close this bookVolume 1: No. 10
View the documentQuery -- Communique replacement policy
View the documentNews -- Jonathan J. King
View the documentOpportunities -- Xerox KBSCC; GTE; IU/NN jobs; headhunters
View the documentNews -- DARPA and NSF changes
View the documentNews -- Pope OK's profit
View the documentNews -- MBA
View the documentNews -- research vs. teaching vs. corporate careers
View the documentDiscussion -- workstyles
View the documentDiscussion -- Bootstrap Institute; groupware
View the documentQuery -- Asian research labs

Leonard Foner ( asked about replacement copies of the Communique in the event that someone loses access to his corporate mainframe files. No problem; just let me know and I will send replacements. Hardcopy newsletters are a bit of a problem, but electronic copies can be sent immediately. Mac-formatted floppies are also a possibility.

AI Magazine (Summer '91) carries an obituary for Jonathan J. King, who died of cancer at age 41. His 1981 Stanford doctoral dissertation was on knowledge-based inference in database queries. He had worked at NIH, and subsequently worked at HP, Symantec, Teknowledge, and Sun. He was also an associate editor of AI Magazine from 1983 through 1989.

Xerox's Knowledge Based Systems Competency Center in Rochester, NY, is seeking people experienced in applying KBS to real-world problems. Contact Mark Maletz, (716) 422-8032. [AI Magazine, Summer '91.]

GTE Labs (Waltham, MA) is advertising for a speech-recognition researcher. [IEEE Spectrum, 6/91.]

A San Diego company is advertising for a Staff Scientist Ph.D. in EE with in-depth knowledge of low-level image processing, motion estimation, brain theory, neural networks (esp. optical implementations), geometrical modeling, wavelet theory, pattern recognition, statistical inference, and software engineering. (Whew!) The work is to design and analyze traditional and neural- network image processing algorithms, analyze the efficiency of implementation on a NN ViP chip, and develop an unsupervised automatic target algorithm for multispectral/multitemporal infrared image data. Listed salary is $1057/week. Send resume to Job @ CR4158, P.O. Box 9560, Sacramento, CA 95823-0560 by June 30. [IEEE Grid, 6/91.]

A similar job in Torrence, CA, with stronger electro-optical emphasis (as well as C and LISP) is listed in IEEE Spectrum, 6/91. Job #MD21153, P.O. Box 9560, Sacramento, CA 95823-0560 by 7/1/91.

Listed salaries are meaningless if you can convince them that they'll make money by hiring you. It's best to leave salary open for later negotiation, no matter what they say. Even sending a resume is chancy -- it just gives the headhunter an easy way to screen you out. ("What, no expertise in brain theory? Too bad, she would have been great.") Try to find out something about the position, then tailor your resume to fit.

Suppose you know the local companies, and could probably hit the hiring authority directly by mail or through your own contacts. Should you do it? Probably not. The boss has hired a search firm on retainer, and has to pay them no matter who is hired. Your direct contact would be forwarded to the recruiter for evaluation. The recruiter, though, wants to prove that hiring him was a wise move (and should be repeated for the next job search). He is likely to find fault with you and recommend several of his own contacts more highly. He may also stall until you've found another job, or sabotage you in some other way. So go to the headhunter first and get him to champion you; only if that fails should you go after the job directly.

A contingency-based recruiter is best avoided, though. He doesn't have an exclusive listing, and profits only if he gets your resume in to Personnel before anyone else does -- so he'll send it to as many companies as he can think of (unless you forbid it). Each such contact adds a substantial fee (e.g., 35% of salary for an executive-level position) to your cost, and that claim lasts for six months. Such recruiters can be useful if you're new to an area or industry, or if you've exhausted all avenues of direct contact. Entry-level people also find them useful, as the cost penalty is small. Otherwise, it's better to conduct your own mail or networking campaign.

If a recruiter calls you with a specific opening, find out whether it's an exclusive listing. If it's not, you can tell him that you'd rather represent yourself -- even if he's told you enough that you can identify the employer. If it's a job that you wouldn't have found on your own, though, the recruiter has earned his fee.

