| About CRA |
| CRA for Students |
| CRA for Faculty |
| Events |
| Jobs |
| Government Affairs |
| Computing Research Blog |
| CRA-Women |
| Projects |
| Publications |
| Data & Resources |
| Membership |
| What's New |
|
Home |
Bi-directional Technology Transfer: Sabbaticals in Industry
| Date: | November 1997 |
| Section: | Opinions |
This note discusses my recent and successful experience with the technology exchange-not just technology transfer-through a ten-month sabbatical in an industrial product group. I advocate that more academics and more companies consider this option. Below I examine normal technology transfer, discuss my sabbatical, and draw conclusions. A version of this note appears in Proc. of the NSF Conference on Experimental Research in Computer Systems, June 20-21, 1996, pp. 201-203, NSF 97-16 and at URL http://www.cs.wisc.edu/~markhill/nsf_es96_pos.ps, reprinted with permission. An assumption underlying my views is that experimental systems research is successful if and only if it eventually affects how some commercial computers are built. Nevertheless, a key qualifier is "eventually." Some researchers should work 3, 6, 12, and 24 years out. Nearer term research will strengthen our country's industrial base, while long-term work can create new industries. To affect how commercial computers are built, researchers must work to get their ideas known to industry. In my view, it is not sufficient to publish a few journal articles. I recommend (and have practiced) the following:
Publish in conferences. Practicing engineers tend to pay much more attention to conferences than
journals. Nevertheless, they do not devote too much time to attending conferences or combing conference proceedings.
Give talks in industry. This can be useful if your research is close enough to get on the "radar screens" of engineers, which, depending on the person, varies from 2 to 5 years out.
Hold annual or semi-annual affiliates meetings. This is very effective. Industrial people are forced to spend a block of time thinking about your project, which enables them to make many useful suggestions. Furthermore, students get excited to see that people care about their work.
Have students graduate to industrial jobs. While this is a slow method, it appears in practice, to be one of the most effective methods of transferring university research into products. This method, however, is not as good at providing feedback to university researchers, in that the feedback is intermittent and delayed (e.g., by going to conferences and sharing drinks with former students).
My Sabbatical:
Mark.Hill@Eng.Sun.Com
A year ago I completed a high-bandwidth, high-latency method of technology exchange between academia and industry: a sabbatical in industry. After getting tenure at the University of Wisconsin, I approached several companies about a sabbatical, where Wisconsin would pay half of my salary and the company the other half.
For several reasons (some idiosyncratic and temporal), I selected Sun Microsystems. An important factor was the vision of Greg Papadopoulos, Chief Technical Officer of Sun Microsystems Computer Corporation, who was willing to hire me for activities to be named later. I began work at Sun in September 1995. Papadopoulos and I decided our goal was for me to work 25% on Wisconsin work, 25% on internal consulting, and 50% time on some Sun project. I was able keep the Wisconsin work limited because my talented colleagues, James Larus and David Wood, stayed home. I found that my academic background gave a more global perspective for the internal consulting, but my effectiveness varied with the flexibility of the listener.
The most rewarding part of my sabbatical was my 50% time project. Papadopoulos and I decided that I would spend the first month selecting the project. I chose to work at co-defining a high-end server product, which, for proprietary reasons, I can't describe. I can say, however, that research by the Wisconsin Wind Tunnel project has proven very pertinent to aspects of the product, but there are other aspects that I had to learn from scratch.
My experience in product design has been fascinating and debunked my belief that academic research is intellectually harder than building a product. Designing a product is easier because you get to make point decisions (and not have to characterize all possibilities) and you only innovate as a last resort. It is much harder, however, because you have to consider many more interacting dimensions, such as reliability, availability, serviceability, power, cooling, physical design, and manufacturability. Furthermore, you have to get it all right, since a product's success seems to be determined by the two dimensions you did least well.
I will be better able to judge the full impact of my sabbatical in five years. Nevertheless, I already see several benefits that will inform my teaching and research:
Equally important, Sun has found my sabbatical well worth their investment. They plan, for example, to increase the opportunity for similar sabbaticals in the future.
In summary, my sabbatical took a lot of time-10 months-but the information exchanged will likely be mutually beneficial for many years to come. While my "sample size of one" does not permit any statistically sound generalizations, my judgment does suggest some advice.
First, academics should consider a sabbatical (or leave) in industry. Even if you get a small fraction of the benefit I did, I would judge it worth your while. I recommend that academics:
Second, companies may wish to consider supporting more sabbaticals too.
While they are expensive-relative to giving a school support for a research assistant-they provide you an employee with a more global perspective and a chance to influence research problems for many years to come. I recommend that companies:
