close this bookVolume 1: No. 25
View the documentNews -- research industry; operating systems
View the documentNews -- FY'92 at NSF/CISE
View the documentNews -- information sharing
View the documentNews -- job opportunities
View the documentNews -- Asian employment
View the documentNews -- calls for papers
View the documentQuery -- corporate research labs
View the documentDiscussion -- Austin; housing costs
View the documentDiscussion -- community

Calton Pu ( sent in a NY Times article (9/5) on the new Microsoft research lab. Rick Rashid (rfr will leave CMU to serve as Director of Research under Nathan Myhrvold. Rick is one of the developers of the Mach operating system. (Could that be a future direction for Microsoft?) The lab's advisory board now includes Gordon Bell, John Hennessey, Ed Lazowska, Doug Lenat, and Raj Reddy. Rick tells me that he is in full recruiting mode, looking to bring in 50-60 CS researchers in the next two to three years.

HP's Dr. Joel S. Birnbaum has returned to the $1.37B post of VP of R&D and Director of HP Labs, after three years in RISC-based Information Architecture manufacturing. (Before 1980, he was director of computer research at IBM's T.J. Watson laboratory.) Birnbaum succeeds Frank P. Carrubba, who has joined Philips Electronics N.V. in the Netherlands. [Ron Wolf, SJ Mercury, 9/7.]

GE Aerospace will cut 2,000 more jobs, for an announced total of 7,200 since April 1990. The defense industry is in an era of shrinking markets (e.g., in missiles, satellites, and radar systems) and emphasis on lower costs. [SJ Mercury, 9/7.] (Increased R&D would be a reasonable response, but an unlikely one.)

Tim Finin ( pointed me to articles about Sun's extension of its operating system to Intel 80386 and 80486 CPUs, by the middle of next year. This will allow Sun to sell its "Solaris" suite of UNIX software on high-end PCs, for a cost of around $1,000. Like most UNIX-based operating systems, this one will likely require plenty of RAM and hard disk. [SJ Mercury, 9/5/91.] (Why not just buy a workstation?)

Digital Research Inc. is improving it's DR DOS operating system, soon to be available for about $70. Now that Novell is buying DRI (if the shareholders agree), DR DOS may become the operating system of choice for networked PCs. It currently has a 5% market share, far behind MS DOS. [SJ Mercury, 9/7.] This merger has the business community very excited, and some are predicting the fall of Microsoft. "Someone needs to be independent of the hardware. ... Novell is in a good position to provide that for DOS, UNIX, and all the other operating systems." [Darrell Miller, Novell EVP, PC Week, 7/22. Soft.Letter, 8/25.]

HP's New Wave program is doing well in Japan because it runs on more machines than does Windows 3.0. Canon is now introducing a Japanese version. [Japan Marketing Group, MicroTimes, 9/2.]

The Defense Department intends to stick with Ada. (An Air Force study consisting of five sub-studies showed that Ada beats C++ in capability, reusability, readability, large-scale software engineering, reliability, maintainability, lifecycle cost, reducing project risk, and availability of qualified programmers -- for DoD-type projects. C++ was superior in efficiency, compilation speed, object-oriented support, ease of use, integration with off-the-shelf software, popularity, and acceptance.) Ada is currently mandated for all DoD projects unless it is not cost-effective. There are 387 Ada projects, with then number tripling every year. The Navy has 161 projects; the Air Force 140; the Army 51; the Marine Corps 28; and other DoD agencies have 7. DoD is also tracking 168 academic, commercial, civilian government, and international projects. [John Keller, Military & Aerospace Electronics, 8/91.]

Here's news from the NSF Directorate for Computer and Information Science and Engineering (CISE). [Charles Brownstein, CISE NEWS, 8/27/91.] Calton Pu helped me get hooked up with this newsletter. Contact for a free subscription.