Several recruiters have recently asked me to update my resume, so it may be that they have some Ph.D.-level software jobs opening up. One is Randy B. Enzian, Epoch Software Systems, Inc., 3080 Olcott Street, Suite 208A, Santa Clara, CA 95054-3209; (408) 492- 1178. He says he's looking for imaging and signal-processing people for defense projects. Another headhunter is Marc Williams, who has recently joined Optimum Executive Search, 488 Waller Street, San Francisco, CA 94117; (415) 863-1014, (415) 626-2568 Fax. Then there's Roger L. Laton, President of Digital Software Corporation, 275 Saratoga Avenue, Suite 260, Santa Clara, CA 95050; (408) 985-0606.

Opportunities -- IJCNN deadline extended; speakers sought at Brown:

The deadline for submission of full papers to IJCNN'91 in Singapore has been extended to June 30. [J.N. Hwang (hwang, Connectionists.]

Bob Weiner ( is looking for speakers for his Future Generation Information Technologies Symposium during the latter part of this year. Topics of interest are hypermedia, visualization, information services, delivery technologies, development environments, intelligent access support (agents, expert systems, neural nets), distributed information systems, and information personalization. [IRList.]

Mark Weiser ( wanted to know what's been happening at DARPA, so I checked with Oscar Firschein. (He's about to take over Image Understanding from Rand Waltzman, as I reported earlier.) Oscar says that IPTO, the Information Processing Technology Office, has now become SISTO (Systems Information Sciences Technical Office?), still under Barry Boehm. The second in command is Erik Mettala, in charge of the Unmanned Ground Vehicle project -- similar to the old ALV project, but with smaller vehicles and an emphasis on team action. [Stan Rosenschein's Palo Alto company, Teleos, has a part of that.] Other program officers are Oscar (IU), Steve Cross (expert systems), Tom Chrystal (natural language), Charles Wayne (speech), and, in August, Geo Wiederhold ( for databases. There is also a program in software correctness. The robotics program under Bob Rosenfeld closed down some time ago.

Barbara Yoon (neural networks) is in a new computation-related office under Steve Squires. Tom Schwartz says there's a rumor that she will soon sponsor a second neural-network competition.

I've also done some checking on NSF's ERC and STC centers programs. Marshall Lih ( says that funding for another round of Engineering Research Centers won't be announced until at least Winter. He's sending me literature on the smaller State/Industry-University Cooperative Research Centers (S/IUCRCs).

Sonja Sperlich ( tells me that Congress has yet to authorize a 3rd round of STC competition. (FY'92 STC funds will be used to strengthen existing centers.) A new competition may be announced in a few months. Bill Harris is still the nominal STC Director, but is spending his time as assistant to NSF's new Director, Walter E. Massey. Rich Hersh has moved to ASC, the Advanced Scientific Computing program.

In his latest social encyclical, Pope John Paul II "embraces market economics with an enthusiasm that, despite important qualifications and nagging moral doubts, marks a new departure in papal teachings." He says that "the free market is the most efficient instrument for utilizing resources and effectively responding to needs," and that "the Church acknowledges the legitimate role of profit as an indication that a business is functioning well." Education and a strong sense of responsibility are needed to make it work, though. "It is not wrong to want to live better; what is wrong is a style of life which is presumed to be better when it is directed towards 'having' rather than 'being,' and which wants to have more, not in order to be more but in order to spend life in enjoyment as an end in itself." [Newsweek, 5/13.]

Think that an MBA degree would help your career? Think again. The MBA is of greatest use as a competitive edge for entry-level jobs. Experienced professionals who interrupt their careers to obtain one may be unable to match previous earning levels. Young MBAs are also finding jobs hard to obtain, and many from the top schools are having to take jobs at small companies and start-ups. [The good news is that small companies may be better managed!] Recruiters still tour the major universities, but they may be seeking half as many candidates as before the recession. One factor is that the supply of MBAs has greatly increased: 21,400 were awarded in 1979; 75,000 last year. Over 700 business schools in the U.S. now pump out three MBAs for every available position, according to the American Council on Education. [Diane E. Lewis, Boston Globe. From the San Jose Mercury, 5/26.]

