Subscribe
Issue No.04 - October-December (2004 vol.26)
pp: 46-61
Ernest E. Keet , Vanguard Atlantic Ltd.
ABSTRACT
The author, a participant in packaged software's early days, worked with IBM up through the System/360 launch and then with Turnkey Systems, an early provider of packaged software. This article traces the author's background from graduate school through the Turnkey Systems sale to National CSS in 1979 and subsequently to Dun & Bradstreet.
INDEX TERMS
Turnkey Systems, software, telecommunications monitor, Task/Master, Key/Master, Docu/Master, National CSS, D&B
CITATION
Ernest E. Keet, "A Personal Recollection of Software's Early Days (1960-79), Part 1", IEEE Annals of the History of Computing, vol.26, no. 4, pp. 46-61, October-December 2004, doi:10.1109/MAHC.2004.20
REFERENCES
 1. Luanne Johnson is president of the Charles Babbage Foundation. 2. Mercator, the product sold today, is a modern object-oriented interpreter of real-time data streams that lets its users link legacy and modern systems with a seamless interface. This product was developed by a company that could trace its lineage directly back to the early Turnkey Systems. 3. We had to account for this as compensation, taxable to the employee, but we "grossed up" the payments so the family travel was at zero cost to the employee. 4. A year and a half later, Jerry's impetuousness got him fired, and Burndy hired Turnkey Systems to provide an interim VP of data processing: me. I did this job for almost six months, with good second-level people who saved my bacon and who were relieved to be away from Jerry's bombast. I hired my replacement— which was my main job at Burndy— who immediately cancelled all contracts with Turnkey Systems and proceeded to make radical changes to the systems architecture that Jerry had built. I'd like to think that by hiring a strong, independent guy we did Burndy a service at our own expense, but the "tear it down" approach seldom fixes problems, and despite later successes I don't think that his approach improved Burndy's systems. As a postscript, Jerry Kaufman left data processing altogether and became a psychologist. 5. Sending the data and its formatting characteristics as separate data elements is now accepted practice. HTML, XML, and other markup languages are based on this concept, but in 1969 it was reasonably innovative. 6. Also see Larry Welke's anecdote in the IEEE Annals of the History of Computing, vol. 24, no. 1, Jan.–Mar. 2002, pp. 85-89. 7. Before the System/360, IBM's products ran only batch jobs. Because the processor ran so much faster than the peripheral devices, most of the time the processor was in a wait state. Multitasking, however, enabled the devices' relatively slow processes to overlap. Consequently, with sufficient memory, enough programs could be run to ensure that the wait state was minimized, which meant that this expensive mainframe computer was optimized. 8. The original team consisted of Steve Ward, Dennis Sisco, Joe Farrelly, and me. In 1971, Bill Zack joined the team briefly, but in the main the design and implementation of Task/Master was the work of the original four. As my programming skills became known by the others, I was relegated to developing the user interface modules. In my "break down the door" manner I got the job done, but the programming was ugly. 9. It remains unique. Its source code is on file at the Charles Babbage Institute, and the source code is as self-documenting and readable today as it was 35 years ago. 10. If we consider one-off projects and system software developed under contract to the hardware manufacturers, the industry dates from the mid-1950s. Computer Usage Corporation (CUC), founded in 1955 by Elmer Kubie and John W. Sheldon, wrote a program for California Research Corporation to simulate oil flow. Fletcher Jones and Roy Nutt founded Computer Sciences Corporation in 1956 to develop systems software such as compilers. Simscript, a simulation language developed by California Analysis Centers Inc. (CACI) was launched in 1962, and the ADPAC compiler appeared for direct sale in 1964. Marty Goetz claims that the true software products industry started with Applied Data Research's Autoflow launch in 1965. Historians will have to sort this out, but it appears that there were no companies mass-marketing software packages directly to end users until the early 1960s, and no real "industry" until a decade later. 11. As noted, Task/Master's source was Cobol, and we distributed this along with Steve Ward's Genesys code-modifying software utility, another unique product Turnkey Systems developed. Genesys produced custom Cobol, unique to the customer's parameters, from a master source listing. This output was then compiled, giving each customer a unique system but from a common master source listing. If they wished, customers could modify their unique Genesys output. So long as these modifications were made in separate linkage modules, they could maintain custom systems across many new releases. 12. Marty Goetz, a former Sperry-Univac programmer, was a founder (in 1959) and later president of Applied Data Research, or ADR, one of the first software products companies. ADR's first product (Autoflow, a computerized flowcharting system) was launched in 1965. Marty later sued IBM for attempting to monopolize the software industry and in 1970 won a $2 million out-of-court settlement. The settlement helped ADR grow and become listed on the New York Stock Exchange. Years later his daughter, now grown and a lawyer, sued IBM on behalf of a medical company Marty had helped found, and won again. I think Marty may be the only person to have successfully sued IBM twice and won both times, using the proceeds to build successful companies. See also M. Goetz, "Memoirs of a Software Pioneer: Part 1," IEEE Annals of the History of Computing, vol. 24, no. 1, Jan.–Mar. 2002, pp. 43-56, and M. Goetz, "Memoirs of a Software Pioneer: Part 2," IEEE Annals of the History of Computing, vol. 24, no. 4, Oct.–Dec. 2002, pp. 14-31. 13. In 1992, I was riding a train from Westport, Connecticut, to New York City. I was reading my newspaper and heard the two men sitting opposite me say "Task/Master" so I put down my paper, interrupted them, and introduced myself. It turned out that they were from American Brands, one of our earliest users. They were lamenting the fact that their management had decided to abandon Task/Master. This was 14 years after we had discontinued any new development or sold a new customer site. If there is a lesson in this, it is that software has a much longer useful life than almost anybody, including us professionals in the field, will ever believe. 14. We never sold the software, we wrote a perpetual use license. This eliminated the monthly lease or license fee but substituted an optional annual "maintenance" fee, in reality the continuing right to enhancements and bug fixes. 15. The original Graphics product was priced at$15,000, the first Task/Master at $18,000, but over time this price increased steadily. It was close to$40,000 when we discontinued it in 1978. 16. The day's sales outstanding, or the amount of accounts receivable, was expressed as a number of days of average sales. 17. Only Bob, after a few years away, stayed at TSI and worked in the successor company. Leslie went to Rolm as a senior development manager, Joe went first to ADP as vice president of research and development, then to RJR Nabisco and Seagrams as CTO, and Bill became CEO of Gartner Group. 18. An interesting side note is that back in my IBM days I had been asked to evaluate the then brand-new National CSS as a credit risk for its purchase of two IBM System 360/67's, a huge order. I met the founders, heard their business plans, and concluded that they would not last six months, which I reported back to IBM— thus almost making my prophesy come true. 19. See R. Weissman,"CCITT Meeting Recommendations," IEEE Annals of the History of Computing, vol. 24, no. 4, Oct.–Dec. 2002, pp. 43-43. 20. We had long before installed a National CSS terminal (a converted Selectric typewriter that IBM called the 2740) to let our people write Cobol programs interactively. The system improved productivity, but our customers would not pay for it, and we could not afford the monthly charges, which frequently ran into thousands of dollars. Also, the system was so slow that Hal Feinberg, one of the VP/CSS developers, added a "tic" that made the Selectric ball spin occasionally, just to let the users think the system was working and not asleep.