Finance - Requirements fer
the Records and cOntrol of a Business
J. W. PONTIUS
Consultant - Electronic Data
On behalf of Accounting Services, and the financial function in the
Operating Departments, I 'Would like to thank you, Mr. Oldfield, and your
associates for the opportunity to participate in this Industrial Data
We are greatly interested in the data processing techniques and systems
which your new organization will be developing. As time goes on, we look
forward to a close, working relationship with your group which, we are
certain, will result in lower operating costs, improved data handling and
information systems, and greater profits for the operating components.
Although the large attendance at this meeting is indicative of the high
interest in the developments which will come out of your organization, let
me reassure you also of the very high
interest in your group on the part of the financial people in the
operating departments. During the many regular visits by members of Office
Procedures Service to operating departments and also
during visits by operating people to our headquarters in Schenectady
for the four months that your organization has been established, the
discussion nearly always has gotten around to the nature of developments
which can be expected from your group. So, we look forward to the message
which we shall receive this afternoon concerning your future plans to meet
business data processing needs.
In our approaches to these needs since the idea became feasible several
years ago, we have tried to be as practical as possible and think in terms
of equipment 'Which is presently available or known to be available as
soon as procedures could be made ready.
I heard a story recently which illustrates better than I can this
matter of being practical. Two men were taking an aptitude test, one an
accountant and the other an engineer. One of the questions on the exam was
this: Suppose you were on one side of
a room and a pretty girl was on the other side of the room. If you
could go only half the distance each time you started toward her, could
you ever reach the goal? Recognizing this as an old chestnut, the engineer
gave the correct mathematical answer, "No, never." The accountant
took a more optimistic view of the situation and said, "No, but I
will finally get close enough for all practical purposes." Now there
isn't any question about the viewpoint of that fellow being practical and
I shall do as well in keeping my comments down-to-earth this morning.
In discussing the business data processing needs as we see them, -the
subject upon which you have asked for comments today, I would like to
divide my comments into three principal parts:
First, a brief review of progress with electronic or automatic data
processing systems in the Company to date.
Second, a few comments of experiences and the lessons learned from
these systems and an appraisal of their relation to present needs.
Third, a short look at the future and the requirements of the period
beginning, say five years from now - the early 1960's and beyond.
Excluding the four large IBM 701 and 704's presently used by
Engineering people principally for the solution of mathematical problems,
there are now three large and fourteen medium sized automatic data
processing systems installed in the Company. The large systems consist of
one Rem-Rand UNIVAC and two 702's. The medium sized systems are all 650's.
The equipment already installed represents somewhere between five and 10
per cent of the commercially available automatic data processing
equipments which have been manufactured in more than single copies.
Excluding again three additional 704's, there are now firm orders
placed by operating components for one large system - a 705 - and
seventeen medium sized 650's, including one 650 on order
by a department of Canadian General Electric.
Although most departments are giving attention to the possible future
use of electronic equipment, we know of no other systems on order.
This review of the systems on order, shows the clear trend away from
the large, more versatile equipment - which was available first - to the
much less expensive medium sized equipment. It is also quite evident that
all of these systems are of the general purpose variety - that is, they
can be applied to a wide variety of problems and applications.
As is well known, the commonest application is the preparation of
payrolls and related reports. Perhaps somewhat more surprising are some of
the totals now being rolled up by the equipment. The payrolls of 22
operating departments with nearly 70,000 employees, or . nearly one-third
of all employees, are now being handled on the
UNIVAC, the 702's and the 650's. The first of these payrolls was taken
over about 20 months ago by the equipment.
With one exception, the number of instructions carried out by the
equipment automatically in a payroll application ranges from a low of 1000
to 2000 instructions for that portion of the work which can be done by the
medium sized equipment to 15,000 to 20,000 instructions performed by the
large scale equipment in handling a much greater proportion of the payroll
work. For a department of, say 3000 employees, the average running time
that portion of the payroll covered by these instructions
varies from about eight hours per week for the medium sized equipment to
two or three hours per week for the large equipment. Improvements in these
times are expected, of course, as personnel are able to take better
advantage of the equipment's capabilities.
One of the most interesting by-products which will probably come out of
this early payroll work by the Company is the acceptance of quarterly
social security reports in the form of magnetic tape by the U. S.