On the other hand, MBAs or company-trained equivalents are in great demand throughout the Asian world. Economic growth has been so explosive that companies are exceeding the supply of managerial talent. Many large companies wish to move from autocratic management styles to Western participatory styles, but there aren't enough university-trained managers to make it work. Western managers have been filling many of the top slots, but short tours of duty may affect their decisions. Asian companies would rather build local expertise, but must first revamp and expand the local business schools. [Fortune, 6/3/91.]

The Carnegie Foundation for the Advancement of Teaching recently published a careful comparison of professors with many publications vs. all others. Those in the top 45% (based on lifetime productivity of journal publications, which appears consistent with recent productivity) are more active in research conferences, association activities, and consulting. They get more federal support, and -- except in engineering -- get the lion's share of institutional and departmental research funding. Their salaries are higher, and outside income typically adds 20% or more to academic salary. Their work week takes as many hours as less-published professors, but they teach fewer courses, spend fewer hours in preparation (although the same amount per course taught), have less contact with students, and spend less time in campus governance. They rate their departments more highly on academic reputation and intellectual environment, and are less likely to believe in teaching as a promotion criterion or to agree that research reduces teaching quality at a university. They are also more satisfied with their career choices. (And no wonder! But the next question is how to get the reputation or the funding that starts the cycle.) [Change, Mar/Apr '91.]

Myron Kayton (Santa Monica) argues that senior engineers are charging far too little for industry short courses. A five-day, leading-edge course requires about 450 hours the first time it is given (20 hours of administrative liaison plus 30 class hours times 10-15 hours preparation and 1 hour of post-class time per class hour). A senior lawyer would charge $350 per hour, or even $700 in the Los Angeles area. Engineers teaching at Southern California schools typically get $2500 to $8000 for a one-week class or a 15-week night class of 25 students. The difference of $150,000 or so goes mostly to industry, which buys "at least a man-year of knowledge" for about $1000. Kayton would like to see tuition raised to $20,000 per student, which would still be a bargain if each man-year of knowledge saves the company more than $150,000 in salary, overhead, and productivity. [IEEE Spectrum, 6/91.]

(Kayton is, of course, ignoring market forces. He is free to offer his services at $350 per hour, $150,000 per week, or $500,000 per course. He might even get it -- if he works exclusively for one company and gives them a competitive advantage. For those who teach ordinary short courses, note that there is little payoff in a course given once. Professional seminar leaders look for topics that can be sold in different cities every couple of months.)

"My business is growing, and I see so many opportunities I wish I were 1,000 people so I could pursue them all." Roger McConochie is one of about 200 trained anthropologists working in corporate America. (He's a business consultant in St. Charles, Illinois.) Compensation levels are about twice those of academic positions, which start at $27K to $32K for non-tenured assistant professors with doctorates and increase to $50-53K for tenured full professors. [Carol Kleiman, Chicago Tribune. San Jose Mercury, 5/12.]

A survey of job-hunting executives (by New York recruiting firm Canny, Bowen Inc.) showed that 45% lost their jobs to downsizing or acquistion; 51% had disagreed with corporate strategy, promotions, or a change in responsibility. Most were in their mid-40s, with compensation ranging from $150K to $350K. [Compensation includes perks, bonuses, retirement benefits, etc. It's how executives keep score.] Severance pay was often six months to a full year. Neither pay nor relocation was a significant complaint, but office politics was sometimes mentioned. The executives were looking for more than just a job: 41% had already turned down offers. Most had done little planning for career changes. Some wished they had obtained advanced degrees, others wished they had done more networking. [New York Times. San Jose Mercury, 5/12.]

