NSF is advertising for a new Director of Computer and Computation Research, the position held by Richard A. DeMillo (with Bruce Barnes as Deputy). Salary will be in the range $87,000 to $100,500, for a career position or 2-3 year appointment. Or you can join as an IPA, retaining your current salary and benefits. (This often works out better for short-term people. Civil service pension benefits are unvested until after two years.) Contact Deborah Munn, (202) 357-7601, by 6/28/91. [CACM, 6/91.]

Integrated Systems Inc. (Santa Clara, CA) is looking for M.S. research scientists in advanced real-time control (modeling, simulation, system identification, control design, and integration). Applications are in manufacturing automation.

Molecular Design Ltd. (San Leandro, CA) is looking for a scientific programmer and application scientists to develop and market distributed relational database software. (That's right: application scientists to "support the marketing of various products through developing and presenting demonstrations, seminars, and workshops.")

Metaphor (Mountain View, CA) is looking for "architects" for multimedia, object-oriented systems and storage, operating systems, distributed systems, applications, and knowledge-based systems.

Unisys Defense Systems (Sunnyvale, CA) needs programmer analysts for satellite-related database applications, human interface, scientific applications, and control. Many of the positions involve simulation and diagnosis.

Loral Western Development Labs (San Jose, CA) is also advertising for satellite-related C3I people.

The Dept. of Linguistics and Computer Laboratory at UCambridge are looking for two research assistants -- an experimental psycholinguist and a computational linguist -- for a 3-year project in post-lexical and prosodic phonological processing. Contact Paul Warren (pw25@phx.cam.ac.uk) or Ted Briscoe (ejb@cl.cam.ac.uk) by June 15. [NL-KR.]

UAlabama at Birmingham, Dept. of Computer and Information Sciences, needs a GRA to study neural networks for EKG analysis. (The department specializes in high-performance computing, graphics and image processing, and biomedical applications.) Compensation is $7,600 and full tuition. Contact Warren Jones (jones@cis.uab.edu). [Neuron Digest.]

The newly created Institut Dalle Molle d'Intelligence Artificielle Perceptive (IDIAP) in Martigny, Switzerland, seeks researchers in speech recognition and image manipulation. Laboratories are being established and international network connections will soon be in place. IDIAP is the third institute of artificial intelligence supported by the Dalle Molle Foundation, the others being ISSCO (UGeneva) and IDSIA (Lugano). The new institute will maintain close contact with these centers as well as with the Polytechnical School of Lausanne and the University of Geneva. Contact Daniel Osherson (osherson @disuns2.epfl.ch), Directeur, IDIAP, Case Postale 609, CH-1920 Martigny, Switzerland. [Connectionists.]

Tim Finin has asked me to mention the 8th IEEE Conference on AI for Applications next March 2-6 in Monterey, California. 5,000-word papers on applications and enabling technologies are solicited; contact Tim for details. Proposals for panel discussions, tutorials, and workshops are also sought. The best submitted papers will be considered for a special issue of IEEE EXPERT to appear late in 1991. (A special issue of IEEE Transactions on Knowledge and Data Engineering may also be available.) IBM will sponsor an award of $1,500 for the best student paper.

The closing date for joining IJCAI-91 workshops is June 15. (Each workshop is AUD$85.00.) Contact me if you need the email address or phone number for any workshop organizer. Topics are:

Parallel Processing for AI [Laveen Kanal] Decision Making in the [NL] Generation Process [Marie Meteer] Software Engineering for Knowledge-based Systems [Bob Jansen] Modeling for Intelligent Interaction [Judy Kay] AI and Business [Daniel O'Leary] Integrating AI and Databases [John Zelezikow] Dynamic Scene Understanding [Tom Huang] Changing Representation in Machine Learning [Katharina Morik] Full Implemented NL Understanding Systems [H.J. Novak] Evolution and Chaos in Cognitive Processing [Dwight Deugo] Fuzzy Control Workshop [Peter Eklund] Fuzzy Logic in AI [Anca Ralescu] Intelligent/Cooperative Information Systems [Mike Papazoglou] Natural Language Learning [David M.W. Powers] Knowledge in Medical Decision Support [Cynthia Sarmiento] Explanation Generation for KB Systems [Nel Wognum] AI in Design [John Gero] AI Approaches to Production Planning [Steve Garlick] Objects and AI [Mamdouh H. Ibrahim] Computational Approaches to Non-Literal Language [Dan Fass] Design of Rational Agents [Mike Georgeff] Reasoning in Adversarial Domains [Christopher Elsaesser] Situation Awareness [Simon Goss] Computer Vision [Ray Jarvis] Interfacing Production Systems/Real World [Franz Barachine] Verification and Validation of AI Systems [Marc Ayel]

Ron Gershon (gershon@vis.toronto.edu) tells me that Matsushita owns Panasonic Technologies (a U.S. company), and that Richard Lipton is the contact person in both sets of ads. It seems that there is only one lab, ITL in Princeton, NJ.

