How It Is: A Memoir

I was in Long Beach, California in 2009.  The expert from the software company that sold software for Computer-Aided Engineering was on a riff.  “It’s no longer good enough to use fudge factors, safety margins, guess estimates when doing a preliminary design of a new product: a car, an aircraft, a consumer product.  Make computer simulation an integral part of your design process.  It’s time to get closer to reality!”   He then explained how both Boeing and Airbus had suffered long delays in the 787 and A380 projects, because design engineers had not taken an integrated approach to simulation, had relied on outdated methods that worked in the past, but were no longer relevant for modern aircraft designs using composite materials.

Yes, if only they had been smarter, I mused.  The 787 would have been flying by now instead of suffering interminable delays.  Boeing would be delivering these new airliners, not having to accept cancelled orders.  Perhaps I could have flown on one coming back from a recent trip to Asia.  I had visited Korea, China, Vietnam and India on business.  My flight from Bangalore to Frankfurt on a Lufthansa Boeing 777 in coach had been particularly uncomfortable.  I'd been unlucky not to get an upgrade to business class.  Would the 787 provide more legroom?   I was very impressed by Jet Blue.  I’d flown on an Airbus 320 down to Long Beach, California from Seattle, and marveled at the legroom provided in coach.  If only international flights would offer the same space.

I had stayed on the Queen Mary because most of the hotels in Long Beach were fully booked.  When I had stayed there before, some twenty-five years before, with my wife and two young children, we had seen a cockroach in our cabin.  No such luck this time.  I enjoyed having time to circumnavigate the ship on its various decks.  It was while I was on the sundeck enjoying the balmy Long Beach evening air that Nigel had reached me on my cell phone.  What had prompted Mr. Nguyen to call Nigel?  Nigel was like a modern Diogenes, going around Asia with lamp upraised, looking for an honest man.  Perhaps he had found one in Mr. Nguyen.

While we were in Vietnam, one of the meetings that the reseller had organized for us was with the major oil & gas company in Vietnam.  Two representatives of this company had flown up especially from Ho Chi Minh City, formerly Saigon, to discuss their computing requirements.  The meeting had gone well.  Now it appeared that Mr. Nguyen, the interpreter, had alerted Nigel to some unseemly relationship he suspected between our reseller and the prospective customer.

Apparently, Long Beach was hosting a convention of California Business Women.  That morning, a taxi driver told me that there were 15,000 women registered.  We had driven by the Convention Center that morning.  Sure enough, large groups of women appeared to be converging.  At the meeting I was attending, only one woman, of Asian descent, was in the audience of about forty people, presumably all engineers.  Women did not appear to be very evident in Computer-Aided Engineering in the U.S.  In China and Vietnam, I noted to myself, women were key members of the management at two of our resellers, and women had featured prominently among the customers we visited.  Today’s meeting had been advertised as taking place in the historic landmark Madison Restaurant in downtown Long Beach.  When I got to the Madison that morning, I was redirected across the street to a sister restaurant, L’Opera.  A film crew was using the Madison as a set for a movie to be tentatively called: “Dinner with Idiots.”  I looked wistfully across the street at the production crew.  At least half of them were women.

After thirty years, I had become cynical about my sales role in the computer industry.  At the beginning, I truly believed in the benefits that powerful, expensive supercomputers would bring.  “If you want to study the moon,” I told people, “Isn’t it better to use a large telescope rather than give everybody a pair of binoculars.”    I was proud of the achievements my customers had made.  Typical of them was Paul Rubbert and his team at Boeing.  They had shown via simulations run on a Cray supercomputer that it was possible to position an aircraft engine up and forward of a wing’s leading edge.  To get Boeing’s new designs off the ground larger engines were needed.  Paul Rubbert told me that it had been an unwritten rule that engines had to hang underneath the wing.  Without that breakthrough in design, bigger engines would have been scraping the tarmac.  I thought about Paul’s work every time I flew on a Boeing airliner.

I was philosophical now because, whereas the breakthroughs I had witnessed owed much to the purpose-built nature of the supercomputers I sold, purpose-built supercomputers appeared to have gone the way of the dinosaurs.   Almost all supercomputers made then, and today are based on commodity microprocessors designed by Intel or AMD and/or NVIDIA.   Our foreign competitors would not have to work very hard to design competing systems.  They would just have to buy chips produced in factories in places like China and Malaysia.  The systems I had once sold were now museum pieces.  In fact, I was scheduled to attend a similar meeting as today’s, which was being held in the Computer History Museum in Mountain View near San Francisco.  Prominently displayed in the lobby would be a Cray-1 supercomputer, the same type of machine that Paul Rubbert’s team had used.