I've heard that executives and middle managers climb toward the top until they're 40; after that, they just try to hold onto what they've got. As the baby boomers age, you can expect fewer to willingly give up their hard-won positions. The Silicon Valley turnover rate in 1980 was 30% for "exempt" employees; now it's about 18%. [Radford Associates/A&ACG.] James Mitchell (San Jose Mercury) attributes this to the work of personnel directors in defining and humanizing their corporate cultures. I find it hard to believe his statement that "high-tech companies have grown much faster than their employee pools." The truth is that qualified people are finding it hard to locate new positions. An 18% turnover rate means that it takes an average of 5 year for any specific position to open up -- and then you have to compete with those in the pyramid who haven't had a promotion in five years.

Edward Dunbar, a UCLA research psychologist, says that many minority and foreign-born engineers in Southern California feel excluded from challenging, high-visibility project teams. They also feel socially isolated, and doubt that they are being considered for management positions. Some lose their ambition for advancement, others lose their commitment to the company. [Michael Wolff, IEEE Spectrum, 6/91.]

David Morgenstern (BMUG) reports in MicroTimes that Apple was pushing its SQL-based Data Access Language at the recent National Database Exposition. (Some two dozen vendors displayed DAL-based front ends to mainframe databases. ACIUS was also there, offering a 4D DAL toolkit for its 4th Dimension database.) The Apple booth was full of colorful software and people, a contrast with the suited-and-tied salesmen and yogurt-serving ladies at the rest of the booths. MIS customers were somewhat reticent to stop at the Apple booth and talk with the men wearing large earrings.

Will Tracz [Computer, 4/91] says that "a prominent psychologist" has found a predominant type among Silicon Valley programmers and engineers: perfectionists seeking control in their lives. In childhood, they either grew up in chaos or were overprotected and never allowed autonomy.

Patti Price, an SRI speech-recognition researcher, has observed differing workstyles among her team. The engineers are interested in quick fixes and rapid prototyping -- ideas that can be tested within a week or two. The Ph.D.-level computer scientists, however, are motivated by long-term projects that will produce papers.

Marylene Delbourg-Delphis (Chairman of Acius) says that Americans are "totally individualistic, but insecure. The more they speak of being part of 'the team,' the more they aren't a team." European managers -- she's French -- expect debate, but Americans want the boss to make a decision so that they can "buy in." "Being part of a team means you're making sure your neighbor isn't stabbing you in the back." [Cris Oppenheimer-Pitthan, San Jose Mercury, 6/2.]

"Everyone looks at me as a leader and all I want to do is run." CEOs of American businesses often suffer from insecurity, afraid that their success is a fluke and can't be sustained. They live in dread of exposure, and can't stop working long enough to enjoy their success. Many are depressed, although they may or may not realize it. Rich Chollet, the founder of Brookstone, recently took his own life. Helping executives reconstruct their personal lives is a new specialty in psychiatry. [Newsweek, 6/3/91.]

Companies with telecommuting pilot programs have exceeded their expectations. Workers are happier, with lower absenteeism, lower turnover, and often greater productivity. Handicapped employees may benefit the most, but workers enjoy the flexibility, extra [hassle-free] non-commute time, and closer family ties. Employers save on recruiting, overhead, and parking space. Pacific Bell's 2,000 telecommuters reported greater job satisfaction (71%) and improved performance (87%). Absenteeism fell 57%, and a majority of supervisors found no difference in managing telecommuters. There can be a loss of teamwork and team spirit, though, so regular office visits are advised. [Andy Shapiro, Computer Currents, 5/21.]

Doug Engelbart's Bootstrap Institute is giving a $900, 3-day seminar, June 18-20, at Stanford (where Doug is also an associate at the Center for Design Research). I had difficulty wading through the hype phrases, but the gist is that knowledge workers should have tools to support distributed group activities. You can hear Doug's thoughts (and get copies of his notes) on enterprise integration, total quality, concurrent engineering, groupware/CSCW, hypermedia, CASE, online document delivery, integrated CAD/CAM/CALS architectures, open systems, interoperability, program/issue/records management, continuous process improvement, cross-enterprise and multicorporate collaboration, etc.