Nico Habermann, Dean of CS at CMU, will take leave to serve as the Assistant Director (i.e., head) of CISE, a position last held by Bill Wulf. Chuck Brownstein ( will stay on as his Deputy. Melvin Ciment will continue as NSF coordinator for the High Performance Computing and Communications (HPCC) initiative. The CISE office number is (202) 357-7936.

Rich DeMillo, Division Director for Computer and Computation Research, is returning to Purdue. Bruce Barnes is Deputy and Acting Division Director. (CCR's Director serves under the CISE Assistant Director (!), who serves under the NSF Director, Walter Massey.) Harry Hedges is stepping down as Director of Cross Disciplinary Activities (CDA), but staying to help out in CCR. (That's jargon for keeping the paperwork moving until they can hire full-time directors with suitable interests and credentials.) NSF is recruiting for both the Division Director and CDA positions. I infer that they are also seeking program directors for Programming Languages and Compilers, Operating Systems and Software Systems, and Software Engineering. In fact, you never know what NSF might be looking for. If you want to work there, it makes sense to contact the division director early and often.

CCR's other program directors are Dana S. Richards, Computer Systems; Zeke Zalcstein, Computer Systems; Kamal Abdali, Numeric, Symbolic and Geometic Computation; Harry Hedges, Programming Languages and Compilers; Nat Macon, Operating Systems and Software Systems, and also Software Engineering. The division phone number is (202) 357-9747. (I can supply others if necessary.)

The Division of Information, Robotics & Intelligent Systems (IRIS) consists of YT Chien, Director; Bruce Barnes, NSF-ICOT Visitors Program; Larry Rosenberg, Deputy and also Information Technology and Organizations; Maria Zemankova, Database and Expert Systems; Su-shing Chen, Knowledge Models and Cognitive Systems; Howard Moraff, Robotics and Machine Intelligence; and John Hestenes, Interactive Systems. (202) 357-9572.

The Division of Microelectronic Information Processing Systems (MIPS) consists of Bernie Chern, Director; John Lehmann, Deputy; Bob Grafton, Design, Tools, and Test; Pen-Chung Yew, Microelectronic Systems Architecture; John Cozzens, Circuits and Signal Processing; Gerald Maguire, Experimental Systems; and Paul Hulina, Systems Prototype and Fabrication. The MOSIS program is vacant. (202) 357-7373.

The Division of Advanced Scientific Computing (ASC) includes Tom Weber, Director; Melvyn Ciment, Deputy; Richard Hirsh, Supercomputer Centers; and Merrell Patrick, New Technologies. (202) 357-7558.

The Division of Networking and Communications Research and Infrastructure (NCRI) is adding a program on the National Research and Education Network (NREN), to be run by Robert J. Aiken. Other members are Steve Wolff, Director; Jane Caviness, Deputy; George Strawn, NSFNET; Steven Goldstein, Interagency and International Coordinator; and Aubrey Bush, Networking and Communication Research. (There are several associate program directors as well.) (202) 357-9717.

The Office of Cross Disciplinary Activities (CDA) consists of John Cherniavsky, Acting Head and CISE Institutional Infrastructure and CISE Instrumentation; Gerald Engel, CISE Special Projects; and Caroline Wardle, CISE Educational Infrastructure. (202) 357-7349.

Budget level is still in the hands of Congress, but NSF and CISE expect to do "well" or "very well indeed." CISE may grow 20% because of the High Performance Computing and Communications (HPCC) initiative. All of the program funds are increments to existing programs which support research, instrumentation, education, etc., in the broad range of HPCC topical areas. Mathematics and Biological Resources divisions will team with CISE to support research in "computational biosciences."

