The concept of an "on-line, total system" has been the dream of every data processing specialist for the effective use of his equipment. Since it represents probably the most significant next step in the introduction of data processing technology in the library, this concept and its implications should be clearly understood by librarians. It combines two distinct, though closely inter-twined developments in the data processing field: the on-line system, and the total system.
The first is exemplified by the typical automated airlines reservation system. A clerk is able to communicate directly with the computer, through a "console" or 11input11 unit, which is physically remote from the computer but designed for easy communication between the computer and the person using it, The most important feature of the on-line arrangement is the rapidity with which communication can take place--usually so fast that the person is able to operate at his own rate of speed. The immediate practical result is the elimination of many routine steps previously necessary in data processing operations, particularly the handling of cards, tapes, and forms. Of greater significance is the possibility of using the computer in ways previously impossible, particularly for jobs which were too small or too specialized even to be considered.
The development of the total system has been a continuing theme in the data processing field: "Include as many of the clerical tasks as possible in a mechanized system, so that all of the desired results can be achieved with a minimum of routine effort outside of it--and particularly of the effort required to put data into it." The result is a total system, which relieves the user of all routine work and allows him to devote his energies to judgments and creative work. In combining the two developments, the data processing specialist visualizes not only performing all routine, clerical effort by a mechanized system, but making it easy to ask it to do so, even for very small tasks.
In the library situation, the total system idea has seemed particularly appropriate. If the descriptive cataloging information could be typed once, directly into the computer at the time of ordering, it could form not only the start of the acquisition record, but of the catalog record as well. With an on-line system, the records could be directly corrected, if necessary, upon receipt of the book itself. Similarly, if subject cataloging information could be made available to an automated searching operation as well as for the automated production of a printed catalog, single entry of it would do multiple duty. If circulation records could be automatically generated, entered directly into the computer at the time of charging, automatically maintained and reconciled, and then utilized for analyzing the activity of the collection--all based on a single entry of reader and book identification--the dream of a total system is close to reality. If, in addition, all of these processes could be carried out by direct communication between the computer and the people involved, the result would be a truly integrated system,
Each of these steps has been introduced in one library or another; each is well proved from an operational standpoint, if not an economic one. And several proposals have been made for implementation of the complete concept, the most notable of which is that for Automation and the Library of Congress, a study by Gilbert W. King (and others) published by the Library of Congress in 1963. But how do we stand today in the realization of the data processing specialist's dream? Is it at all practicable, especially at the level of the small library?
International Business Machines, Inc., has been experimenting with such a library system (diagrammed in Figure 1) at their Advanced Systems Development Laboratories in Los Gatos, California. Because the system is still under development and implementation, it is not appropriate to speak of it yet as a total library system.
Of the parts of a total system described in Figure 1, the ordering and cataloging parts are in operation "on-line.'' They are therefore of particular interest, because they represent probably the only existing library functions which are now in operation by direct communication with a computer. Order and cataloging information, both descriptive and subject, for books and reports is typed using a "terminal"--a typewriter similar to the IBM Selectric Typewriter--which transmits typed information directly into the computer.
The order information is written either by the requester or by the order librarian on a single, multi-purpose card (Figure 2). The information on the card is then typed onto a worksheet (Figure 3) using the terminal. The format of the worksheet defines for the computer the significance of each piece of data typed on the form. The data is communicated directly to the computer (an IBM 1460), and stored in a magnetic disc, direct-access memory. The data then is available for output, reference, addition, or change at any time during the day.
At the end of the day, the information is stored, at present, on punched cards which are then converted to magnetic tape. (This last step will undoubtedly be eliminated in the near future, and the information will be stored directly on tape,) Various clerical functions are then automatically completed by the IBM 1401 and IBM 7090 computers. These include the printing of orders and notices to patrons that materials requested are available for circulation (Figure 4).
Upon receipt of the book or report, the descriptive cataloging information is corrected, if necessary, to conform with the characteristics of the item actually received. The subject headings and descriptors or index terms are added, The conversion sub-system, designed to transfer existing manual records to machine form, also uses the worksheet for input data and was the first "on-line" operation. The collection at Los Gatos consisted of about 6,000 volumes; the manual records for over 5,000 of these have been converted to machine form in less than a year.
Spine and book pocket labels, shown in Figure 5 (page 7 ), are printed by the computer and an initial set of circulation cards (five for each volume) are punched. Acquisitions information is also printed as part of a daily newsletter. The other portion of the newsletter, which consists of announcements, current periodical articles of interest, and new periodicals in the library, is printed by the terminal. The unique features of the terminal permit editing at the time of typing, ease of error correction, and right-hand margin justification for printed text.
The circulation system has some special operating features. The five cards printed out as a result of the initial cataloging are all placed in the pocket of the book or report. The first card shown in Figure 6, is used to enter the pre-punched book identification into the computer for the first circulation. The second card is used to reconcile the records upon the return of the book. The third and fourth cards serve comparable functions for the second circulation. The fifth card is used to enter the identifying information on the third circulation; it contains a special punch (an 11011 in column 80) indicating that a new packet of circulation cards should be produced by the computer. Upon the return of the book, the new packet of five cards is available at the circulation desk for successive circulations. Each day a complete listing of all volumes in circulation is printed, sequenced by author. A separate listing for reserve requests, also sequenced by author, is printed indicating the names of the individuals requesting each book or report. These listings are used to check on the availability of material, for placing a reserve, and for notifying the holder of a book that it is wanted by someone else. When a reserve or return is requested by a borrower, a card (Figure 7) is punched with his identification, and entered into the computer, The three-part form shown in Figure 8 is produced, The first part goes to the present borrower of the book asking for its return; the second goes to the requester notifying him that the return of the book has been requested; the third goes to the requester, upon the return of the book, notifying him of its availability. The computer also maintains a check to see whether or not the original borrower returns the book for the requester by the due date. On a monthly basis, the form in Figure 9 is printed out by the computer. It lists an inventory, by person, of each book with which he has been charged.
