|
Looking Back at the Computer Symposium
by
W. L. MURDOCK
Consultant.
Operations Research and Synthesis
Consulting
Service Management Consultation Services
The sense of the symposium seemed to be that the standard component or
building block idea of engineering computing equipment is to be adopted by
General Electric. I should like to comment on some of the results of this
decision as I see them.
ENGINEERING
Ken Geiser expressed the opinion at the symposium that we would be able
to design the building blocks with their properties and impedances so
completely specified and standardized that application engineers would be
able to assemble them in the field into a particular system for a
particular customer. This seems critically important to me. Otherwise,
Engineering would be heavily involved in most customer orders and costs
would be high. On this point, I believe, hangs the clear competitive
advantage of this equipment philosophy over the general purpose machine
idea. Some special packaging, such as is done in the Specialty Control
Department on welding controls, will be necessary but Engineering will not
have to organize a special effort for the great majority of the orders.
It will be of interest to see if the engineers will be able to make the
basic building blocks correspond to the elemental functions of
communications: counters, multiply circuits, divide circuits, buffers,
additive storage units, standard relay circuits, negative and zero balance
devices, etc. The interest derives from the fact that each customer will
have his own special combination of needs but these needs can all be
served by such elementary operations. Not only will building more complex
special logical units tend to increase the special engineering but it
would also imply that the customer will have a greater problem if ever he
wants to change or add to his system. I believe that we can assume that
the customer will want to add to and change his system continuously. Ford
Dickie expressed the opinion that the majority of customers will not ask
for high speeds in their equipment. I believe that the customer will tend
to prefer equipment which is just adequately fast, very reliable, easy to
repair, easy to add to and modify and understandable in terms of his
business operating functions. Optimization of speed and cost with regard
to any particular function, such as matrix multiplication for production
control and costing, will not impress him if it
implies a rigid system which has to be reconsidered by the equipment
factory whenever he wants to make a change.
PRODUCT PLANNING
It would seem that product planning would be simpler if we are able to
design compatible building blocks in terms of basic functions.
31.
|
| If we must design whole complex logical
units then Product Planning has the chore of analyzing the systems of
customers in the markets chosen and coming up with widely marketable
units. Of course, even customers with much experience in automatic data
processing cannot tell us what they want in the way of real time
equipment. The burden is on our own Product Planning. But the problem is
much simpler in terms of the dozen or so basic arithmetical and logical
operations. If we can design in terms of the basic function then the
problem is one of providing the application engineers with components so
designed that they can devote their attention to the customer's operating
system with a minimum of attention to compatibility and requirements
between components. The application engineer needs to be able to write
schematic diagrams which faithfully portray the customer's needs but with
nearly all the compatibility and impedance problems which the schematic
implies presolved for him in the integrated design
of the line of basic components. Perhaps a symbolic notation, a sort of
operational calculus, will be developed which will readily allow him to
string equipment together, paying chief attention to the customer's
system.
APPLICATION ENGINEERING
A major question here is to know in what terms the application engineer
and the customer will communicate. The ordinary procedures chart which
identifies piece of paper and machine is not sufficient. All information
must be completely described as to the kind and number of characters,
timing and frequency. All arithmetical and logical operations must be
completely and precisely described. Operating business people are not used
to stating their procedures in these terms. The application engineer must
be able to extract this information from the customer without an undue
amount of time on the customer's premises. For one thing, this seems to
mean, that he must understand the logical and arithmetical aspects of,
say, production control or order service, as business functions, to an
unusual degree. This suggests the wisdom of our marketing effort picking
out certain markets, such as manufacturing control, which
it will thoroughly understand and leave others, such as retail store
communications, to another time or Department.
SALES
Herb Grosch brought out the disadvantage of salesmen who can speak only
in general terms. These men should understand the logic of the customer's
need and the systems designed by the application engineer. Clearly defined
markets, thoroughly exploited, will permit a sales effort of refreshing
depth of content to the older customer of data processing equipment.
Since the basic building blocks are s0 universally required and
since the characteristic of the product is in the system designed for, by
or with the customer, it may be that a single component manufacturing
organization could serve a number of systems design and sales organization
each of which are pointed at a different area of application.
32.
|
| I believe the Engineering work which lies
ahead is more clearly apparent than is the marketing work. Planning for
the latter seems more difficult and more obscure. A good way to gain
evidence would be to carefully follow at regular intervals for several
years the usage by customers of delivered equipment.
We in Services are seeking more ways for the "customers", in
this case General Electric Departments, to use presently available
equipment to improve communications. One promising approach is to
teach standard punched card equipment, not machine by machine, but rather
to teach the basic components: counters, selectors, comparing units, card
feeds, etc. of which any particular punched card machine is just a
particular combination. Information flow charts with all logical and
arithmetic operations specified, integrated card designs, card programming
techniques, whole business planning of machine procedures, are aspects of
the approach. The natural advantages emerge in the ability to relate the
counters and selectors (relays) in the machine directly with business
operating functions requiring arithmetic or logical choice whereas it is
not clear what a particular machine, say the IBM 407, has to do with a
business activity, say Order Service. But speak in terms of selectors in
the machines and in terms of a specific business function, like charging 2
per cent tax in Idaho, on the other and it is possible to establish a
direct connection. It will be recommended that managers learn what
selectors, counters, and several other components are, that procedures
people have a complete functional understanding of all components and
timing of the machines and that technicians and board wirers be upgraded
to a more professional level. Integrated punched card procedures are
uncommon and component usage (not machine usage) is typically quite low
due to a great abundance of too simple machine runs.
33·
|
| (Editors Note: This
title does not really match the table of contents... exactly, why?
we do not know, perhaps the table was done and the paper submitted
at the last moment....ed-) |
| BACK
TO INDEX OF SPEECHES |
| |
| |
| |
|
