'Trevor began a Science course at the University of Western Australia that was interrupted when he joined the RAAF in 1942. He was selected for a special course on Radar at Sydney University, which led to his serving as the CO of two radar stations, the first in Geraldton and the second in New Guinea.
After the war Trevor completed his BSc with 1st class Honours in Physics, and after four years of research joined Weapons Research Establishment (WRE) and was sent to England to assist in the design and development of a classified computer, one of which he installed in Melbourne in 1957. After this he led a team completing the design of a range-safety computer for the Woomera Rocket Range.
In 1961 he entered the commercial world, joining E.L. Heymanson, a Melbourne-based distributor, to manage their computer marketing activities. He convinced Heymanson to seek representation for Control Data Corporation, a blossoming US computer company based in Minneapolis, Minnesota. Trevor then spent two months in Minneapolis in 1962 becoming familiar with CDC Products and culture..
"All who knew this remarkable man benefited, grew and expanded their horizons".
The above preamble is from the Complete Eulogy for Trevor by John O'Neil
The following document is a compilation of speech notes given to Ron Bird by Trevor in late 2006. They are Trevor's 'spoken' reflections on the development of the digital computer industry in Australia commencing circa 1946. Ron has added supplementary notes taken from several phone conversations on this subject with Trevor in 2006.
CSIRO - CSIR MK 1
In 1946, CSIRO embarked on a course, which resulted in the CSIR MK 1 for the Radio Physics Division. This was one of the earliest, truly automatic, stored program computers. CSIRO had foreseen the need for faster methods of computation and analysis of its research data.
The CSIR was completely 'home-grown' and at some 10,000 miles distant from the mainstream development in the UK and USA. Bearing in mind the 23 patents, which were issued by the Australian Patents Office between 1909 and 1940, such progress augured well for Australia's future in what turned out to be a very significant Australian Industry.
At this time, even Thomas J Watson, the father of IBM, could not accurately interpret the signs - he is said to have remarked that the world would never need more than six electronic computers. So, it is small wonder that CSIRO, in 1953 pulled back from developing a successor to the CSIR MK 1.
Australia's ability, in the commercial field, "to snatch defeat from the jaws of victory" is notorious; CSIRO's action was an early example although it must be said that the reasons for the decision were complex and understandable at the time.
Fortunately there were other factors at work. Thanks to the early work at Sydney University by people like David Myers, and the well-known efforts of Trevor Pearcey and Maston Beard on the CSIR MK1, Australia got off to an early start in computers. Of course we squandered it to a degree.
WEAPONS RESEARCH ESTABLISHMENT - ELLIOT 403
In the early 1950s the Weapons Research Establishment sent a few people including John Allen-Ovenstone, Don Overheu and myself to the UK to work under Wilkes at Cambridge and Tommy Flowers at the British Post Office. Others like John Bennett at Cambridge and Manchester and Barry de Ferranti at GEC had already found their own way to the UK, but, apart from Pearcey, there wasn't much local hardware activity until the Weapons Research Establishment installed the Elliott 403. This was a one-off that turned out to be a bit of an orphan but it did the job of reducing the large volumes of data from the Woomera missile range.
SYDNEY UNIVERSITY - SILLIAC
Next, in 1954, came the inimitable Harry Messell at Sydney University. Utilising the UK computer experience of John Bennett and Barry de Ferranti and with STC's manufacturing help, they built SILLIAC, a derivative of the University of Illinois ILLIAC machine which was built and installed in a commendably short time. This was a successful pioneering effort which, like the CSIR MK1, did much to provide early programmer and user experience. The Defence Signals Branch (at that time top-secret) sent two people to the UK to help design and build INFUSE - a large but unconventional pipeline processor.
Other organisations, as exemplified by the NSW University of Technology and the Commonwealth Bureau of Census and Statistics, were, for their own diverse reason becoming involved in electronic computing. The Bureau of Census sent one of its bright young statisticians to the US Bureau of Census whose pioneering efforts to utilise computers (UNIVAC, in this case) on the reduction of large bodies of data had considerable influence later on Australia's data processing policies.