CISE intends to increase the average duration of award from 2.4 years to 3.0 years. Content and need will be the deciding factors, as always, but program literature will ask for longer requests. (This means you will win fewer awards and with longer gaps between them, unless you are very skilled. If 20-25% of new proposals are currently winning 2.4 years of funding, something like 16-20% could win 3.0 years -- or an even lower percentage if people submit more proposals to compensate for reduced success. Large proposal-mill departments will do very well, of course, but new faculty members may not. Prestigious lab directors will be happy, NSF's paperwork will most likely be reduced, and Congress will have the warm feeling that NSF is moving to satisfy perceived needs. This will shift funds away from the second and third- ranked schools, however, unless coupled with political pressure to grant more awards to such applicants. Either there will be no outcry and less-successful universities will eventually quit trying for grants, or there will be an enormous outcry and Congress will pump more into research initiation grants and special programs for women, minorities, and smaller schools. Either way, most researchers had better start looking for industrial or Commerce-department funding.)

Dan Corkill ( says that his AI Expert article "Introducing Blackboard Systems" is now out (Sept 1991 issue). It was edited down to meet page requirements, but Dan can provide full text and complete references by request.

IEEE Expert's August issue continues a special track by Bob Simpson ( on DARPA's computer-vision and autonomous navigation research.

Computer Science Press is looking for reviewers for an introductory AI textbook, The Basis of AI, by Donald Tveter ( The book views pattern recognition as fundamental, unifying symbolic, neural, and case-based approaches. Contact Nola J. Hague ( [, 8/25.]

IEEE Expert is looking for book reviewers. See the magazine for instructions.

Debora Weber-Wulff ( is starting an nqthm-users mailing list for users of the NQTHM Boyer-Moore theorem prover. Contact or to join. [, 9/8.]

If you are a user of NeuralWare products, contact Suraj C. Surendrakumar ( about joining the free NeuralWare User Group International (NUGI). [, 9/8.]

Keith Pierce ( is looking for large, non-Ada software products that can be put in the public domain for instructional use. He would like to develop course materials similar to the Ada-based EM-1 software-engineering suite from the Software Engineering Institute (SEI). [, 9/5.]

IntelliSys Corporation is looking for an MS/PhD developer for a geographical information system (GIS). Relevant experience includes expert systems, fuzzy logic, image processing, data compression, and air pollution control. Work will be in Taiwan, New Mexico, and/or New Jersey. Dr. James Liu, (505) 298-2333. [Howard Hou (,, 9/7.]

UMinnesota's Geometry Center, the National Science and Technology Research Center for Computation and Visualization of Geometric Structures, has three $40K research fellowships available. [Mark V. Meuer (, m.j.o, 9/4.]

Bellcore's Multimedia Communications Research organization needs a PhD CS researcher experienced in low-level system issues for its Information Networks Research group. Contact Jonathan Rosenberg ( [m.j.o, 9/6.]

UIllinois at Urbana-Champaign has an AI post-doc position available in its Knowledge-Based Engineering Systems Research Laboratory (KBESRL). Contact Stephen Lu ( [R. Bharat Rao (, m.j.o, 9/6.]

BioCAD Corporation needs database and algorithm software engineers for computer-aided chemistry. Sondra Card (sondra [Gary Chappell (, m.j.o, 9/7.]

Honeywell's Corporate R&D lab needs a senior research engineer in database systems and software engineering. Satya Prabhakar ( [m.j.o, 9/6.]

Thomas Search Consultants is looking for MS/PhD AI/Planning/KR people. One East Coast position involves AI and decision support applied to Army map databases. Chad Thomas (c.thomas, (703) 643-2226. [, 9/10.]

Matsushita Electric Works, Ltd. (a sister company of Matsushita Electric Industrial Co., Ltd.) is recruiting MS/PhD EE/Computer Engineering scientists for its Osaka and Tokyo laboratories. A 2 to 3-year commitment is expected; Japanese ability is not required. Contact Howard Yamamoto (recruiting by 10/18/91. [Tom Borgstrom (tom, m.j.o, 9/9.]