Mark Weiser (mark_weiser.parc@xerox.com) forwarded a message from Hector Garcia-Molina. Hector, Rafael Alonso, Kai Li, and Richard Lipton have taken leave from Princeton University to start the computer-science research lab. It will eventually have about 40 people, half with doctorates. The university and CS department are supportive, and ITL hopes to initiate joint projects. Early plans were to have one initial project in hardware architecture, another in databases and distributed computing. Open publication is planned.

Mark sees a strong parallel with the Xerox PARC labs. Lab managers ("area managers") are autonomous and can shift direction radically, spend money, get grants, hire, etc., without prior approval. A lab is 30-70 people (Mark's is about 50). The new Japanese-funded labs are all smaller, around 20-30 final size, so perhaps autonomy and lack of defined projects is not unusual.

The Japanese have always mined information and technology whenever they could, and they have paid for research by U.S. consulting firms such as SRI. (Often they have the same research done at multiple sites, here and in Japan, to increase the chance of success.) They've now discovered that top scientists are cheaper as employees than as consultants, and have found a way (other than enrolling students) to link with our universities. If they get some patents from this, it is their legitimate due for supporting U.S. researchers.

If there is a problem, it seems to be entirely with the failure of U.S. industry to support similar labs. Our government can't force U.S. industry to be competitive, and industry sees research as a cost rather than a profit center. Why should they support basic research if they can get a better return from investments, then license basic technology from the Japanese?

I'm afraid that much of our problem is optimism. Each company believes that its industry will survive (through protective legislation if necessary), and each believes that it can do as well as its U.S. competitors. Most of our high-tech companies are not aggressive in foreign markets -- partly because of U.S. bans on technology export -- and most are not sufficiently wary of foreign companies taking over the U.S. market.

There is also a personal optimism. Executives from finance and marketing (or from MBA programs) are trained to run any company. If one high-tech company were to fail, or even an industry, the executives would find work elsewhere. There is little personal incentive to ensure the long-term future of a company, whereas there is great pressure to build a good resume (and personal fortune) through short-term maximization.

The good news is that small business is alive and well. Most entrepreneurs have a long-term outlook, identifying strongly with their businesses. The bad news is that entrepreneurs seldom have the capital to hire scientists or to compete in high-tech markets. Perhaps the lesson is that we must work for the Japanese or work for ourselves.

I've been reading Inside Track, a 1986 book by Ross and Kathryn Petras. They interviewed over 400 people in 17 industries, compiling dossiers on almost 60 corporations and bureaucracies. Two were government agencies, clearly a distinct environment. The rest were mature corporations in mature industries, except for a few computer companies that I discuss below. I don't consider software development a mature industries, and I doubt that salaried software developers are prepared for the changes that are coming.

The key point is that traditional, well-established career paths exist in most leading corporations. You graduate from a prestigious university, perhaps take a summer internship in the industry, then (with luck) join for several years of training and low-paid grunt work like typing, filing, fact-checking, and arranging client meetings. If you're good, you move up to several years of slightly better-paid work with a bit of intellectual content. If you don't love it, you drop out -- possibly going to a less-prestigious firm in the same industry. The best of those remaining compete for the top jobs. Winners get prestige, job security, reasonably good pay, and satisfying work -- but the hours are long and the pace remains hectic.

In book publishing, for instance, one path is from editorial assistant to assistant/associate editor to editor. Even an editor makes only $30K to $60K, but it's a highly sought position. Advertising and legal firms give their best people partnerships, but the financial rewards may be balanced by continued hard work, politics, and insecurity.

Everyone (except an independent owner) answers to someone, and every job function is competitive. Pay is just high enough to motive the people at the next lower rank. Every career path includes long hours, hectic work, and intense competition with peers for the few slots that open up as people retire. Supporting staff positions are less competitive than the prestigious career paths, but they have their own cultures and constraints.

You could not be managing editor of a major journal without first being an editor there -- or managing editor at another major journal. Managing editor is not an entry-level position, and it's almost impossibly stressful even for someone with a lifetime of experience. Likewise, you could not be an editor without first having been an assistant editor or a writer. It works down the chain, with very little chance that anyone is going to enter the pyramid above the apprentice level -- except by having followed the same chain at another corporation in the same industry.

Entry-level jobs had 1986 salaries of $15K to $25K, sometimes even $9K or $35K. The better-paying jobs require advanced degrees or the ability to accept greater responsibility quickly. Competition is strong for all jobs, or they wouldn't be able to get a New York City-based editorial assistant for just $9K. A professor couldn't expect to start as a book editor at Harper & Row, Random House, or Simon & Schuster; he or she probably couldn't get in even as an editorial assistant. Only young/ inexpensive, talented, hard-working people with a dedication to the field and a history of related activities (e.g., college newspaper work) have a chance at the career path.