Doug claims that design complexity and urgency are increasing exponentially, and that organizations failing to invest in groupware will soon be unable to cope. He sees online collaborative knowledge development as a way that heterogeneous project teams can coordinate their work to identify needs, find solutions, and incorporate learning. He proposes an open hyperdocument system (OHS) to support collaborative knowledge development, and will demo the AUGMENT prototype. Such a system would support a project team's knowledge base: continuously updated plans, specs, documentation, decision trails, status reports, etc. The seminar will include design considerations for sharing of hypermedia files and screens among widely distributed knowledge workers. For more information, contact info or uunet!! [Reported to IRList by William Daul, infmx!]

For a cautionary view, see Captain Stephen B. Sloane's report, "The Use of Artificial Intelligence by the United States Navy: Case Study of a Failure," in the Spring '91 issue of AI Magazine. The captain of a new aircraft carrier tried to implement a knowledge system that would record all operating procedures aboard the carrier. His goal was to capture operational knowledge via incremental refinement, with each new crewman able to learn from the knowledge base and to continue where his predecessor had left off. The system also had AI components to help with real-time scheduling and tactical decisions. Sloane's conclusion is that the knowledge system failed because crew members had no interest in recording their expertise (and no time to do it); the AI component failed because it had insufficient perception of the true constraints and variables in tactical situations. Overall, the project failed because a centralized control system did not match the Navy's use of distributed responsibility and authority.

Ron Gershon ( would like to know if there are corporate relationships between the labs that I mentioned in V1 N9. [Is it Mitsubishi or Matsushita that owns the Panasonic name? Is there really going to be a Matsushita lab in Cambridge? Is there any tie between the Panasonic lab and Lee Giles' NEC lab in New Jersey? What is the funding source now for the International Computer Sciences Institute (ICSI) in Berkeley? And who has signed on at the new Canon lab in Palo Alto?] Also, how do the other Computists feel about these labs snatching top researchers in the U.S.? Is Japan just tapping the world's best basic research, or is this a political move to address balance-of- trade criticisms and hedge against technology export laws?

If the U.S. labs worry you, you might consider whether you prefer the new Asian science parks. IEEE Spectrum (6/91) lists seven in Japan (Tsukuba Science City, 21st Century Plaza, Eniwa Business Research Park, Nagaoka Research Park, Toyama Center, Kanagawa Science Park, and Kurume Techno Research Park). Japan also has the Fuzzy Logic Systems Institute in Iizuka, near the Kyushu Institute of Technology and its 200 computer science faculty. MITI is planning some 40 other semi-rural technopolis centers for engineers and computer scientists.

Then there's Daedok Science Town in South Korea, Singapore Science Park in (you guessed it!) Singapore, and Hsinchu Science- Based Industrial Park in Taiwan. (The Hsinchu park may soon spin off a Software Industrial Park. A core feature of Taiwan's Science City in Hsinchu will be optical-fiber links to a supercomputing center.) Hong Kong is also building a small science center and has plans for a science park. Two new parks are opening in the Philippines, and Shanghai is planning a huge science, technology, and trade zone.

India is particularly focused on information technologies, including telecommunications, multimedia communication, and large, distributed databases. India has an "electronics city" in Bangalore, where 43 engineering colleges turn out 11,000 engineers each year. The Indian Institute of Science there is a major center for basic research. Many foreign high-tech companies are building labs, including Texas Instruments software research center for semiconductor chip design.

Many factors are at work here. One is that a growing economy can afford research, thus ensuring future success. (Have you ever played with the old Club of Rome simulation of world technology? Crank research spending as high as possible if you want to avoid global economic and environmental collapse.) Another is the relatively low cost of foreign research labs. Larry Press reports [CACM, 2/91] that a university-trained C/UNIX programmer in India costs $12,000 per year, with typical programmers only $2,400. (The industry has grown 59% per year since 1985, to $10- 15M in exports, but now faces a shortage of programmers and lack of experience with big projects and turnkey or packaged software.) Corporate-paid training (university, in-house, or via job rotation) is also feasible since lifetime employment policies mean that knowledge assets will not be lost. If you were building a large software lab, where would you put it?