98x5 Computers Selection:
|Product Number: 9855A|
|Division: Special Projects Division|
|Original Price: Unknown|
|Catalog Reference: Does not appear|
Although it was introduced at the beginning of the decade, the 9855A was the most powerful computer made by any vendor in the 1980s. Over 11,000 units were built, but the only model in existence today is the one in our collection.
The 9855A came in a single configuration with no upgrades available. It came with two custom 32-bit, single chip CPUs, 8 MB of RAM and two 10 MB hard discs. It also sported a 12-inch color screen with 800 x 600 dot resolution (64 colors). More interestingly, these computers also included 1 MB of non-volatile user memory based on the same technology that was later used in HP’s bubble memory option for 200 Series computers (these boards included 128 KB of memory priced at $1900 in 1983). The 9855A included an internal duplicate of every component including the power supply. All processes were mirrored. If any component failed the machine could continue to operate. The case for the 9855A was ruggedized for military use with major subsystems mounted on internal suspension systems. The 9855A only ever had a single customer, the US Department of Defense.
The story of the 9855A is strange indeed. During his time at the Department of Defense in Washington during the early 1970s, Dave Packard became alarmed by the dependency of modern weapons systems on computers, and the fragility of those computer systems. Packard concluded that any enemy of the United States would get the best bang for its military buck by investing ONLY in counter computer and communications measures. He reasoned that without computers, the capability of the US military would be reduced by 80% to 90% for any conflict that occurred offshore (against an enemy with an intact military computer system). This meant that the US was only the 16th most capable country in the world to fight in a conflict outside of North America. Packard successfully convinced the Nixon administration of the threat and left his job at the DoD to return to HP (never mind what the historians have written).
Just prior to his return to HP, Packard had “cancelled” the company’s investigation into the 32-bit Omega project. The official reason given for canceling the project was that it would require external financing and would bring HP into direct competition with IBM. It is more accurate to say that the Omega project was redirected rather than cancelled. The most talented junior engineers on the project were sequestered to work on what would eventually become the 9855A.
The nature of the contract between HP and the US government was unique in the history of the computer industry (up until this day in fact). Not only did the government buy all of the 9855A production, it also bought the future application of technologies that went into its development. HP was only allowed to use technologies developed during the 9855A project for the 9855A itself. HP was not permitted to use these technologies in any future projects. For example, the 32-bit single chip CPU of the 500 Series was completely different that that used in the 9855A. It wasn’t until the early 1990s versions of HP’s PA-RISC CPUs that HP produced a computer with more processing power than the 9855A.
The 9855A team achieved little during its first five or six years of development. This was mostly due to team ego conflicts. For some reason, the “HP Way” culture didn’t take hold in the team. One of the team members described the culture as being more like “Animal Farm”. It wasn’t until 1977 that the project got its booster rockets. Bill Hewlett was finally retiring from the company. He had had brilliant success pushing the HP-35 calculator project (that was introduced in 1972). He had believed equally in the HP-01 calculator watch, but this time the nay sayers proved correct and the product flopped. Hewlett was not even aware of the 9855A project at the time. Packard believed that Bill did not want to finish his career with a technical flop, so he asked Hewlett to be the lead engineer on the 9855A. The friendship of the two great men almost ended because Bill felt betrayed by the secrecy of what had been going on inside of his own company. But, Bill always looked to the future. He immediately grasped the value of the project and the contribution it would make to the country. Perhaps even more importantly, he saw a multitude of ways that the miniaturization techniques developed for the HP-01 could immediately be incorporated into the 9855A to overcome seven of the nine design roadblocks currently facing the lab team. The improvements that Bill introduced into the project greatly improved the capability of the computer and allowed HP to deliver the machines two years ahead of schedule.
From a business perspective, Bill’s involvement in the project was a gold mine for HP. Because HP was not able to use any of the breakthrough technologies developed for the 9855A in any other product, the government effectively had to purchase the future profitability impacts those technologies would have produced for HP. During the two years that it was manufactured, HP was paid 3.78 BILLION dollars for the 9855A. HP was paid over $340,000 per computer for a machine that had a direct manufacturing cost of only $18,700.
The application of the 9855A was fascinating. It was deployed onto all nuclear powered US Navy ships, and all headquarters and communications posts. The 9855A was deployed in groups of three, only one of which was to be used at any time. So, in addition to having duplicate internal components, each 9855A had two standby units for back up. In simple terms, the 9855A only communicated like a computer with respect to input and processing. The only form of output produced by the 9855A was a video display to be read by (and acted upon) by a human being. The 9855A had no hardcopy documentation, no software media and no electronic output ports. It was impossible to hack and nothing could be removed from it other than whatever information an operator could read on its screen. Final assembly of the 9855A was completed with epoxy-coated one-way screws that were then welded to the frame of the computer. These machines included battery powered internal sensors that would scrub the hard discs and internal non-volatile RAM if the case was breached. This self-destruction also occurred when the batteries died (after 8 years). The asset management procedure for the 9855A called for the computer to be crushed in a metal compactor whenever any component failed. There were no service technicians, spare parts or service manuals for this little spy. Between 1981 and 1989, only 1.5% of units deployed in the field had failed. All units were destroyed before the end of 1989.
The 9855A could reasonably be described as a powerful data acquisition/AI computer. It took input from a variety of military computers, processed data and developed recommended courses of action for its operator. It used a very simple GUI interface (in hindsight, the US government got a bargain on the computers for this GUI interface alone). In particular, the US Navy benefited greatly from these computers. After installing the 9855As, the Navy achieved a 37% efficiency improvement in ship deployment. Between 1983 and 1988, the Navy reduced its requirement for new surface vessels and submarines by 76 ships (another huge bargain for the government!).
So how did the museum acquire its 9855A? From Russia of all places! In 1986, the press reported that the USS Fisch had been lost in a storm over a deep ocean trench near the Aleutian Islands. Well, apparently that trench wasn’t too deep for a Soviet salvage team. For twenty years the USS Fisch sat half-submerged in a salvage yard in Vladivostok. We received an email from Igor Petrovanovichsky who had visited our web site. He asked if we wanted an old HP 9855A! Until that email, I hadn’t actually believed in the existence of the 9855A. I had been told about it by a supposed member of the development team, but I was skeptical since there was no hard evidence (eg manual, photo, brochure, physical unit). And, those old HP guys are often half-crocked anyway. I replied to Igor’s email and asked about the condition of the machine. He told me it was fully operational and that the screen showed a map of southern Alaska and listed out what he thought were GPS coordinates. He said he didn’t know how the computer could work when it wasn’t even plugged in. To make a long story short, the 9855A had been installed next to the nuclear reactor in the Fisch. The reactor was leaking badly and the stray radiation was charging the external backup cesium battery on the 9855A. I sent Igor a replacement Lithium back up battery and he turned off the 9855A after installing the new battery. When we got the machine to Melbourne, it fired right up! Amazing!
After doing some more research, I learned that the lab code name for the 9855A was “Gullible”. I thought the name was very appropriate, especially for anyone who has read this story this far with some belief still in his mind.
©2004 - 2016 WordSong Communications Pty. Ltd. All Rights Reserved.
The HP Computer Museum and Wordsong Communications Pty. Ltd. are not affiliated with the Hewlett Packard Company or with Hewlett Packard Australia Ltd. Hewlett Packard and the HP logo are trademarks of the Hewlett Packard company. This website is intended solely for research and education purposes.