EMERGENCE OF THE AUSTRALIAN INDUSTRY
Computing in the 1950s was adequately delivered by three superchargers of technology: curiosity, enthusiasm and the need for ever improving performance. But beyond the standard motherhood attitudes to all curiosity driven efforts in technology there was little government reaction to computing. Perhaps this was to be expected at this time.
Computing was perceived simply as a scientific and engineering activity; its significance as an industry was yet to emerge. There was certainly no hint of any relevance to productivity, balance of payment issues, the standard of living or contributions to the "clever country". Government policy as a practical issue, at that time, was non-existent!
The office machine companies that were already established were followed in the 60s and 70s by a deluge of other US and UK manufacturers operating through subsidiaries and distributors.
BURROUGHS IBM NCR & ICT
AND THE BIRTH OF THE SERVICE BUREAU
Beginning in the mid to late 1950s the Australian subsidiaries of the office machine companies, Burroughs, IBM, NCR and ICT began their efforts. Well-established as tabulator, cash register or calculator suppliers they were in a good position to take their customers into the promised land of automatic data processing.
Some local distributor companies were also active in the area - for instance Ron Jelbart initially distributing Flexowriters, and having shrewdly trademarked "Data Control", "Data Print" and "Data Centre", got into the early service bureau business along with IBM, NCR, GE and ICT.
Service bureaus served a dual purpose - they enabled customers to develop their programs while waiting for their own machines and they served as effective demonstrators. But they also performed useful work. Roy Hadley at NCR in Sydney set up an NCR 405 bureau in 1959.
These systems were relatively primitive - 1K of nickel delay line memory, magnetic film storage (not tape), a paper tape reader and an off line film to printer device. To maintain some reliability 3% of the hardware was recycled (after testing) through the parts store each day, and all the programs were written in machine language.
Tony Benson has vivid recollections of the difficulties, but real problems were solved; For instance, 30 years' of data were analysed for the Ionospheric Prediction Service and significant engineering design work was done for the Snowy Mountains Authority.
Similar stories can be told of ICT with a 1201 in their Melbourne Bureau and IBM, who installed a 650 in the MLC/Sydney Bureau in 1958.
Barry de Ferranti was also significant in this capacity as IBM's assistant branch manager in Sydney; Barry, I might add was a real pioneer having begun his computer career in 1948. In the early 1960s he, with Cliff Bellamy, ran Ferranti's Australian subsidiary, and he was still in business as a consultant in 2006 - that's a career spanning almost 60 years so far!
INTERNATIONAL MARKET RECOGNITION
Although Australia's distance presented logistics problems in training and support the international suppliers began to recognise that there was money to be made here.
The University of NSW installed UTECOM (an English-Electric DEUCE) in 1956, the ANZ Bank in Melbourne bought a GE-225, the Weapons Research Establishment acquired an IBM 7090 and perhaps most significantly, some of the major Commonwealth government departments began sounding out suppliers for some really hefty purchases. These included the Bureau of Census and Statistics and CSIRO with others, like the Department of Health in the queue.
In 1961 Honeywell led the charge. The Defence Department, under John Ovenstone's guidance, acquired a pair of Honeywell M-H 800s. The associated tendering process had attracted all the US and UK majors; ICT, English Electric, Ferranti, IBM, GE, Bendix and RCA (through AWA), Univac, NCR, CDC and possibly others.
The following year CSIRO and the Bureau of the Census RFPs confirmed Australia as a market worth taking an interest in. Although CDC/CDA won both those contracts there was a lot more business in the offing particularly from the private sector, which then constituted about 40% of the potential business. By 1963 all the Defence Department sales hopefuls had subsidiaries or distributors and these were followed by others such as DEC, Amdahl, Facom (later renamed Fujitsu), Data General, Prime and many others - a total of more than 30, in fact.
This period saw the emergence of additional data centres: ADAPS merged with Kim Jelbart's original Data Centre, Univac put in an 1108, CDA a CDC 6600 and the universities began installing larger machines - IBM 360s, CDC, Burroughs and ICL for the most part.