Department of Health, Education and Welfare, commonly known as the Social
Security Administration at Baltimore. Reports for the first quarter-of
1956 were submitted both on magnetic tape and on the customary paper
forms. No discrepancies were disclosed and the only barrier to
discontinuance of the old paper reports is final approval by the Internal
Revenue Service which is expected any day. This development is all the
more significant when one realizes that the Department of Health,
Education and Welfare never would accept reports in the form of punched
cards. This is a far cry from the day remembered by Al Simpson, the
Manager of Office Procedures Service, when a former Comptroller of the
Company rejected certain proposed punched card procedures, saying,
"We shall not payout our good money over boles in punched
Probably the most dynamic automatic data processing job being done in
the Company is the material control application on the UNIVAC at the
Appliance and Television Receiver Division. The first department to have
used the new system for one year recently cited the following
1. The inventory turnover rate was increased from
12.6 to 18.0 times a year while supporting an increase of 64 per cent in
The investment in inventory was reduced by $2,000,000 not withstanding
the higher production volume.
2. Other less frequent applications in actual
operations are preparation of reports of responsibility accounting,
budgets, expense statements, financial statements, sales and inventory
ports, and others.
One of the first lessons which has been learned is the amount, the cost
and the complexity of paperwork are, if anything, even more astounding
than was originally believed. It is interesting to note that in some
instances these realizations have earned financial people a new respect on
the part of the experienced engineering personnel who have crossed over
into procedures and are working on the design of new electronic payroll
systems. The fact that most of the paperwork performs a legitimate, useful
function unfortunately only wakes the problem of reducing its volume more
We have all been aware of the high cost and growing number of clerical
workers. What was not quite s0 clear at the outset was that a very
substantial portion of this clerical work is performed not in the
financial function - where it was no surprise to dis-
cover that a very large amount of clerical work is done -
but rather in other functions as well. What bas been forcibly brought to
light is that much of the information being handled by one function was
substantially similar to that recorded or processed in others. The need
for integration .f procedures, then, appears in sharp focus..
Another lesson being learned daily is that to use the presently
available equipment at all is a long, bard job. To use it well is much
more difficult but often extremely rewarding. It is incorrect to merely
translate existing procedures, perhaps mechanized procedures, to
electronic procedures in order to produce substantially the same results.
This mere translation from bookkeeping to punched card procedures is not
apt to be particularly rewarding in the light of the economic or operating
advantages which may be gained. In many instances the results will not be
sufficiently rewarding to warrant undertaking the costs and other problems
of making the translation. It is far more likely that to use the equipment
well we must do something more or better than we are now doing. This leads
back again to integration, doesn't it.
As examples, we must find better ways of capturing the raw data in
machine - sensible form at an earlier stage of the clerical process. Or we
must find a means to save clerical costs by consolidating records, perhaps
of inventory, which are maintained, let us say, in the accounting,
purchasing, production control, and warehouse areas. Or we must rearrange
our records to provide in one place in a machine-sensible form a great
deal of information which formerly was not particularly usable because it
was scattered all over a number of different records and therefore not
readily accessible. We must then use this new accessibility to correlate
the data in order to get new or better information. Or, in the sense that
more timely information is also better information, we must use the
electronic equipment to speed up the cycle of time
required to produce information, particularly that information which
involves the handling and manipulation of masses of
figures or the use of complicated analytical techniques.
A further lesson which has been learned is that the real demands are on
the skill, ingenuity and experience of the procedures people who prepare
for the equipment as well as on the equipment itself. In making a major
installation there are usually about eight stages.
