It's those endeavors outside of the 3000 arena that have enriched what
brings to his HP 3000 experience. For a time he worked at the Nuclear
Laboratory in New Mexico's White Sands Missile Range, where he said his job
calculate how much destruction a full nuclear exchange would yield. After
end of the world in so much detail, he's been happy to lead a company whose
application, QueryCalc, is often used for a task as mundane as sending
because "the HP 3000 sending invoices is an enormously pacifying event." In
posting carefully color-adjusted pictures from Mars on his Web site --
design Voyager-class spacecraft areoshells in the '70s -- he began to
questions on the HP 3000's place in a productive information systems
faith in the capability of the 3000 to serve as the most important resource
in a company's
information environment is strong, based on a scientist's proof instead of
Ordinarily our Q&A consists of several questions, but Atmar answered what
he felt was
the most important one in full detail. We'll print his other answers in
You're one of a small group of developers who's never offered software for sale on the Unix or Windows NT platform, but stuck with the HP 3000 for well over a decade. Why, and do you think your choice has a message for HP 3000 customers tempted to go with solutions on other platforms?
The question of whether or not we should stay with the HP 3000 is one that we've constantly asked ourselves almost every day during the past 10 years, although, quite pleasingly, it's one that we haven't found it necessary to ask ourselves nearly as intensely or as frequently recently. However, 10 years ago -- which was perhaps the blackest time in the history of the HP 3000 -- the answer wasn't nearly as clear.
That was a time when Wim Roelandts, a former vice-president of HP, began visiting HP 3000 VARs and telling them that it was time to move off of the HP 3000 and to begin moving instead to a Unix-based, "open-systems" platform that supported an SQL RDBMS. Indeed, anyone who advocated sticking with the HP 3000 at that time could well have been labeled a fool. Staying with the HP 3000 certainly wasn't a philosophical position that was being strongly advocated within HP itself.
As we all remember, that was also a time when HP rejected all suggestions as to how to improve IMAGE. SIGIMAGE had disbanded in the middle 1980's because nothing it did or suggested made any difference. No feature enhancements had been added to IMAGE by HP during the first 15 years of IMAGE's existence, and HP was certainly not now about to put any new investment into IMAGE.
Perhaps one of the most telling moments came a couple of years later at the Boston Interex meeting in 1990. In one set of rooms near-riots were breaking out. It was the first time in the history of the Interex meeting that HP's managers were being booed from the floor. At precisely the same time, in a different room, Wim Roelandts was telling you in an interview that "We believe that customers in the future are going to buy and do all of their development in SQL [on Unix boxes] and won't need Turbo... The problem is that Turbo has been tuned for over 15 years and there are not many ways we can improve it anymore," although the list of enhancements desired for IMAGE (CIU, b-trees, DDX, etc.) was well-known, quite long and growing.
Ten years ago was also exactly the same time that we began investing heavily in the development of QueryCalc, a new form of report writer that would represent the first major new product development on the HP 3000 in some time. If there was any time in our corporate history when we could have easily been called fools for what we were doing, this was certainly it.
But we didn't blindly go about creating a product that we knew was going to ultimately require several million dollars of investment. We carefully investigated every option: open-systems vs. proprietary, the various operating systems offered by DEC, HP, and IBM, as well as a generic Unix. And we allowed each of the various primary vendors to drop by and present their respective cases as to why we should leave HP and become one of their VARs.
In the end, the decision became simple. We merely decided to do what was best for the small business customer. And we resolved to make all of our business decisions and all of our design decisions based on that single premise.
The very nature of the HP 3000 was core to this approach. I believed then, and only believe more strongly now, that the HP 3000 is the finest commercial-grade database engine in existence. It is nearly the ideal platform on which you can trust your business and operate it with the greatest assurance of results, at the least possible cost. If a computing platform is simple and reliable, then it can be made simpler and more reliable yet. If it's not, then I'm not sure that anything could be done to much improve it.
Nonetheless, there was clearly some significant risk associated with this strategy, given that we knew then that there was a sizable contingent within HP that was trying to kill the HP 3000. We have been VARs with HP since 1976 -- and we'd just gone through a previous, very serious episode where HP had killed the platform on which we had built the greatest portion of our revenue-generating business -- and by all means, I didn't want to do that again.
In 1976, because of our previous experience in building products for the Altair microcomputer, HP invited us to become OEMs for their new 8080-based HP2649 OEM modifiable terminal and suggested to us that we should perhaps convert it into a "microcomputer." We gave that idea serious consideration, but in the end we rejected it because the end product was going to be significantly outside of the price range of the people who were buying PCs at the time.
