Integrity in Industrial Research

When Lucent Technologies’ Bell Labs management convened a committee last spring to investigate allegations of misconduct in experiments reported by Jan Hendrik Schön and his collaborators, some scientists said they thought that the pressure to produce in an industrial lab could have accounted for fabrication of data or other scientific misconduct. I believe nothing could be further from the truth in this case. Schön did almost all his work in Germany while waiting for a visa to allow him to work at Bell Labs. Surely there was not much corporate pressure while he was still in Europe. In addition, Bertram Batlogg, his mentor at Bell Labs, was about to leave for ETH Zürich throughout most of the time when the experiments were being done. Surely under these circumstances neither person was subject to corporate pressure to produce.

It is true that Bell Labs management was delighted to see new and interesting results coming out of the basic research part of the organization. But frankly, most of the management was paying more attention to applied research because we really were trying to deliver technical value to the company. So, we could be accused of neglect in that sense, but, at the time, we certainly were not putting pressure on basic research scientists to produce applied research. We also were not thinking of using Schön’s results for usable devices. His results were too far on the research side to think of them as practical in the rapid time frame needed for developing a product.

Some people have asserted that, when a company’s business is in a downturn, the pressure to produce is so overwhelming that it’s not surprising if a scientist becomes deceitful. My experience is quite the opposite. Eventually you have to sell a product to a customer and it has to be worth buying. In a downturn, companies generally get conservative and try to produce safe products, that is, products that are closely related to those currently on the market and that have an assured demand. I believe that fraud, abuse, and deception in business occur mostly during times of economic expansion. Look at what happened during the Internet and dot-com boom of the late 1990s. We’re still not finished sorting out all the industrial misconduct during that period.

Basic research in industry

Is it true that academic-style research cannot be done in an industrial lab? Certainly ivory tower research that has no connection to technology is not possible in industry, but I believe that an industrial lab can do basic scientific research that is connected to the technology of the company. To have a broad-based research organization requires at least two characteristics, and very few companies qualify. First, the company must have sufficient revenue that it can afford to support longer-term research. Second, the company must have a breadth of business units that can take advantage of the technologies that come from the research organization.

The type of scientific research done must have a connection to the needs of the company. In the mid-1920s, at Bell Labs, Clinton Davisson and Lester Germer were studying the properties of cathodes for vacuum tubes. That work led to their discovery of electron diffraction in 1927. More recently, Bell Labs did a lot of work on III–V semiconductors. While lasers were being fabricated for optical communications, basic research in III–V semiconductors led to the discovery of the fractional quantum Hall effect by Horst Störmer and Daniel Tsui.

The advantage of corporate research is that it often brings a different perspective than university research. When the first transistor was fabricated at Bell Labs by John Bardeen and Walter Brattain, the inventors were worried that their discovery would be scooped by Purdue University scientists while Bell Labs was patenting the device. Considerable research had been done at Purdue on germanium during World War II, and some of the underlying effects that made the transistor possible had been observed there. Later, when the Purdue scientists were asked why they hadn’t invented the transistor, they said they simply weren’t thinking in terms of devices!

It is now popular to claim that the only reasons a corporation would want to fund a central research laboratory is for advertisement or public relations. This argument ignores the fact that central research labs like Bell Labs and the IBM Thomas J. Watson Research Center have delivered enormous value to those companies. I can cite a long list of products that were made possible by Bell Labs research, and I am sure that IBM has a long list as well. A central research lab brings at least two attractive features to a company. First, it can attract the best scientists and engineers in fields relevant to the company. Second, it offers an environment in which the next day’s product is not pushing the research staff every day. I remember that, at one stage, a Bell Labs group created the best semiconductor device simulation program that was on the market even though our group was much smaller than groups doing simulations at other corporations. I asked one of the researchers how he thought this was possible. After a couple of days thinking about the question, he said it was because the group did not need to have something the next morning but could work on a longer time scale.

Intellectual property

Robert Laughlin recently asserted that intellectual property cannot be sold unless it is kept secret (see Physics Today, December 2002, page 10 ). This assertion seems to ignore the entire patent process, which allows many companies to make large sums of money. If you are doing industrial research and not development, the patent process allows you to reveal the research results without worrying about losing your rights. However, it is true that when detailed designs and processes are developed to make a product, such technical knowledge is not revealed. I do not see how withholding detailed technical knowledge used to build a product is relevant to longer-term industrial research, which is openly revealed. Thus Laughlin’s claim that the source of fraud and deception in industrial research is the necessity of secrecy is completely erroneous. In fact, I find it difficult to see any difference between the amount of secrecy in university research labs and the amount in industrial labs. You could argue that secrecy in university labs is a more serious issue than anything in industry, because universities are supposed to be open and based on unrestricted intellectual inquiry.

In his Reference Frame, Laughlin said that researchers in industry are often not told what is important to the company because knowledgeable technical people do not reveal the problems to them. These people are supposed to be afraid of losing their jobs if someone else solves their problem. That may have happened to Laughlin, but I have seldom found it to be the biggest hurdle to working with development groups within the company but outside the research organization. In my experience, the main issue was that someone who was running a development project needed to be sure that he or she had all aspects of the project under control. It is difficult for such a person to simply hand over an important aspect of the project to someone he or she hardly knows. Thus, the best approach for the researchers and their managers is to develop a trusting relationship with the scientists and engineers in the development organization. That way, they will feel comfortable depending on the research organization and giving it responsibility for a part of a project.

I do agree with Laughlin when he says that physicists should be obsessed with the truth. One of the great accomplishments of the 20th century is the development of the scientific method for getting to scientific truth. Physics is the standard bearer of this method. I recently visited an exhibit called “A Question of Truth” that was put together by the Toronto Science Museum. The accompanying material said that “because modern science was developed by Western European males, it is necessarily biased.” Such a statement indicates a lack of understanding of the scientific process, a process that has held up well throughout the world. Scientific facts can be verified by anyone—Indian, Chinese, African—or even Western European. The process in and of itself is the ultimate guardian against bias, fraud, and any other abuse. I hope all physicists, no matter where they work, will remember the basic tenet of how modern science should be performed and how scientific truth is established.