In 1981 I was in charge of computing in BP Chemicals and was
invited to give the keynote speech to an audience of computer
manager at Univac's International Executive Centre near Nice. The
title was to be a rather ponderous "Data Processing through the
1980s in a multiple site process manufacturing company".
In that long-forgotten speech I made a number of predictions,
and having only recently unearthed the script I have been tempted
to try and see to what extent, if at all, those predictions have
come to pass in the intervening 28 years. First, however, I must
explain, very briefly, how things were at that time, since those
predictions had to be made from the status quo ante.
The portfolio of applications run on mainframe computers in 1981
had mostly been developed during the previous 10-15 years, and many
were still batch-based. They had little interconnection with other
applications and, for some, their data entry was still by means of
punched cards encoded from specially designed input documents.
Essentially, they had their origins in the basic design
philosophy which gave rise to the old punched card systems; had
been developed piecemeal; and had links between them added on only
when the need was identified. They reflected the fragmented growth
of Data Processing in those early years whereas, in contrast, our
process manufacturing companies operate processes which by their
very nature are integrated.
The challenge we faced was to integrate (ie interconnect) our
computer systems so they could provide quicker and better
information for those directing our company's affairs. And because
their reactions impact and repercuss elsewhere in the company's
processes, this input and the changes consequential upon it had to
be handled by our computers without delay.
I foresaw the implication of those demands as requiring the
widespread use of database technology (still in its infancy in
1981) with access to it freely available via VDUs having effective
software for data retrieval, manipulation, and report generation
purposes. All this with sufficient computing capacity that access
could be gained at any time without having to wait.
In my speech I proposed the following to meet the above
requirements:-
(a) An 'outer circle' of mini computers being used at
outstations for dedicated functions, and using software packages
developed specially for those particular functions.
(b) Computers located in the major (regional) sites being used
specifically for site purposes and using the standard package
approach to applications.
(c) A corporate mainframe computer holding corporate data at
fairly elemental level in a structured database, supported by good
data retrieval/report generator software.
All three layers to be connected by a communications network
linking them together. The blueprint I postulated shows the
coalescence of computing and data communications, functions which
in 1981 were usually organisationally separate - as was office
automation. Sooner or later, I argued, their functions would
inevitably be brought into one grouping, which would probably be
called information processing rather than data processing.
At my present age of 80, I have been away from the computing
world for many years and do not pretend to be au fait with current
technology or trends. I notice, though, that we now hear and read
about IT, which embraces the different functions I mention above.
And where I predicted the increased use of VDUs (mini computers),
every office seems to have a PC on every desk. The considerable
advance in Data Processing since 1981 is, to me, most noticeable
when I visit the bank and my transaction is immediately reflected
on the print-out of my account entries I sometimes request. I
marvel at the speed of communication, and the efficiency and
versatility of the software embedded in a database in some distant
centre.
Modern computing has progressed considerably over the past 28
years, but I like to think that I may have been more prescient than
I ever realised when I gave that keynote speech in 1981.