High-tech companies hire entry-level people for product design and development, R&D, production and manufacturing, and finance, marketing, and sales. Research, design, and development people generally start with engineering and CS degrees. Software people enter as applications or systems programmers. They work in loosely structured teams with flexible hours, at 1986 pay ranging from $25K for college graduates to $35K for advanced degrees.

Non-software career paths are also an interesting part of the industry. Production and manufacturing people may start as production supervisors, then move up to production managers and plant managers. Salaries match those for design work. Responsibilities range from the technical to personnel management.

Marketing people start as assistant or full product managers at $28-$32K, working with the technical staff, sales department, and finance in developing and introducing a marketable product within allowable budget constraints. They may advance to product marketing managers, overseeing a line of products or supervise an entire division.

Sales staffers may be responsible for one product or a category of products. They typically start with a small territory or account at a pay level of about $25K plus commissions -- which, in mature industries, are often based on quotas. Commissions improve with experience, which also brings larger territories (or national accounts) and higher volumes.

Financial staffers start as financial assistants at about $28-$32K, but often have a fast track toward corporate management. Entry-level work includes analyzing budgets and doing cost projections.

Of the computer companies, IBM (Armonk, NY) is the most "corporate." It had about 400,000 employees in 1986 and was 6th in the Fortune 500. The company takes good care of its people, but demands in return that they work hard for IBM. Families are also well taken care of, and community activities are considered important. Promotion and pay are merit-based, with separate tracks for technical and management types. Promotion is almost exclusively from within, so outsiders can only join at the bottom. To get hired, you need interpersonal skills, technical knowledge, professional appearance, a good academic record, and an eagerness to do a little more than the next guy. To get ahead, you need initiative, motivation, dedication, and hard work. If you're IBM material, it's a pretty good life.

Hewlett-Packard (Palo Alto, CA) has about 82,000 employees. It used to be touted as the best place in America to work, and it's still near the top. HP is more into instrumentation than computers, but their 1000/3000/9000-series has done well. (1000 is for factory use, 3000 for business, and 9000 for engineering workstations.) The development work is done by small teams with closely involved managers. The environment is unstructured, with no one telling you how to get your job done -- but teamwork and consensus are highly valued. Jeans and open-neck shirts are common, and you don't have to work 9 to 5. Employees are young, honest, earnest, helpful, and almost relentlessly upbeat. Politics and secrecy -- even privacy -- just aren't the HP Way. In some ways its a very "Japanese" company. You advance through hard work and flexibility, not political savvy, with product managers doing everything from conceptualization through development, introduction, pricing, training, and customer contact. Employees are encouraged to move within the company to where they can make the greatest contribution. Salaries are low (considering the talent on board), but the surroundings are pleasant and the cultural benefits are good. Turnover rate is the lowest in the industry. Entry-level screening is very tough, checking for technical competence, leadership, achievement, and an ability to take risks.

Apple Computer (Cupertino, CA) had 5,400 employees at the time the Petras were interviewing; it's about 15,000 worldwide now. Apple had just gone through rough times (failure of the Lisa, the departure of Wozniak and Jobs), and it wasn't the fun company it had been. Even so, it was "working 90 hours a week and loving it." Most employees were hired out of college, with design and development staffers running around in Hawaiian shirts and jeans. [What's the difference between Apple and a Boy Scout troop? The Boy Scouts have adult supervision!] But now John Sculley and the executive staff wear suits. Employees are optimistic, ambitious, and entrepreneurial. They are self-starters who work hard, work smart, and can work long hours without supervision. (Training and teamwork are also part of the atmosphere, though.) Employees are expected to blue-sky with the team, take part of a project and perform, and keep one eye on ease of use and the other on marketability and mass sales. There is no place to hide and no one to blame. Pay is pretty good, and the perks are even better. People are hired on the basis of technical skill, past initiative, and personality. Many employees start out as summer hires. There is no set career path, and for many the path is spin-off to their own companies. (Much of Apple's productivity and success depend on independent developers, much like a 1930's movie studio or a Japanese firm's use of independent manufacturing shops.)

Microsoft (Bellevue, WA) is a much smaller, software-only company supporting a much larger base of fielded products and platforms -- both DOS-based and Macs. It buys technology when it can, develops it when it must, and is a master of guerrilla marketing in the U.S. and abroad. The company is more structured than it used to be, but program developers still work in a freewheeling, creative, relaxed, and very individualistic atmosphere. The pressure to perform is intense, but it comes from within. Seventy to 80-hour weeks are common. Technical employees start as programmers, moving up to "technical leads" and then managers. If you can't make it as a manager, you can drop back to technical lead. Pay is above average for the industry. To get in, you really need to know your technical curriculum and you need solid references. You may also have to write a program, and will be screened for creativity, ambition, ability to handle stress, and willingness to do whatever it takes to get a job done. A history of past successes is also helpful. Once hired, there is no formal training; you're just thrown in. In other ways, though, managers take good care of their charges. "It's almost like a college dorm."