1. Study of procedural and economic feasibility, using flow charts, work
distribution charts, volume statistics, and an analysis of the output and
2. Decision to acquire equipment.
3. The recording of the present system in every detail.
4. The design of the Dew system from data input, through the
data processing and final1y the checks, reports, invoices or other
5. Preparation of detailed block diagrams of the sew
systems and the coding or writing of the instructions to the automatic
6. Preparation testing and debugging of the new procedures.
7. Installation of the new equipment.
8. Conversion to the new system.
The demands will vary as to type of skill, and the Dumber or
individuals and the time they are required. For the most part they are
procedures analysts. Quality rather than quantity is to be preferred in
Experience to date has given us adequate confirmation of the fact that
there are important by-products results to be gained from any
investigation which brings together top quality people, often with
different skills to look at an area which formerly was the more or less
exclusive bailiwick of persons with one type of skill. It is common place
to find that traditional delays in supplying information are challenged
(the monthly time cycle for example), standards of accuracy are
questioned, the use of sampling is proposed, and a reconsideration of the
basic kinds of data being furnished to management is urged. Somewhat
facetiously, but making his point well, the Comptroller has said several
times "Order a computer, do all the procedures work required, and
then cancel the order for the computer. By that time you will have
achieved a very large portion of the savings otherwise attributed to the
computer." In such a case the department would react to produce
results, which, while they have nothing to do with electronics, can
nevertheless be directly attributable to the catalytic action which is
produced by the pressure of a proposed computer installation and the
questioning, intelligent review that precedes it.
A great deal has been learned about the economics of using automatic
data processing systems. The installation of the new equipment is normally
accompanied by the highest ratio of fixed to variable charges of any type
of office equipment. The site preparation, conversion costs, rental, the
relatively fixed operating and maintenance crews all tend to produce high costs
and thereby lead to the drive for a high percentage of utilization. This
in turn leads to the selection of (1) high volume areas, such as payroll,
which can provide a good work load and (2) areas where there are good
possibilities for substantial reductions in operating costs. Users of the
equipment must guard against the tendency to use the equipment - once in
place and justified to handle types of clerical work
which cannot produce savings to offset their full share of the fixed and
variable costs. Thus we find a type of marginal costing applied to the new
equipment. The economics of using the equipment seem to favor the use of
general purpose equipment - in the present state of the art. The
developmental and tooling costs for special purpose equipment are so large
that it seems probable that the latter type of equipment can be used only
where the number of potential users for a single application is so great
that they provide a major market for the equipment.
We have learned to a large extent the speed and accuracy of
the equipment are as great as advertised. The people operating the UNIVAC
and the two 702's know of no errors made by the equipment. We have been
impressed with the compactness and the relative accessibility with which
data can be stored, the long sequence of instructions which can be carried
It was not surprising to learn that while, in theory, the equipment can
handle almost any clerical task, there are actually a number of types of
clerical work for which the equipment is not particularly attractive. At
the top of the list of tasks done least well or Dot well at all, would
have to be sorting. The equipment sorts fast but at a very high price. On
the other hand, what has been impressive has been the frequency with which
these high powered equipments can be used on clerical work that involves
little more than simple arithmetic. There are probably two reasons for
1. What we have formerly thought of as being simple Jobs have been
reconstituted into more complicated, integrated areas. Over many years, am
intensive effort was made which succeeded in breaking down what is in
essence a clerical job of some complexity into a series of rather simple
jobs which were within the abilities of the people and equipment assigned
to carrying them out.
2. The second reason is that, to a greater and greater extent, it has
been realized that while these routines involve only simple arithmetic,
they do involve a great deal of data handling - often based on the
comparison of two or more code numbers. This comparison in machine
language is done almost entirely in arithmetic terms. Arithmetic is also
involved in utilizing the decision making abilities of the equipment. In
most cases, as a matter of fact, it can be said that unless advantage is taken
of the equipment's ability to handle data and perform logical operations,
it will be impossible to utilize the other abilities of the equipment to
any considerable extent in clerical work.
We have learned that the ability of the equipment to handle exceptions
is a powerful one but that this ability is not so powerful
as to be able to make all decisions or take care of all exceptions in a
complex situation. The attempt to handle all exceptions will either carry
the coding beyond the capacities of the equipment or require equipment of
such great cost and physical characteristics as are not warranted by the
frequency of reference to the exceptions. Here is a case, then, where the
equipment has failed to live up to the advance billing. The alternatives
seem to be to attempt to reduce the number of exceptions by changing
business policy or to arrange to handle them outside the equipment.
The equipment is enforcing a new high level of data discipline on us.