Nevertheless, we did take HP up on their offer. We rewrote the internal operating system code in an HP2645 Intelligent Terminal and converted it into the AICS System 2000 Word Processing Terminal. With this device, we now had a serious business machine -- one that provided great value to its users, even though the price of the System 2000 was $12,000 per machine in 1980 dollars. When we first released the System 2000 in 1978, our primary competitor was the IBM Mag Card word processor, and it proved to be no real competition at all.
There are perhaps only 10 of the System 2000s still in operation now, 20 years later, but they are still working and we still service them. We personally have two that remain in use in our own operation. The fact that these machines are still capable of running on a daily basis after such a period of time is nothing less than very powerful testament to the quality of products that HP has traditionally built.
But there is a very dark side to HP, too. In approximately 1980, HP apparently decided to kill the HP2645 series terminals. However, they didn't come out and explicitly tell anyone that they had made that decision. Rather, what they elected to do was raise the price of the terminal, gradually at first and then precipitously later, so that by 1982 we were paying $6,000 for an OEM-discounted terminal that we had previously bought for only $1,500 three years earlier.
Only then, in 1982, did HP duplicitously announce that the 264x Series terminals were to be discontinued. "Orders," they said, "had dried up and there was no longer any demand for the devices."
When HP announced the discontinuance of the 264x Series terminals in 1982, we gave very serious consideration to purchasing the "rights" to continue manufacturing the terminals here in Las Cruces. It wouldn't have been nearly as difficult as it might seem. We had already purchased and were well familiar with the source code. More importantly, there were no specialized parts in an HP2645 terminal. Every individual component piece was a standard, "off the rack" device. Indeed, all that we would have been buying from HP was the list of contractors that HP used to manufacture the circuit boards and the various pre-formed plastic parts that appeared in the machines.
If we had followed through with this plan, we estimated that we could have reduced the cost of manufacturing an HP2645 to well under $1,000. If we had done this, we would have been able to not only significantly lower the retail price of the System 2000 but perhaps continue to manufacture and sell HP2645 terminals under the AICS logo at a price point well below HP's earlier retail prices as well.
It wasn't due to a lack of faith in our ability to manufacture HP2645 terminals, at the prices we needed, that eventually persuaded us to abandon the plan. Rather, it was the apparent end of standalone word processors that the rise of the PCs predicted. If the market was coming to an end for standalone word processors, as it ultimately did, continuing on would have made no sense, no matter what quality of device we produced.
I tell this story because in 1987 and 1988, HP began saying very much the same things about the HP 3000 they had said several years before they discontinued the HP2645 terminals. This time, however, I was clearly well tuned to listen to them and hear what they were saying.
The "Boston Riot" was to a degree manufactured. People such as Steve Cooper, Alfredo Rego, Rene Woc, Nick Demos, Ken Sletten and I had grown increasingly more disappointed with HP during the five years prior to Boston. HP simply refused to listen. You could talk to the designated VAR liaison and he would patronize you with such statements as, "That's a very nice idea, Wirt. I'll bring it up at the next meeting," but you could be sure from the tone of his voice that he threw his notes in the trash, if he made any at all, as soon as you hung up the phone. Arrogance and contempt were the two words that best described HP during that period.
It was only after every internal avenue was taken that the issues that concerned us most were finally made public in a series of articles in Interact and The HP Chronicle. But if there was a clear beneficiary to the uproar, beyond the obvious benefit that the users derived, it was HP itself. Before HP was able to enact any tangible changes to IMAGE, the mere fact that they were clearly listening to the users became a source of general pride within the user community.
The difference in attitude that HP displayed as a company before and after Boston was like night and day. Much of that difference is due to a single person, as things often are. HP is a large corporation, and like all large corporations, the last person you speak to comes to represent the entire organization. No one, however, has ever represented a company better than Jim Sartain did during his tenure as director of the IMAGE lab in CSY.
Jim listened very carefully to the users, and he never responded with arrogance or patronization. More importantly, he never lied or misrepresented the truth in any manner. Singlehandedly, Jim restored a profound sense of trust between the users and HP, and he is to be sincerely congratulated for that. His was an extraordinary achievement -- and the change in attitude that Jim helped foster was one of the primary driving forces that very much helped resurrect the HP 3000 from the grave. The positive attitudes and levels of expectation that people feel generally now about HP, CSY, and their enhancement requests being taken seriously is in good part attributable to Jim Sartain.
But to answer your question, "Have we ever been tempted to move onto another platform?" the short answer is: No, not in any serious manner.
That's a choice every software manufacturer has to make. A manufacturer has to choose to develop a strong commitment to one platform or write software that will run anywhere. The choice really boils down to software that runs extremely efficiently on one platform-- but no others-- or runs rather poorly on a very large number of different platforms. We chose the former. But what we really chose was the best machine for our customers.