The theme of the meeting I attended appeared to be that no excuses needed to be made now for computer simulation.  Computer simulation was so good that regulatory agencies were going to insist that manufacturers use it.  The simulation tools were only going to get better with even more fidelity and an ability to couple multi-physics, such as when designing cars, simulating suspensions, rolling tires on a bumpy road and the comfort of passengers inside a vehicle.  I mused on this.  If only one could use a supercomputer to predict what was going to happen in life.   Could simulation deal with matters like inadvertently working for a crooked company.  I doubted it.  It remains true that “you learn from experience!”  

In the job I was hoping to accept, I would no longer be selling supercomputers.  The Battelle Memorial Institute is a major, not-for-profit, research organization.  A typical Battelle scientist cost Battelle about $300,000 a year to maintain, back then.  My job would be to help connect such scientists working on renewable energy, carbon sequestration, climate studies and related fields with more funding sources.  Nigel was fond of talking about “a race against time” as the reason why customers should invest in supercomputers.  In my new job, I would be helping to accelerate R&D that was focused on trying to slow climate change, a true race against time.

Kendall Square Research (KSR) was the ill-fated startup I had joined after spending fourteen years at Cray Research.  The bulk of their funding had come from Bill Koch, the billionaire winner of the Americas Cup.  After things fell apart at KSR, Bill Koch took the helm.  Wisely so, as he had deep pockets.  Investor lawsuits were piling up with his personal fortune as their target.   I had received a contract from Battelle for a KSR purpose-built supercomputer.  A group at Battelle had developed a set of software tools called Global Arrays that facilitated computational science.  Much of the work was subsequently carried out on the KSR supercomputer I had sold them.  Now, Global Arrays software tools are widely used in computational chemistry applications used to develop new materials and drug molecules.  Back in the mid 1990’s, Bill Koch had flown into the Tri-Cities to visit Battelle.  A gentleman of the old school, he had given them the option of cancelling their order if they were uncomfortable about proceeding with a company so badly tainted as we now were.  A widely read recent article in the Wall Street Journal had been particularly damning.  But Battelle still wanted our machine.  Its special design offered great attractions for the development of Global Arrays.  This had been a major competitive win for me.  We had beaten IBM on this deal. 

The young IBM salesman who opposed me at Battelle, was Peter Ungaro.   Later in his career, Peter Ungaro had been recruited by our Board of Directors to become the CEO of the third startup I had worked for, a Seattle-based company called Tera that had bought the assets of Cray Research from SGI and renamed the resulting company, Cray Inc.  I had been blissfully unaware of our prior history until pointedly reminded of it on the first occasion when I had an interaction with our new executive leader.  This points out a moral.  Be careful whom you outsell in business.  You may end up working for that very person one day.  By taking Battelle’s offer, I would be putting all of this history behind me and starting anew.  

A medical doctor was now giving a very interesting presentation on how to use computer simulation to better design prosthetic legs.  He showed some videos of a double leg amputee patient skiing downhill on prosthetic legs that were instrumented with sensors to collect stress and loading information that could be fed back into the design process to further refine the design.  I am somewhat of a late bloomer, I thought.   I had taken up skiing in my early sixties.  Now that I was fully enrolled in Medicare, instead of retiring I was about to embark on a new career.  

My pending job offer was being delayed while a degree exception form was passed around Battelle management for signatures.  If I had stayed the course, when I was enrolled in a part-time degree program at the University of London, this would have been unnecessary.  I had dropped out of taking courses because I was convinced, I wanted to write poetry full-time.  My early work in the computer industry had financed a couple of years of freedom to write poetry, but, as Mr. Micawber predicted, spending one penny more than you earn on a regular basis is a sure path to bankruptcy.  I was spending several pennies, if not hundreds of pennies for every penny earned!  I remembered my bravado in telling people at the time that all a poet needed to survive was his voice!  I was full of notions about medieval poets wandering the courts of Europe and singing for their supper.  It was ironic that I had become a successful salesman perhaps because of a convincing voice.