The input data must be correct. To use the equipment to detect input
errors is highly inefficient. It is essential that procedures be
established to insure a high degree of accuracy before the data reaches
the equipment. This will avoid the situation which exists when the
equipment and the operator have what amounts to large quantities of
"unfinished business." The use of "On line"
input devices - usually without the provision for verification
of input data - increases the opportunity for delays due to
In trying to automate paperwork, a number of paral1e1s to factory
automation can be seen. Just as the factory way find it necessary to
reduce the number of models, standardize, find new materials, and throw
out products not automatable,(??) sp must the
office expect to tolerate, yes, encourage, uniformity of procedures,
standardization of output papers and reports, and so forth. It would be
highly desirable to have simpler, standard calendars, clocks,
measurements, etc. Also, the data processing can't begin until the raw
data are there - all of it and on time. The respect for deadlines will be
more exacting. There can be little capitalization of the programming of
others, the repeated use of the same programs, or much automatic
programming until much wore standardization of procedures is evident..
If you have gotten the impression that many of these comments have been
concerned more directly with procedures analysis and systems planning than
with the equipment, you have been placing emphasis where it was intended.
The big job is procedures, not coding of instructions to the equipment. In
general, present procedures have not been documented very widely in the
Company. We know of only one or two
departments which have flow charts for all of their
paperwork. This means that there is a tremendous job to be done in finding
out what present procedures are before the task of selecting advanced
equipment can proceed very far.
Now getting a little farther from general principles and techniques,
here are some examples of the needs that have become evident for the
present and the next one to five years:
1. Due principally to ease of handling and sorting, the punched
card appears to be here to stay for a longer period than has been
expected. What is needed is the capacity to store mere than the present 80
or 90 characters of information. Perhaps the card should be made of
plastic with capacity, then, to store hundreds and hundreds of characters
magnetically. The facility for interpreting, or printing, a large portion
of the increased amount of data originally stored in the card should be
2. The need for an improved means of preparing original data in a
machine-sensible form was mentioned earlier. The coding systems which
presently use magnetic ink, fluorescent ink, optical reading, the optical
system to be used with the future ERMA's, and magnetic patterns are some
of the approaches developed so far. These devices suggest the need for
uniform sizes of papers - such as the issuance of all internal and
external orders, bills, and checks in the form of punched cards or uniform
arrangements, of machine-sensible coded data on other paper forms.
3. Further improvement in the facilities for estimating data processing
times prior to installation are needed. The knowledge that a given
equipment takes so many milliseconds or microseconds
for the execution of various kinds of basic instructions
has hot proved sufficiently accurate to date. It's the over-all processing
time that is the determining factor, of course.
4. The time required for alteration of programs of instructions must be
reduced. Using payroll as an example, too much time has been required to
incorporate the new insurance plan, new wage agreements, etc. into the
list of instructions.
5. There seems to be a need for a sort of universal black box which
will enable users of punched cards and electronic equipment of various
manufacturers to be hooked together and integrated.
6. There is no question now about the accuracy and reliability of
the equipment. Neither is the need for any appreciably greater speeds as
evident. What is needed is lower rental or purchase price,
lower initial requirements for air conditioning, etc., equipment of equal
capability but occupying much less space, much lower maintenance costs,
greater mobility of components, and freedom from breakdown of critical
components as systems become more intricate.
7. Provision must be made for the earlier closing of books and the
issuance of financial statements following the end of the specified
financial period. Since the limitations of equipment can no longer be
offered as an excuse, it is evident that the solution lies mostly in the
area of procedures analysis and ingenuity. One approach which has been
considered is the prompt, daily recording of all transactions - with the
daily reflection of that transaction in all files and records affected -
perhaps as many as several hundred places. We should then be able to
extract financial statements and other management control reports within a
day or two following the end of the period.
Looking to the period from 5 to 20 years ahead, it is much more
difficult to foresee the requirements.
From our standpoint, the engineering design and capabilities of the
equipment appear to be beyond Our ability to date to train men and learn
to use the equipment. As was pointed out earlier, we are probably moving
as fast as any industrial user, but this does not mean we are moving fast
enough. We must know how to make better use of the tools available so that
the designers can go ahead to exploit the invention farther.
The need for providing random access in an electronic form to a very
substantial volume of information, s-0 as to obtain answers within
thousandths of seconds, has probably received as much attention in the
past few years - without arriving at any supportable conclusions - as any
phase of automatic data processing procedures. The experience of these
years would seem to have altered our views somewhat in this respect. If I
reflect them correctly, present views are that increased amounts of
randomly accessible storage will be practical in the large quantities
originally desired only if they can be produced at relatively low cost. In
most cases it just will not be practical to pay a high price for storage
which is not actively used. On the other hand, when
activity is high, the cost of high or medium speed storage is normally
well worth the price.