Japanese graduates are in such demand that Sony recently hired 100 liberal-arts majors without asking which universities they had attended. Some 400,000 jobs went unfilled in the recent hiring season, with U.S. companies having an even harder time than Japanese companies. (IBM Japan does rank in the top 20 most- desirable places to work, though.) It is common to spend over $7,000 courting an applicant, with the average hire getting 1.4 job offers. Recruiting expense at Tokyo University may go as high as $45,000 per prospect. "When 'good' students visit our plant, all of the appointments and deadlines of the day are automatically postponed." Apparently, Japanese companies are convinced that these top students really are worth the cost. (In the U.S., of course, companies can't count on a fast-track hire staying for very long. Japan is different.) [Kay Itoi, Newsweek, 9/16/91.]

On the other hand, many foreign companies are having success in recruiting Japanese employees. DEC Japan hired 295 graduates last year, and expects to hire 400 this year. It's true that there are 143 jobs for every 100 job seekers, but U.S. companies can draw recruits with 10% better pay, shorter hours, more flexibility, merit-based promotions, and more jobs for women. Also, foreign companies are building better cultures now that they have stopped confusing English fluency with managerial competence. [Robert Neff, Business Week, 6/24.]

Incidentally, NSF has been very well served by it's staff in Tokyo. One man in particular -- I regret that I don't know his name -- joined decades ago when NSF first opened the office. It was a brave thing to do. He has made a career of serving NSF well, and has provided translation skills, office management, and many other essential services for the rotating staff. It is difficult to find such dedication in the U.S.

Japan is short about 500,000 programmers, which means that companies are losing business and may even be cutting corners. Many companies are moving their programming to other Asian countries where salaries may be 80% less. China's programmers are familiar with Kanji; Malaysia's have often been trained in the U.S., Canada, England, or Australia. Singapore, India, and the Philippines are also popular. Andersen Consulting has had a 500- person business applications software house in the Philippines for six years, and finds it cost-effective for simple jobs requiring at least 1000 person-days. 20% of their work is for the Japanese. NEC opened its 40-person Singapore Software Development Center in 1985. IBM Japan Ltd. has groups in China, India, and Singapore. A job that would cost $120K in Japan might cost $50K in India and $35K in Shenzen. [Lori Valigra, ComputerWorld, 8/26.]

While U.S.-based venture funds are looking for investments in Asia, many Japanese funds are looking in the U.S. The Japanese prefer minority investments in technology niches and market niches. "They are more likely to develop new businesses that involve technology infusion, or maybe cross-licensing, than to fund raw startups," according to George Koo, chairman of the SOURCES 1991: Conference for Asian Financing and Funding. Asian investors look for business development rather than immediate return on investment. The large Asian immigrant population in Silicon Valley helps draw Asian venture capital, which often often comes from high personal savings rates abroad. [Murali Gometam, MicroTimes, 9/2.]

The SIGART Bulletin is eager to print abstracts of recent Ph.D. dissertations. Send your abstract with author, title, granting institution, and author's current address to Include your ACM member number if you have one. [Lewis Johnson (,, 8/23.]

Genetic algorithms; IEEE Expert. John J. Grefenstette ( This is a continuing "special track" rather than a single issue.

AI in text-based information systems; IEEE Expert. Contact Dik Lee ( or W. Bruce Croft ( by 10/25/91.

THE INFORMATION SOCIETY covers political, social, economic, and cultural implications of the emerging information society. Published quarterly by Taylor & Francis, $80 (or $40 personal rate), (800) 821-8312. Submit contributions to Dr. Robert H. Anderson (, Editor-in-Chief. [Rob Kling (, comp.infosystems, 9/7.]

Neural networks for oceanic engineering; IEEE Journal of Oceanic Engineering. Contact Patrick K. Simpson ( by 2/2/92. [, 9/4.]

Parallelization in Inference Systems, a book in the AI subseries of Springer Lecture Notes in Computer Science. Technical papers and half-page project summaries are solicited. Contact Bertram Fronhoefer ( by 10/15/91. [, 9/10.]

