Re-inventing the Corporate Research Lab

Last month, my colleague Paul Horn announced his retirement from IBM.  He had been the leader of IBM Research since 1996, a position that John Kelly has now assumed.

I have worked closely with Paul for many years.  His accomplishments are impressive.  Under his leadership, IBM Research developed the Deep Blue supercomputer, which in May of 1997 defeated chess grandmaster Garry Kasparov in a six day match.  A few years later, the Blue Gene initiative was launched, a family of highly parallel supercomputers aimed at solving some of the most complex problems in science and industry.  Since 2004 Blue Gene has been number one in the Top500 list of the most powerful supercomputers in the world.

In technology, the Research division pioneered a number of semiconductor breakthroughs, such as the use of copper wiring to help build faster microprocessors and the use of self assembly, - inspired by biological mechanisms, -  in semiconductor manufacturing. 

Beyond individual products and technologies, Paul led in the creation of entire new disciplines, most famously Services Sciences, which is aimed at helping us better understand the nature of services,  - by far the largest sector in the economy of most countries, - in order to improve their productivity and quality through the use of technology, science and management principles. 

Paul played a major role in the launch of the Autonomic Computing initiative, aimed at making complex IT systems increasingly self-managing.  In 2001 he published the Autonomic Computing Manifesto, a challenge to the IT industry, and the research community in particular, to deal with the growing complexity of IT systems - the largest obstacle standing in the way of continuing progress. 

But, impressive as these accomplishments are, I believe that the key legacy from Paul Horn's tenure as director of IBM Research is his leadership role in transforming the very culture of the institution, and in the process laying the foundation for the corporate R&D lab of the 21st century.

Paul realized that increasingly, the toughest problems, - requiring the kinds of breakthroughs you get from the very best technologist and scientists, such as those in IBM Research and other top labs and universities, -  were out there in the real world, - in the marketplace.  So, if that is where the problems that inspire breakthroughs are, then that is where the research people should personally go to learn about them and hopefully come up with elegant, innovative solutions to the problems, as well as new ideas that might lead to fundamental advances in science and technology.

Paul seemingly agrees.  In a recent interview he said that " . . . his proudest accomplishment has been opening the doors of IBM Research to customers for the first time and encouraging its scientists to collaborate with clients to solve real-world problems.” 

Paul and I actually share a common background - we are both former physicists and we both came to IBM from the University of Chicago, - where he was in the faculty and I was a graduate student.

In physics, you need a very close connection between theory and experiment.  Whatever ideas, models, and equations you come up with, they are only as good as how well they help explain the results of experiments, and shed new light on the physical world around us. 

Similarly, the messy, unpredictable, complex marketplace of business, people and services should be viewed as a kind of real world, full of important problems that need explanations and solutions.  Consequently, that is where the research people need to go and come up with their own theories and solutions to those problems.  In the knowledge economy, the marketplace is our lab.

Most corporate research labs were first established in the years after World War II, and were modeled after research universities.  Their job was to push the frontiers of knowledge by conducting basic research, which was over time applied to develop and bring to market new products.  In this model there were large time gaps between research, product development, and the marketplace, but since both technology and markets advanced at a relatively slow pace, there was enough time to make the needed transitions across the gaps. 

But, the model has been falling apart over the last twenty years.  First, the rate and pace of technology advances has significantly accelerated, most prominently in the IT industry.  The hand-offs and elapsed times to take a technology from the lab to the marketplace inherent in the old model are no longer competitive.  Start-up companies have done away with the gaps altogether, significantly decreased the time-to-market for new products and services, and put huge pressure on those enterprises still operating under the old, kindler and gentler rules.  These competitive pressures, which have been further exacerbated by the forces of deregulation and globalization, have made it unaffordable to continue supporting corporate research labs that have only a loose connection to products and customers. 

It is not that fundamental research, knowledge, and top talent are no longer needed - in fact they are more important than ever given the rising complexities of the problems to be solved, and the increased opportunities to apply new technologies and science to solve them.  But the culture of the research labs had to drastically change.  They had to become far more involved in helping develop the highly sophisticated product and services that their breakthroughs hopefully lead to.  Above all, they had to go out to the marketplace and learn first hand about the increasingly complex problems that desperately require their breakthroughs, knowledge and talents. 

Few corporate labs have made this tough cultural transition, and that is why they have declined.  A number have disappeared altogether or are now shadows of their former selves.  IBM Research not only changed its culture as required, but, in my opinion, it actually re-invented itself into an organization that is now more relevant than ever to both the world of science and technology and the world of IBM and its clients.  I truly believe that this is Paul Horn’s overriding legacy. 

The arc of Paul's and my career continue to move in similar directions.  We both retired from IBM this year, within a few months of each other, and we are both seeking new challenges in academia, where Paul is now headed as Distinguished Scientist in Residence at New York University, while I am getting ready to teach my new graduate seminar at MIT starting in a few weeks. 

After a very successful career at IBM, I wish Paul all the best in his new endeavors.

Comments

[Chris Ward]

The 'commercial' world is splitting apart from the 'scientific' world rather badly.

IBM as a profit-seeking corporation is bounded by the three commercial laws of copyright, patent, and anti-trust.

It employs copyright lawyers with tbe brief to prevent IBM from infringing others' copyrights ... when IBM ships software, it is IBM software, written by people on Sam Palmisano's payroll. Friends and relatives cannot be allowed to help.

It employs patent lawyers whose brief is to ensure that IBM has enough patents to preserve IBM's commercial freedom to engage in business; anyone else with a patent needs to be proudly met with a cross-licence agreement, not with a pile of dollars to buy IBM out of a hole.

It employs anti-trust/competition lawyers with a brief to ensure that IBM's prices are high enough that every business is designed to make a profit. To ensure that when a business commoditises, as the Personal Computer business did, then IBM will not drop its prices below break-even and engage in a loss-making business. Lenovo can run that business profitably, and IBM handed it on.

And if you want it differently, you have to ask the US Congress to fix it. They make the public policy laws; the corporations work under their rule.

Now, the scientific/university world is upside-down on all these counts. They develop software to share with all who will take it; see the One Laptop Per Child project. Not sure about patents; but most scientists treat 'published papers and citations' as the currency which will keep them in salaried employment. As for anti-trust, the universities generally attempt to disseminate knowledge; they don't charge market rates for adult education outreach services.

We are now in the position of Linux, OpenOffice.org, and IBM Lotus Notes wiping the floor with the previous IBM solution (OS/2, SmartSuite, and IBM Lotus Notes); whether it also wipes the floor with the office solution from the Pac North West is just a matter of getting the pricing right, and is hanging in the balance. The idea is to cream the profitable business off the top, and leave the competition with the deep-discount stuff.

So where next ? As a knowledge-based economy, we obviously need the innovation and the people from the Universities to be injected into the Corporations.

How to stick it back together again ?

[Freddie Moran]

Article says it well for all the good things Paul did for IBM Research - a nice summary and read.