call this information up from the files using terminals to upgrade the cataloging. The present procedure requires filing temporary cards in a card catalog, keeping cards with each book, upgrading the card as the descriptive cataloger, subject cataloger, classifier and shelflister all do their necessary work.

In the future, working from the process file, each of these activities can be done with a great savings in time and in space now occupied by card files.

Here we are speaking for 4 positions, 2 GS-9 programmers, a GS-7 programmer and a GS-6 programmer, automation wise, to maintain and keep that file up, get the total cataloging files in working order, and also at the same time to support the Order Division automated system.

We now have book ordering procedures under automated control. Mr. BENJAMIN. Are you leaving us on bibliographic at this point? Mr. APPLEBAUM. No, sir. I am still on page 63. Skipping to the 2 positions for the multiple-use MARC system, Mr. Howard mentioned name and subject authorities. It is vital for the cataloging procedure to have in machine-readable form the name and subject authority file. This is the list of main headings, the authors, which are in our records, so that as new books are cataloged, that cataloging can fall under the appropriate heading. Otherwise we might have books filed under Library of Congress, under U.S. Library of Congress, under any possible variation, Congressional Library, for example, or variations in personal names.

MULTIPLE NAMES PRESENT PROBLEMS

Mr. WELSH. Here is an example of our problem, Mr. Chairman. There are almost 50 Hans Muellers on that list. Each name represents a separate author. One of the problems that a cataloger finds is trying to associate the particular author with a particular publication so that is an example of the authority file, that there are that many different Hans Muellers in the collections of the Library of Congress. Every time a publication comes in by that name, you have to determine which one of those Hans Muellers it really is. Mr. APPLEBAUM. The remaining two positions are a GS-13 telecommunications specialist and a GS-5 clerical assistant to help in the planning and development of a national linkage of libraries with bibliographic data bases.

WHAT SECURITY MEASURES ARE USED

Mr. BENJAMIN. With regard to that nationwide linkage of libraries, my question is then: What steps are being taken to prevent unauthorized access to library data from unsupervised computer terminals?

Mr. NUGENT. I understand there have been questions about the CRS file of Member's requests. Within the on-line files there is no information concerning Members and their requests. In the SDIF file containing information privileged to the CRS, such as management information on the activity of the service, there are the names of the people who prepared those reports, but there is no identification connected with the member.

The only terminals to this file are hard wired, that is to say, one distinct telephone line coming in from one distinct terminal, so we always know that is the proper source. No dial-up terminals, those that use the common telephone, can access that data base.

The other sensitive file that we have are files of the Senate Foreign Relations Committee, For over a year they have been using private passwords that they themselves assign so no one who is not privy to their password or their terminal identification may get on those systems. We have had this password system in existence for that one group only for about a year.

As of March 15 we are going to make the password system uniformly apply for dial-up terminals within the library. We have heard from our counterparts in the House and the Senate that there is a strong thrust to insist upon the use of passwords by dialup terminals in the House and the Senate.

So we feel by this method we will have a system that is secure from what we would call normal prying. It would not be secure from, say, an intelligence agent who wanted to tap the lines, but from all normal curiosity seekers and pryers this would be as safe as we can make it.

Mr. BENJAMIN. I guess I would analyze your reflection that there is a security problem?

Mr. NUGENT. Only in the DOD military intelligence sense, not in the normal sense of prudent file security, sir.

COPYRIGHT APPLICATIONS

Mr. BENJAMIN. Continue on.

Mr. APPLEBAUM. Three positions for copyright applications. These positions were appropriated for the Copyright Office in the supplemental appropriation to implement the copyright revision bill for the purpose of supporting the automated copyright applications and this is merely a request for a comparative transfer of these funds from the appropriation for the Copyright Office to the appropriation, Salaries and Expenses, Library of Congress.