LOCAL HARDWARE AND SOFTWARE DEVELOPMENT
Associated with all this activity was the emergence of Australian hardware manufacturers and software houses. There were about twelve hardware companies of that era and sad to say - not one of them has survived - with the possible exception of Eracom but I doubt if that company was one of the originals.
Computer Power was the first of the software shops and although it will always (and properly so!) be associated with Roger Allen, it was in fact started by an unusual American called Jack Vale.
Somewhat more sedately, Lyndsey Cattermole, with an IBM background, set up the successful and respected Aspect Computing some time later in 1974. There were others including Ferntree but none were more successful than Computer Power and Aspect and possibly the Whitehorse Strategic Group
EARLY AUSTRALIAN DISTRIBUTORS AND SUBSIDIARIES
The CDC subsidiary, like Honeywell, began through a distributor. In distributor talk, the 'principal' was the primary supplier. From the principals' perspective, distributors, especially the naive ones who started off believing that principals had principles, were a good idea; they didn't cost anything until commissions had to be paid, they did all the heavy work of finding the first customers, and, when successful, they could be dispensed with as soon as their 90-day or 180-day or whatever "Notice of termination of Contract" could be applied.
CDC happened to be a very respectable company so their distributor (E.L.Heymanson and Company was treated properly as they should have been but this certainly was not the universal experience.
On the other hand, distributors usually had a distorted idea of what the systems were capable of. They promised too much, they usually did not have the financial resources to recruit and train support staff, maintain expensive stocks of spares, or, worst of all, stand the strain of long - twelve months or more - sales cycles before installation and then the agony of acceptance testing and the wait to be paid. For these reasons, even later on, successful distributors like Lionel Singer with Prime, Pyramid and Sun found themselves replaced by subsidiaries.
The subsidiaries seemed to have widely varying modus operandi; some, particularly the UK subsidiaries were tightly controlled. Others received almost unlimited freedom. CDC had never expected to sell anything bigger in Australia than a 160-A, (equivalent to an IBM 1620 or a Ferranti Sirius) and, never having had a subsidiary anywhere else in the world did not quite know what to do, so they listened to our advice. They sent money to pay newly recruited staff, they answered technical questions properly and they trained Australians in the US to return to Australia to train the others who were needed.
Data General, I am told also gave their local management total carte blanche.
My impression is that this was pretty much the norm, although as might be expected, it was usual to send a US or UK national to run the subsidiary, at least in its formative years.
I remember these CEOs as we would now call them as a decent lot. There was not much fraternising but contacts were 'friendly' and the Americans, in particular, were a different breed from many of those who came later.
EARLY MEDIA INFLUENCES
In 1965 CDC invited Bob Forrest, the editor of Datamation to Australia and he ran three low-key cocktail parties in Sydney, Canberra and Melbourne so that he could meet and talk to both users and suppliers. All involved were invited and they all came with the exception of one company: sipping with the enemy was, it seems, like sleeping with the enemy and they didn't do that sort of thing. Datamation later put out a special edition devoted to computer development in Australia. There is an interesting sidelight to this, which should be mentioned.
In the early 60s both The Australian and The Financial Review ran a weekly feature on events in the computer field. Then, as now they were mostly concerned with who was buying what, from whom and who was getting rich (or thought to be getting rich) as a result. Industry visitors from overseas suggested it was all a bit prurient, and pronounced it unique to Australia.
The practice has persisted to this day with a much wider range of publications. Among the later practitioners were Computer World and Computer Pacific Weekly. Susan Coleman, CPW's 'fearless dragon' lady, was a pillar of support for Australian companies in their mostly unsuccessful attempts to survive the stranglehold of the multinationals, and an unrelenting critic of governments which, she correctly averred, never did enough to help.
Susan kept up the pressure from CPW's inception in 1972.
I cannot resist quoting from a 1991 letter to CPW from John Marquet which concludes "Nevertheless, a government that won't distinguish between Australian-owned Australian companies and foreign-owned Australian companies when handing out grants, bounties and other assistance, is being perverse, and upsetting Ms Coleman and me, and in Susan's case they had better watch out."