Matthew Witten ( would like to know if companies other than Microsoft are building research groups with no expectation of direct return. Is corporate/government/ philanthropic pure research still alive in the U.S.? Or internationally? Does it exist at universities and NSF-funded centers? Where can one find scientific communities in full-time pursuit of new understanding?

I'd also love to get some insight on this. I know that CSLI is having trouble finding funding for linguistics research, and the SRI AI Center has been hurt by military/DARPA cutbacks. Xerox PARC seems to be doing well, although hiring has been frozen (the last I heard). AT&T Bell Labs, Bellcore, and MCC are healthy, but more applied than they were. Small expert-systems companies are being bought out and made to serve immediate corporate needs. Can anyone describe the research groups at HP Labs, IBM's T.J. Watson center, Lockheed's AI Center, etc.? Who's hiring? Where does one find non-academic pure research?

Matthew also commented on life in Austin. Drivers are careless about signaling turns and stopping at red lights, but aren't as bad as Boston drivers. [In Boston, you can pull halfway into traffic and stop or stall with no danger. It's expected.] Transplanted Easterners like Matt do not always adapt easily. Texas SAT scores are low, and Matthew's three kids -- 6,8,13 -- get much of their real education at home. Other than that -- and the poisonous spiders, scorpions, snakes, fire ants, poison ivy, oak, and sumac -- it's a good city to raise kids. The weather gets hot and sticky, and allergists are the number one medical occupation. Housing is reasonable, utilities somewhat high. Income is not taxed by the state, but everything else is.

New statistics from the National Association of Home Builders show that the 13 least-affordable housing areas (for the local income distribution) are all in California. San Francisco heads the list: the median income of $49,900 could support a $139,720 house, but the median house is $279,000. Only 9.2% of households in San Francisco can afford to buy a home there. Outside of California, New York City is the least affordable; urban areas of the Northeast are also high. [AP. SJ Mercury, 9/10.]

I've just seen a play based on Willa Cather's 1913 novel, O PIONEERS!. It's about immigrants scratching a living on the Nebraska plains. In the early days, it was important to share knowledge of crops and cultivation, home building and religion. Each new immigrant was a resource, each that left was a loss to the community. Then prosperity came, and people divided into ethnic and religious cliques. Elders resented that some had prospered more than others, and the young had time to wonder what they really wanted in life. Gossip became less positive, and customary civility less binding. Shared values gave way to legal wrangling. Without a common goal -- survival, prosperity, civilization -- there was little common culture or sense of community. Having reached a decent level of civilization, there was nothing left to work for.

I grew up in Kansas, and I sense truth in the portrayal. I was a math professor's kid, and was essentially outside the community of farmers and the businessmen who served them. The farmers, though, felt outside the currents of modern civilization -- perhaps I was more mainstream than they. Economic survival was (is) still a struggle, but the sense of community was weak. Churches were failing to draw young people. Few people knew their church organist, and few still knew the mayor.

Now I live in cities, among strangers. Most of us do, because that's where the corporate and academic jobs are. Who cares if you succeed in business, buy or sell land, or add a room to your house? You can join churches and organizations, but it takes at least three years -- sometimes a lifetime -- to develop a sense of belonging. My kids are in soccer and karate and other activities, and raising kids is always a tie to other parents. Still, there's little sense of shared struggle. And few people who know or care enough to celebrate our triumphs or to mourn our setbacks.

And what about AI? You can subscribe to any number of journals without anyone noticing your name. You can attend conferences without anyone bothering to read your name badge. There was a time when people with diverse backgrounds enjoyed swapping tips and stories, but now the field is broken into specialties. Biologists talk with genetic-algorithm enthusiasts, and neurophysiologists with neural-net theoreticians; that's about it. Where is the sense of community? What is our struggle? Are we too civilized? -- Ken