Ashton-Tate (Culver City, CA) is a database software house that has prospered through acquisitions and the licensing of marketing rights. In 1986, at least, the company was characterized by informal dress, HP-style "management by walking around," and a party at least every quarter. Over half the employees were in marketing, and good many of the remainder were in finance. Pay was pretty good, with additional stock options, 401K plan, and bonuses. Most recruiting was through headhunters, and screening interviews could take several months. One senior executive looked for people with proven accomplishment who could run a division within eight years. The candidate had to know the right questions to ask [a characteristic of experienced or fast- track managers] and couldn't expect to work 40-hour weeks.

Each of these companies has a fairly clear career path for executive managers, or can at least swap managers with other computer-related firms. I don't see the same career paths for programmers and other software people in the microcomputer industry. Maybe I (and the Petras) just don't know enough about the industry, or maybe it's so new that no programmer has ever matured.

In other industries, programmers advance to systems analysts, then perhaps to MIS directors or program directors. In the software industry itself, though, outstanding programmers are recruited by marketing-oriented managers in the same way that computers are purchased elsewhere. You jump in at an entry-level salary, work 70-hour weeks until you burn out, and then what? Move to a smaller company, in all likelihood. Or become a systems consultant. Or find a "vertical market" job in the defense or manufacturing industries.

For programmers, there doesn't seem to be tenure, partnership, or other prestigious position with security and good pay. It's work until you drop, or up and out to another career. A few programmers like Bill Gates (of Microsoft) make it to management or ownership, a few probably get rich on software royalties. But on the whole, I don't see much of a future for today's programmers. Tomorrow's programmers will be younger, cheaper, and more energetic, and today's will be out of luck.

IBM's workforce has been cut by 47,000 (10%) since 1986, mainly through early retirement and the sale of its typewriter and printer businesses. 65,000 employees were transferred to sales and marketing positions. The company is losing market share, and will ask 10,000 workers to take one-week unpaid vacations this July. Another 10,000 workers will leave voluntarily in the coming year, at a cost of $2.3 billion. Not to worry, though: IBM earned $6 billion last year, the most profitable of all U.S. companies. [Time, 5/20.]

IBM Japan's pretax profits were down 21% last year, to $1.1 billion. It now trails Fujitsu and NEC, although it has 30% of the mainframe market. The company will transfer 400 employees (of 25,000) and cut recruitment of college graduates by 50% (to 500). [San Jose Mercury, 6/9.]

Apple Computer has decided to lay off 1,500 people, 10% of its workforce. It isn't that sales are off or that Apple is forced to downsize. (The company has about $1 billion in cash.) Apple just wants to run lean now that its products are competitive and profit margins are slimmer. Remaining employees will be expected to maintain productivity. Business Week says that the reduction will cost Apple $75M, but may save $200M per year.

The SF Chronicle reports that 750 of the layoffs will be immediate (400 from Apple USA), the rest delayed a few months so that people can seek reassignment within the company. Executives may be given a 10% pay cut. (John Sculley's salary last year was $2.2M, not counting stock options.) The company may close 20-30 of its 70 leased buildings in Cupertino, and cuts have been proposed in day care, fitness centers, off-site conferences, and company cars.

Industry analysts tend to approve of downsizing to run lean and mean, but published statistics show very little effect on average productivity (across all industries). Employees who remain have low morale and distrust management. [Robert Kriegel.]

Alvin Toffler, in his new book "Powershift," claims that more and more companies are bringing manufacturing operations back to the industrialized nations. High-tech, low-volume products require increasingly sophisticated factories, educated workers, high-quality components, and just-in-time delivery. Less developed nations just can't meet modern standards and schedules. (Toffler cites one warehouse in China that held 40% of the nation's steel during a recent critical shortage. The steel was unavailable because the warehouse makes deliveries only twice a year.) [Christian Science Monitor.]

Toffler's migration doesn't mean that jobs are coming back, though: factory automation has become a fact of life. Perhaps secondary or tertiary suppliers will use hand labor, but such jobs may well be done in the less developed nations. Software production has not yet been automated, but commercial or proprietary software libraries and object-oriented development might have that effect. If salaried programming ceases to be an important industry, what happens to computer science departments?

"Employees of electronics companies ought to save more and prepare themselves psychologically for interruptions in their careers. In the 1990s, they're likely to be hard to avoid." [James J. Mitchell, San Jose Mercury, 5/26.]