Present circulation averages a total of 5 to 10 per day. Such a low circulation figure makes it impossible to arrive at any valid conclusions concerning the utility of the circulation procedures.
Moreover, this circulation system appears to be somewhat cumbersome and is probably subject to many operational difficulties, It should be recognized, however, that the system as a whole is certainly not dependent upon the particulars of this experimental approach to circulation. Any of the well-proved book card or transaction card systems could be used equally well.
A serial control sub-system is under development. It has been delayed by the need to complete the set of holdings records. Unfortunately, this has been done without the use of the computer. When the serials sub-system finally is introduced, the holdings records will need to be completely converted. The intent of the serials sub-system will be to provide automatic renewal of periodical orders, automatic generation of claim notices, control of binding, printing of routing slips, printing of a title-arranged listing which includes holdings, and printing of listings by subject or other special arrangement. There is no attempt at present to provide any indexing of journal articles other than manual indexing, although plans are in process for both indexing and a new information retrieval system.
The entire area of information retrieval is outside the scope of the present system. This last point deserves some emphasis. There is a tendency to identify the use of data processing in the library with the automation of complex intellectual processes, such as "information retrieval." Librarians have been as prone to do this as the most avid machine enthusiasts. Yet the fact is that use of data processing on clerical operations-even in an on-line, total system--is completely separate from the issues of information retrieval, unless the librarian wants to include the latter. (Incidentally, some on-line retrieval systems are also in experimental use-specifically in the Lockheed Technical Library in Palo Alto, using an RCA data input and computer system, and at Massachusetts Institute of Technology in their project MAC.)
The equipment involved includes the IBM 1460 as the computer with which the input typewriters communicate, and the IBM 1401 and IBM 7090 for all the other types of processing. The IBM 1311 "disc-pack" storage unit is used as the direct access memory.
It should be emphasized that this system is intended as experimental, and primarily for the purpose of experimentation with the equipment. It therefore does not represent an ideal library system in any sense, nor is it intended to do so. However, it does illustrate several of the significant aspects which must be considered by anyone planning to develop or to introduce a total system.
Specifically, the system developed at IBM Los Gatos is presently possible almost solely because of the unique situation of a computer development center working on this type of on-line equipment which is interested in its potential application to library operations: It requires the availability of a computer capable of on-line operation. Since it utilizes an almost infinitesimal portion of the machine's capacity, the library must be chargeable only for that small portion of the capacity which it actually uses. The system as designed is appropriate only for a very small, special library operation. The problems in conversion and reconciliation of records are substantial and will delay significantly the implementation of any system, despite the fact that, in the case of IBM Los Gatos, the system itself has been designed to facilitate the conversion.
On the other hand, it must also be recognized that the approach of direct communication from typewriter keyboard to computer will probably allow the typical library to take advantage of a centralized computer to an extent that no other approach will. The day of "service centers" providing the use of a computer, virtually as a utility, is not far distant, Libraries may well find that when it arrives, all the dreams about the value of computers will have come true.
As has been indicated, the concept of a total system is highly valued by the data processing specialist, and rightly so, since the more a system can be designed as an integrated whole, the more efficient and economical it can be, However, there are basic problems in trying to achieve this integration. First, no system can be designed as a completely integrated one: the realities of system design as a technical problem are such that a system must be analyzed into component sub-systems, each of which is then designed separately. At best, the analysis into sub-systems is made in light of the over-all goal. This is precisely the path which the development at IBM Los Gatos has taken.
Another problem is grounded in the psychology of the designer and of his desire to implement a successful system. It arises in its worst form when the designer has been unable to develop an economical system—one which will even be competitive with the present ways of doing the same job. He immediately searches for ways in which his system could be augmented with other jobs and functions which would utilize the same capabilities, the same data, and the same equipment in such a way as to make the resulting extended operation a competitive one. Almost without fail, the results are even less competitive, and the designer continually extends the scope which he must cover in order to be efficient.
Those whose profession is that of system design will protest that such an eventuality is the exact antithesis of good system design--and they are correct, Unfortunately, it is not the antithesis of the approach of many a system designer, and it especially characterizes those who most loudly sound the call of "total system" design.
In summary, the day of the total on-line library system is still ahead of us. There are many cautions and pitfalls of which the library contemplating such a system must be aware, Not the least of these is the simple problem of poor design approach, in which the "total system concept'' is used as a-convenient escape-hatch from uneconomic systems.
On the other hand, the operation at Los Gatos should prove that the on-line library system is no dream. In general, the development of input devices, of adequate methods for digital communication over telephone lines and microwave, of computers capable of serving a large number of users simultaneously--all have been achieved and demonstrated in library operations. It is still necessary to make these devices known to librarians in applications suitable to the individual library, and to demonstrate their economic feasibility in particular cases. The manufacturers are developing ever more versatile computer systems which can be adapted to an ever widening range of situations. The development of computer "service centers" will almost certainly meet the requirements of economic practicability, even for the smallest libraries.