By definition the computer manufacturer subsidiaries here were multinationals; they came for no other reason than to make money. Nobody really thought that the boards of IBM, Univac, Honeywell, ICT, CDC or GE were lying awake at night worrying about Australia's economy. However the multinationals had a lot to teach us and they did. Although it seemed more important at the time, the transfer of technical knowledge may not have mattered as much as what they had to impart in management, personnel and engineering methodologies, although that was more applicable to the US than the UK suppliers.
Critics of the multinationals saw their behaviour in a different and more sinister light. The multinationals, they claimed, took every opportunity to gouge us - they didn't give the same discounts as in their home markets, they charged too much for maintenance, they extracted management fees from their subsidiaries, they wouldn't ship by Qantas but insisted on Pan-Am, they even put their insurance through US companies.
A good deal of this was true, but it need not have been so.
LACK OF GOVERNMENT SUPPORT
The public sector today represents about 40% of the computer market - back in the 60s it was more like 60% and that was plenty of leverage for the State and Federal governments had they put their heads together to stop to some of those understandable but less than friendly practices.
But at that time, the Federal Government had only two instrumentalities that took an interest in what was going on in the computer field. One was the Stores Supply and Tender Board which negotiated the contract after the dreaded Inter-Departmental committee - the IDC - agreed that the Department's choice of supplier could go ahead. Beyond wanting to see value-for-money, which was the responsibility of the Stores Supply and Tender Board, there were no visible Government policies in the early years.
MORE RECENT OBSERVATIONS
Some 40 years later I venture to comment that the intervention of Government, beginning in the mid-70s has achieved very little and one should not expect that to change much in the future. Having watched the Federal Government's IT machinery at close quarters for 4 years I believe there is no single sinister culprit to blame for the fact that Australia does not have an IT industry adequate for exploitation by its many extremely talented IT individuals. A proper industry would not only avoid the annual $l0 billion or so trade deficit that overseas IT purchases represent, but could be a wealth producer into the bargain. (Ed Note: In 2007 the annual Australian IT purchases deficit is 20 billion!)
The problem starts with politicians who, almost without exception, don't understand, and this is exacerbated by the bureaucracy, where public servants soon to move to different jobs think that they have acquired sufficient knowledge of a very sophisticated global industry, to recommend policy settings to government.
It is small wonder that nothing much ever happens. Susan Coleman and many like her have been quite right to complain. Only the government can fix the problem and continuing inaction at this level has gone on for too long;
Ministers should stop listening to Bill Gates and start taking notice of our own IT people. And they should stop calling for more reports and read one or two of the 20 or so already in existence; they all say much the same thing.
To be fair I must mention one really good move although it is not solely applicable to the IT industry; I speak of the halving of capital gains tax.
But getting back to the early subsidiaries, it seems remarkable that there were no disasters of any great magnitude, although I suspect there were many bad moments. The DSB computer, for instance, was fabricated in the UK, then, to use later jargon, drop-shipped. When tested for the first time, it didn't work! It took 2 months to solve what turned out to be a power supply problem but in that time management nearly had heart failure.
There were real problems with the disc storage on the Elliott 403 at Weapons Research Establishment. The story was that the Elliott people gave up to go home for Christmas and left it to the locals to fix. There were many others. When the Bureau of the Census was acceptance testing its CDC 3600 (then claimed to be the worlds most powerful), it wouldn't run its sort program - although the 3600 at CSIRO could.
Census certainly wasn't going to pay for the machine until what was obviously a hardware problem was fixed and not a lot of imagination was required to work out that if the situation became known to the outside world, both the Bureau and CDC would be in real trouble. Fortunately, CDC's then Australian software manager, one George Karoly was called in to help the hardware people. After determining the point at which the program halted and establishing what ought to have happened in the next few clock periods, George told the customer engineers, "I think you've got a wire missing". And that, indeed turned out to be the case!
Now there was nothing very unusual about this in the days of many small printed circuit cards and mammoth backplanes but there is a point to the story. Computers are complicated devices and there are millions of reasons why they won't work.
As one of my early mentors told me, there is only one reason they will work and that's when you have taken care of all the reasons they won't work!
