to support us adequately. He illustrated the point by making reference to all the people at Yorktown Research who would be available to configure equipment and to locate wires in the walls for us. He appeared unconvinced by our assertion that we are entirely capable of doing such things for ourselves, and that our doing so in the past had often provided us with background knowledge and insights that had proven valuable in themselves to our work. Tuesday, November 25 This was Mike's final day at Yorktown Research, and it started off with a bang. As we arrived in the area of the office we'd been using temporarily, Mike was buttonholed in the hall by Daniel Abensour, Irving Wladawsky-Berger and Frank Moss, a project manager in Systems Communications. They said Andy Heller had relayed to them what we had told him concerning our visit there, and that they felt our statements were unfair. Irving said that from his viewpoint we had been "stonewalling" them, that our observations concerning Yorktown Research and the Systems Laboratory were merely an expression of our reluctance to leave San Diego, and that we ought to adopt a more flexible stance because FSD has placed us in a position of compromise. Mike responded that from the beginning our position had been that moving our project to Yorktown Research would involve considerable risk on many fronts, and that any form of success would be far from assured. Assuming the project is in fact of value to IBM, any means of reducing these risks should be sought. One such means might be to establish a temporary working arrangement with Research to avoid an immediate requirement for physical relocation, which would explicitly state the goals and business purposes of the arrangement. Daniel replied that he was anxious to discuss possible accommodations, but that he had pressing business to attend to first. He suggested that we speak with Frank Moss until he could join us. Frank began by saying he couldn't understand our concern regarding the particular organizational environment in which our project is to be carried out. We repeated the explanation that systems work inevitably reflects the qualities, values and priorities of the organization from which it emerges, and that there are plenty of specific examples of pressures giving rise to the effect. We went on to repeat that the clearly stated purpose and easily observed quality of the Systems Laboratory at Yorktown Research is at odds with the nature and purpose of our project, citing explicit reasoning. Having beard that argument, he acknowledged he could understand our concern and that it had been properly raised. Frank then presented an overview of the technical problems in SNA being addressed by the Systems Laboratory, and he gave us several papers to read on routing and flow control. We declined to accept the IBM Confidential documents he offered due to difficulty in handling them on business trips. We then met with Daniel Abensour, and were joined shortly by Irving Wladawsky-Berger. Daniel said he regretted we had not been better received, and that he was preparing to pursue two types of offers: 1) a permanent transfer and relocation to Yorktown Research; 2) a six to eight month assignment on a per diem basis. He asked us to inform him of any decision we might reach concerning his informal offer, and we agreed to do so. Leaving Daniel Abensour's office, we walked to Herb Schorr's area and met Peter Capek who checked to see if Herb might have time to meet with us. Herb did have an hour, and we all joined him in his office. Ed began by describing the history, objectives and status of our internetworking project, and went on to relate the current difficulties with FSD's posture toward it. Herb recommended that we again contact Al Babbitt to verify that his directions are being carried out. We explained to Herb our reasons for being reluctant to move the project, and asked if he could help us arrange to continue the project in San Diego under the auspices of the Research Division. Herb expressed some understanding of our position, but said he was pessimistic that such arrangements could be made. He said it might be possible to work out an interim agreement for continuation of the project for eight to ten months, but that some explicit objective or purpose would have to be established in order elicit concurrence from Ralph Gomory, Research Division President. He doubted that any temporary arrangement could be made unless we could provide assurances that we would relocate to a Research Division site at either San Jose or Yorktown Heights at the conclusion of the temporary arrangement. Wednesday, November 26 As the purpose of our trip had been accomplished, Mike left to return home late Tuesday afternoon. Dale and Ed returned to Yorktown Research the next morning to meet informally with several colleagues there. We spent the late morning with Peter Capek, and the three of us had lunch at the Yorktown cafeteria. Peter said that he shared Herb Schorr's and our doubt that a temporary assignment to continue our project at Yorktown Research would be a good solution to the problem, that he understood and agreed with our reasons for believing we should continue our project in San Diego, and that the best solution would be to find some means to do so. Dale suggested that perhaps the San Diego DPD branch office could take on managerial responsibility for the project. Peter replied that he felt such an arrangement would amount to a good solution, and he recommended that we propose it to the DPD Branch Manager on our return to San Diego. Shortly after lunch, Andy Heller's secretary gave Ed a message that Andy would like to speak with Ed on the telephone as soon as possible. Ed immediately called Andy, who presented an informal offer for a temporary assignment to continue our project at the Yorktown Research Systems Laboratory. The assignment would last until June 1981, at which time we could permanently transfer and relocate to Yorktown Research, or else we would revert to the same situation we now face. Ed replied that such an arrangement would seem to meet his personal minimum requirements, but that he would have to discuss it with the other members of the project and many detailed issues would have to be settled before such an offer could be accepted. Ed said he would discuss the tentative proposal with Mike and Dale when all three had returned to San Diego, and further responses would follow. Conclusion We must conclude that we have not yet found an organization willing to accept administrative responsibility for continuation of our project in San Diego. Our overall judgment is that the Systems Laboratory at Yorktown Research is not an appropriate environment in which to carry out our project as it is currently conceived. Perhaps five to ten years ago there existed justification for moving a project such as ours in order to centralize equipment resources for purposes of efficiency. The computer industry has since evolved such that equipment costs are dramatically reduced and high quality network communication is readily available. Unfortunately, it appears that IBM management techniques have not similarly evolved to take advantage of these advancements. If IBM is to maintain an ability to attract and retain highly competent professional personnel, new management techniques must be established to free individuals from the rapidly increasing burden of repeated physical relocations, many of which result from nothing more substantial than relatively minor changes in organizational posture and direction. We believe there is nothing standing in the way of the establishment of such techniques besides force of habit. Gernot M. Engel Edson C. Hendricks Dale R. Johnson December 19, 1980 (amended 1/15/81 -- E.C.H.)