CENTRAL COMPUTER OPERATIONS

The major request is for 25 positions for our central computer operations. At the present time our computer service center is operating pretty much around the clock, three shifts a day, seven days a week, one shift on Sunday night being for preventive maintenance. However, some of the late shifts are very thinly staffed, two or three people, and given the enormous increase in the workload and the necessity to have people handle this large flow of work to accomplish more work on each shift, we are asking for eight computer operators, one senior tape librarian, one telecommunications specialist to do performance checks and preventive maintenance tests, one customer service coordinator, who will work with all of our users in the Library and on Capitol Hill who have direct teleprocessing equipment, and two other customer service specialists, again working off-shifts to meet the needs of our users, one equipment configuration specialist who will monitor and control the ordering, installation and management activities in order to assure that we are getting what we require and are assigning it

properly, and one scheduling specialist to handle the increased batch-processing workload and to improve scheduling activities as they relate to customer services.

The amount of work that is being undertaken by this group-I think last year I used the expression, nothing succeeds like success-is an indication of what is happening. This chart shows the number of terminals connected to the Library of Congress computers. We began teleprocessing work in Fiscal 1974 and as you can see, the Senate initially got on line before the House, but the House by 1977 had passed the Senate in terms of number of terminals connected to our computers, and we feel by the end of Fiscal 1978 we will have close to 1,000 Library of Congress terminals on line, 350 House terminals, 150 Senate terminals and 25 terminals from other legislative agencies.

QUANTITY OF TERMINALS GROWING

By next year, by the year of this budget request that we are speaking to, we expect well over 2,200 terminals to be hung on these computers. It is an automated age. I don't know whether you had occasion to see the most recent issue of Time magazine. I have it here. It's main story is The Computer Society, and this is truly what we are in. When I was going around picking up some of that information at the American Library Association meeting, people were wearing, pins saying, "Library Automation Is Here." It truly is. There is no turning back from it. We either do it and do it well or we do it and do it poorly, but I don't think we can ever return to the manual systems.

NUMBER OF COMPUTERS

Mr. BENJAMIN. How many new computers do you have?

Mr. APPLEBAUM. New computers-we are presently hoping to upgrade one of our computers. We did get a computer upgrade about a year ago. We have two large central processing units in the main computer service center.

Mr. BENJAMIN. They are what?

Mr. APPLEBAUM. One is an AMDAHL 470 V/6; one is an IBM 370/158. The Cataloging Distribution Service has a smaller computer, an IBM 370/135, at the Navy Yard for its current order work and card production work. There are also four mini-computers in use by various elements of the Library.

Mr. BENJAMIN. The on-line system, what do you use that for? Mr. APPLEBAUM. That is the teleprocessing system; that is the online system, providing high-speed, on-line communications. The other system, the 158, is the batch system and both presumably are providing some backup for each other.

Mr. Nugent may be able to expand on that.

Mr. NUGENT. Our batch services are running now three shifts per day, seven days a week, with the exception of one shift for preventive maintenance.

Mr. BENJAMIN. What do you mean by "batch"?

Mr. NUGENT. This is a programming service where a user will request that a job be run and anywhere from two hours to a day later he will get that job back so that the scheduling can be left up

to the Computer Service Center. This is in distinction to what we call on-line systems, where there is an immediate interaction and immediate response from the computer, and these are always conducted from terminals.

Mr. BENJAMIN. Let's go back to that for a minute.
How many jobs do you have under batch for 1977?

Mr. NUGENT. A job, sir, is not a good metric for this system.
Mr. BENJAMIN. Any measurement you want.

Mr. NUGENT. Basically, we have been growing in excess of about 50 percent per year and between 1975 and 1976 it was in excess of 100 percent.

Mr. BENJAMIN. Are you talking of personnel?