Fixing hardware problems, particularly the intermittent ones, and especially if there is more than one, or if there are timing glitches, is a God-given art that some engineers are particularly good at. The same sort of comment applies to software problems.
My experience, and I am supported by others, suggests that Australians are more than unusually talented in this area. If true it is hard to find an explanation. Specialisation has lots of advantages and particularly so in dealing with computer problems, but specialisation calls for skills in partitioning the problem to be followed by good team work.
Some really difficult problems are not susceptible to the team approach and require a single individual for their solution. The Americans and the Europeans, with their large populations have, very sensibly, exploited specialisation whereas we, a sparse population with and agricultural tradition of having to turn a hand to a wide variety of tasks, have perhaps developed a little more self-dependence, initiative and ingenuity.
If this were not primarily an Australian audience I probably wouldn't get away with this, but, based on observations in this country, the US, Europe and parts of Asia, I believe Australian IT people, both hardware and software, generally speaking have unusual talents and this is one of the reasons why problems were solved before they became disasters in these early days of this industry.
Another issue worth mentioning is the question of where the subsidiary was to be based; Sydney or Melbourne? Ferranti and CDC chose Melbourne - in CDC's case because they had had a Melbourne distributor and there was no point in moving to Sydney - the Stores Supply and Tender Board was then in Melbourne and all its prospects were either in Melbourne or Canberra.
The others were either in Sydney or, like Honeywell, Univac and DEC chose Sydney. And who wouldn't? To this day Sydney is the centre of the computer business. Hewlett Packard, the last major company with its head office in Melbourne has only in the last couple of years moved to Ryde NSW. A very senior executive in CDC told me that some 'know-all' in the US had told CDC that setting up in Melbourne was a serious marketing handicap. Wayne Fitzsimmons has told me that Data General had come in for the same criticism. Melbourne certainly didn't do either CDC or Data General any harm; or Hewlett-Packard, for that matter.
It might be thought that there are better things to worry about than Sydney/Melbourne tiffs but sibling rivalry is an odd phenomenon, which will probably be around for a while yet.
In conclusion perhaps we should ask what lessons, if any, were to be learned from these events, which take us back 30, 40 and 50 plus years. I think there is really only one lesson worth noting. This was succinctly stated in a note pinned to the plastic cover of the first CDC 3200 delivered to Canberra in 1964:
"We haven't got this to work yet. You're on your own bud. Good luck" *
And that's right. The Australian IT industry is on its own - and good luck - and I think we'll probably make it in the end or perhaps even earlier.
Compiled and edited from several ETR notes and conversations
R G Bird 22cnd July 2007 Rev E
Similar to the first delivery 3200 note above "You're on your own bud. Good luck"
In 1962 ICT were running late delivering their 1301 to Cadbury in Tasmania and were being threatened with contract cancellation and giving the order to IBM for a 1401.
To avoid this, ICT shipped the machine that was, in effect, incomplete with parts still being shipped from the UK to Claremont for about 8 weeks after the machine first arrived in order to allow us to finish building the system on site.
To add to the problem of late hardware, the software did not work reliably (MAC (Manchester Auto Code) which was an assembler language and "Rapidwrite" which was a COBOL look alike).
To avoid a looming conflict with Cadbury, four of us (3 engineers and 1 analyst) burnt the midnight oil and wrote the main invoicing application in absolute machine code as well as trying to get the hardware running. As it turned out we managed to get this all working within about 10 weeks of the machines arrival and narrowly avoided the cancellation of the contract.
This invoicing program ran 3 invoices across a page and ran 2 shifts a day. Cadburys at that time invoiced every retail outlet, including every corner milk bar in Australia directly as they did not use distributors for their products at that time.
RGB 22 July 2007
RETURN TO 'INDUSTRY STORY'
THE DEVELOPMENT OF THE AUSTRALIAN COMPUTER INDUSTRY
By Trevor Robinson
This is a Hard Disk! In 1956 IBM launched the 305 RAMAC. The first computer with a hard disk drive It weighed over a ton and stored 5MB of data.