Sometimes disparate events occur almost simultaneously and assume a psychological significance.  As I got ready to leave Cray, it seemed these synchronicities were starting to pile up.  For example, on the same day that I told Cray I was leaving to take a job at Battelle’s Pacific Northwest National Laboratory (PNNL), Inderjit Badhwar sent me an e-version of Gfiles. The cover story of this issue of Gfiles concerned setting up a national gas grid in India.  A quick glimpse at the archive showed that the magazine was tackling many of the same national security, energy policy and clean energy issues that I would be concerned with in my new job at PNNL.   A good omen, I thought, thinking of the metal statuette of Ganesh, a Hindu god associated with new beginnings, that I had acquired in Bangalore on my last trip, hopeful that Ganesh would help me successfully transition from Cray to PNNL. 

  “We be of one blood, ye and I.”  Mowgli’s master words to use in the jungle when encountering potentially hostile animals were taught him by Baloo, the bear.  When I was young, my master words to survive in life situations were taught me by people like Indi Badhwar.   Indi considers Rudyard Kipling to be a great Indian writer, no matter that Kipling was born and bred in England.  I was born and bred in England and admire Kipling’s work, especially the Jungle Book.  We be of like mind, Indi and I, and, of like age.  I first met Indi, years ago in New York, when he was a married student at Columbia School of Journalism, and we lived in the same apartment house.  From a warrior caste of maharajahs, his upper-class behavior was a great influence on me, the son of a British Army soldier, who had once served in India, and a domestic servant, once a maid to a lady-in-waiting at Buckingham Palace.   Indi’s noisy confrontations with the landlady we shared in New York were a great inspiration.  

Another synchronicity, on the very same day that I heard from Indi, Shambhu Singh contacted me by email from exile in his new post somewhere in Nagaland, India.  When I first met Shambhu, he was an official in the Indian Civil Service’s Department of Science and Technology.  He asked me if I had read George Orwell’s Burmese Days.  I had not.   You must read it, Shambhu had said.  That book, even though published in the 1930s, would provide a very good description of Nagaland, his home State, even today.  If Shambhu had intended me to see the story as one about class distinctions and corruption in colonial India, I saw instead, when I read the book soon thereafter, a parallel between the protagonist who commits suicide out of despair and Shambhu himself, who, threatened with having to leave a prestigious post in New Delhi and return to his home State, was drinking heavily. 

Whenever I think of Shambhu, I remember his patent method for disposing of files.  In the Indian Civil Service, a file is the key unit of governance.  To hear that a file has “moved” or “is moving” is blissful news to anybody depending upon a decision or an outcome related to that file.   Such files start with perhaps a single sheet enclosed in a cardboard folder tied with a colorful red ribbon.  They grow to be bulging reams of paper as the Indian bureaucracy processes the file.   During our first meeting, I watched Shambhu append his signature to a page that already had collected several others.  It seemed that this file was destined to be passed on for further review to a long list of as yet unsigned names.  Then he closed the file, tied the ribbon that enclosed its bulging contents with an elegant bow, lifted it from his desk, and tossed it in a slow ponderous, practiced trajectory that arced across his office and culminated with the floorboards not far from the door.  Several other files had already been tossed over in that direction and lay scattered against the wall where a minion would come to remove them.  I observed all this with interest.  Overhead a large fan slowly circled.  A fly buzzed around a plate of Indian cookies that had been brought in to accompany hot tea.  The muted cacophony of Indian road traffic could be heard in the distance.   Shambhu put on his jacket and then, as it was nearing 5pm, we left to go get a drink.  

On my most recent trip to India, one of our resellers had talked about having a project with the State Police in Nagaland.  This had to do with software to manage phone taps authorized by a magistrate.  The phone companies involved would make electronic copies of any phone conversation associated with certain phone numbers.  That same reseller also wanted to use our technology to study social networks.   On returning to the U.S., I attended a conference, just before I left Cray, where a distinguished American Professor, David Bader, addressed this very topic.  People in the audience had laughed when he began his presentation showing his online Facebook page.  They were puzzled after he clicked on the Facebook statistics icon to show that over 300 million people with accounts on Facebook might be online at that moment.  Then he described how his research was developing software that could monitor trends through an analysis of Facebook traffic.  Such software could also show relationships between the kinds of people who were prone to set off bombs in places like Nagaland.