Mr. NUGENT. I am talking of workload. The reason I say that jobs are a difficult measure is because we have one job called CICS, which is the entire teleprocessing workload and that just counts as one job. Similarly, a system called ROSCOE, which handles all the on-line programming which is done on a batch computer, is also called one job, and similarly what might be a 30-second run just to produce a single page is also called a job; so a job does not really tell you what the workload is. It is really the CPU utilization that is an indication of the workload, and right now our CPU, during prime shift hours, is almost invariably 100 percent utilized. In other words, there is zero capacity left in that machine to produce any more work.

CAPACITY EXPANDED BY UPGRADING MACHINES

Mr. BENJAMIN. If you have zero capacity, the point that you are at in your prime time, how can you have any more employees? Mr. NUGENT. Well, this is the machine that we expect will be shortly upgraded, so that we will have more capacity.

Mr. BENJAMIN. Upgraded by how much?

Mr. NUGENT. By a factor of four to five.

Mr. BENJAMIN. Four to five?

Mr. NUGENT. In terms of capacity.

Mr. BENJAMIN. What is your capacity now of the machine? Mr. NUGENT. This is in terms of through-put of jobs and typically we would run in excess of 10,000 jobs a month.

Now a batch job can be measured at full utilization by how much CPU time it consumes, so one of our difficulties is right now, during prime time we end up building up a queue of work that will not get processed until after prime time.

Mr. BENJAMIN. What agencies get your prime time?

Mr. NUGENT. Basically it is the Library of Congress that we serve. The batch processing is entirely devoted to Library and Congressional Research Service.

Mr. BENJAMIN. Give me an idea in the Congress of who requires batch processing.

Mr. NUGENT. Largely through the CRS. We obtain requests that are part of Member requests. It is part of the way the CRS satisfies Member requests. We have a particular kind of job called Code Red jobs, for instance, that must be run in one hour because these are congressional demands, so this is where our prime time is being used.

We also have things that we run in off hours, such as payrolls, usual administrative management information and things that we cannot normally process during the daytime hours. Batch processing is running 24 hours a day on most days. This is in contrast to the teleprocessing system, the on-line system, if you will, which runs roughly 87 hours a week. During weekdays it is available from 7:00 a.m. in the morning until 9:30 p.m. at night. That, in turn, creates additional batch loading because the more data bases that we put on for on-line access, the more they have to be updated, which is a batch process; the more maintenance they need in changing additions which are batch processes, so there is a strong interrelationship between the two, even though we divide them into two distinct services.

BATCH PROCESSING EXPLANATION SOUGHT

Mr. BENJAMIN. What is your initial input into your batch? How do you start this off?

Mr. NUGENT. This is a request for programming services that one of the authorized users of the Library would send over. In some cases it might be a totally new program. In other cases they would say, "Please run this job with these new parameters", so they would submit a job control card and that would be the initiation. that would then go into scheduling. They would tell what kind of a job it was, how long it was expected to run and if it was a short job we would squeeze it in or attempt to. If it was a longer job we would schedule it for the nighttime.

Mr. BENJAMIN. Are most of these jobs just recitation of information you already have, or is it an evaluation of data?

Mr. NUGENT. It is, I would say, combinations of information we already have, because we have such a large data base collection that often it is the assimilation of information from different places that creates the value of it; in other words, it is selective rather than just an exhaustive listing of everything.

What a researcher will try and do is to say, "Don't just tell me about energy; tell me about energy specifically concerning solar, and specifically concerning the heating of houses," and all of those will come together to create useful information.

PROGRAMMING EFFORT

Mr. BENJAMIN. On an average, how many new programs do you write a day?

Mr. NUGENT. A major program normally takes a team of two or three people several months to finish.

Mr. BENJAMIN. You are talking about a very comprehensive program, aren't you, at that point?

Mr. NUGENT. I am talking about a typical production-type program that does a job that one would expect to endure for some time. In other words, it wouldn't be a one-shot process; it would be something that would be used many, many times. We do not write programs on a one-shot basis.

Mr. BENJAMIN. When you say you don't write them on a one-shot basis, I imagine you maintain them if they have to be used again,