There still appears to be certain kinds of work (in banks, for example)
which may Dot be practical without this type of storage. However, we have
learned that it is possible, either through changes in policy, or by
skillful classification of high access demands in
accordance with the frequency of inquiries, to minimize the need for high
or medium storage. As a matter of fact it is interesting to note that in
many instances equipment with fairly substantial amounts of randomly
accessible information are not using this storage so much to answer
interrogations rapidly as to improve equipment efficiency by minimizing
sorting and collating work.
Another field of high interest is the increasing awareness of the
complexity of communication problems and the frequent interdependence of
solutions to these problems and the success or failure of automatic data
processing. Under some concepts, it is difficult to
draw the lines between the two fields. It is clear that communications
requirements are -no longer thought of as involving
only miles or thousands of miles, but also as involving feet - the
distance from one area of the plant to another or to the data processing
center. It is clear also that the facilities available for communications,
and our ideas and concepts of what is essential data, i.e. what needs to
be transmitted, will both have to advance substantially before we can feel
a satisfactory solution has been reached in this area.
It seems clear also that the ultimate automatic data processing is far
from here, but that the progress being made is far from abating. As a
matter of fact, progress can't slow down, because you gentlemen who have
decided to launch the Industrial Computer Section and a new set of
vigorous competitors 'Won't allow it and the demands which motivate the
government won't let it.
I think you will agree that if you move a clerical function as much as
ten feet it takes a number of days to adjust the new situation so that
operations go on smoothly. Similarly, the final successful design,
development and debugging of even early information handling and data
storage systems will make substantial improvements in carrying on business
and will require the efforts of experts in the running of that particular
kind of business, working for years in close cooperation with the
engineers who understand what data processing is all about and what modern
electronic equipment can accomplish. To bring together these experts in a
hard working project is not easy.
We must continue to develop at a fast pace, a greater number of
new kinds of people to do the job. When the revolution has finally taken
place, whether it be in five to ten years - which I think is too
optimistic - or the twenty or thirty years which is closer to the truth -
will be possible for us to realize that the colleges and universities are
graduating new doctors who are systems and data processing experts,
products of an entirely new
course which probably combines unusual study of the human
brain, the physical sciences, mathematics, business, industry and
economics and the requirements of government laws, rules and regulations.
Ultimately the problems of business data handling will be solved - memory,
sorting, simple computation in general, lots of adding and multiplying and
putting together of data on the various forms going in and out of machines
and data input devices.
Standard black boxes will probably be cabled together; the situation
will be very similar to that of the telephone networks in which it bas
gradually become possible to design standard equipment which is reliable
despite the millions of input and output points. In fact, only by
understanding these problems well enough will we be able to arrive at
standard devices at the hope to make any real headway. Systems are bound
to fail in the end that is, to be uneconomical in application, unreliable
and littered with false starts - if in big data handling applications it
is necessary to evolve too many compatibility custom built arrangements.
When we have this twenty years of development behind us, then I believe
that it will be rare to walk into any factory in a
company of substantial size and not observe boxes throughout the factory
containing electronic gear which is simple and reliable. The management of
such organizations will have almost immediate visibility as to the status
So there will probably be a substitute for the present paperwork dilemma.
Large scale devices, possibly highly interconnected with each other, will
provide running control and data for the management staff. I believe that
it is very difficult to try and compare the way the things will be when
this business data processing industry has matured with the way they are
It is going to be necessary and inevitable that this whole field be
attacked on all fronts mere or less simultaneously.
The really big strides are going to be made only after there has been a
solid effort by the large number of people including technical and
business experts. We are talking here about the application of new
techniques to business and industry that will pay off and are paying off
wherever applied in cutting costs and simplifying procedures and providing
better service to customers, providing better visibility to management,
being compelled to make uniformed decisions and better planning in
business and industry. Because of tremendous complications, many years of
unusually high caliber effort will be required to realize the big payoff
of major replacement of drudgery human effort by electronic business data
processing systems. Someday it will be true that business and industry
will be revolutionized. I do not think we can expect it in the next few
years but we can expect to see gradual effects on our businesses from now
TO INDEX OF SPEECHES