Coincidentally, on the same day I gave notice at Cray, I was scheduled to visit PNNL.  A Cray CX1 computer system I had sold them had problems.  The problems, in the main, resulted from Cray’s decision to outsource both the design and manufacturing of the CX1 line of computers, supposedly a server that was quiet enough to house in a office environment, and its warranty support to third parties.  The word “outsourcing” conjures, usually bad, memories of phone conversations with foreign-sounding employees of help desks.  But it was to Canadian and US firms that Cray had entrusted its good brand name.

Cray had ordered some of these CX1 systems for its own internal use in Chippewa Falls, Wisconsin.   Chippewa Falls is where Cray has its roots.  It was the home of the legendary supercomputer designer, Seymour Cray, who founded the original Cray Research back in 1972.  If there is one place on the planet where you don’t want to expose shoddy computer design, it is Chippewa Falls.  Somebody had quickly noticed that the Cray-branded CX1 equipment shipped from the Canadian supplier had an Underwriters Lab (UL) safety certification “pending” sticker where the normal UL certification plate should be.  After due diligence, it was discovered that the supplier had not even begun the process of UL safety certification even though they had been shipping the equipment for about six months. Were they hoping that “UL Pending” would be good enough to last for the product’s normal life?  Once noticed, a six-month long process was begun to get the product properly certified.  In the meantime, Cray had frozen shipments of the CX1 except in cases where a customer, like PNNL, had an urgent need for the system and was prepared to accept what was now called an “engineering sample”.   During the ensuing safety certification process insisted on by Cray, it was discovered that changes needed to be made to every chassis and to every blade in each chassis, ever shipped.  These changes constituted a logistical and support nightmare!  

It was now the turn of the third-party support company to show its stuff.   Local contractors, hired for the occasion, were going to each customer site to replace uncertified parts and to overlay the UL Pending tags with ones saying UL Certified.  Their success rate was about 50%.  For every system they fixed, they left another one broken.  To my mind, depending on contractors hired for the day to carry out the maintenance was akin to picking up day laborers from the parking lot outside Home Depot and expecting them to do sophisticated work without training!  In PNNL’s case, the contractors had been onsite on three separate occasions.  They were not being invited back.   A skilled Cray technician and I were now going over to PNNL in hopes of fixing the problems and returning PNNL’s system to productive use. 

The day I gave notice was also the day when the fourteenth anniversary of my joining Cray was announced at the weekly Company meeting.  This was my second fourteen-year stint at Cray. During my first fourteen years there, I had grown disenchanted with Cray Research’s arrogance about what customers needed and what customers, like Boeing, actually wanted to buy.   The era of very expensive, purpose-built supercomputers was ending as systems based on mass-produced microprocessors took over.  Back then, I had organized a visit to Cray HQ by the head of Boeing’s mathematical science group, Al Erisman.  He had given a keynote address to the Cray management team and explained how Boeing saw merit in, and needed access to, systems based on powerful microprocessors; and he pointed out, Cray’s purpose-built systems were falling behind.  At the time, Cray’s management team was unreceptive.  Seymour Cray had left Cray some years before to start a competing company.  Another key designer, Steve Chen, had spun off his own supercomputer startup company together with ex-Cray colleagues.  Exciting new organizations such as Intel’s supercomputer division, Thinking Machines and NCUBE were taking business away from Cray.  Cray management was offended that Boeing was closely monitoring both Seymour’s new startup and Steve Chen’s efforts and dismayed that Boeing was consulting with the Japanese company, NEC, one of Cray’s competitors. They were also offended that key Government customers had stopped buying big Cray systems and were instead buying large systems from Intel and Thinking Machines.  Despite mounting evidence to the contrary, for the next several years, Cray would continue to insist that the so-called “massively parallel” systems built using microprocessors could not compete with the purpose-built high bandwidth systems that Cray produced.

On the same day that I gave notice, an email was sent out to all Cray employees asking them to complete a survey and contribute their views about what Cray, as a company, could do to recapture the spirit of innovation for which the company had once been famous.  Implicit in this was the widely held view that Cray was no longer an innovative company.   

Now it’s 2024.  Cray was acquired by HPE.  I got a degree in Classics while working for PNNL in a collaboration at the University of Washington.  I am now working at Trovares Inc. a small startup company commercializing software developed at PNNL.  It’s purpose-built software for graph analytics that simulates a supercomputer (like Burton Smith's machine - but that is another story).  Just as in 2009, Boeing is having quality control issues.  The new big thing is Generational AI.  A chatbot will be our co-pilot as we go on.   Self-driving cars will transport us.  A metaverse will surround us.  Computers have come on.

END