|Volume 1: No. 06|
Charles L. Wilson (NIST Advanced Systems Division; (301) 975-2080) is soliciting input for a report on new information processing technologies (NIPT). This is in response to a NIPT initiative within Japan's Sixth-Generation Computer project. The NIST report will help form U.S. policies regarding massively parallel computing, distributed processing, neural computing, optical computing, and fuzzy logic, including identification of emerging technologies , assessment of economic impact, assessment of our position relative to Japan, and consideration of national security. Email responses to firstname.lastname@example.org are desired by May 15. [Connectionists.]
The Initiative for Managing Knowledge Assets (IMKA) has announced availability of its object-oriented representation technology -- for $5,000 to $25,000 per license. IMKA is a consortium of Ford, Carnegie Group, US West, TI, and DEC. [Lance Eliot, AI Expert, 5/91.]
Philips, Matsushita, and Sony have announced that over 180 Japanese companies have joined to develop CD-I technologies. Similar consortia are to be formed in Europe and the U.S. [Spectrum Newslog, 5/91.]
The Open Document Architecture Consortium (Brussels) will develop international standards for multimedia documents. Members are IBM, DEC, Unisys, Bull, Siemens, and ICL. [Spectrum, 5/91.]
Computer (3/91, pp. 81-4) has an excellent report on the growth of electronic mail networks in Eastern Europe and, very slowly, in the Soviet Union. Bitnet and UUCP/Fidonet links are the current standards. [Snyder, Jarmoszko, and Goodman.]
IBM and AT&T are working on bridge software that would unite their telecom and data networks. [Spectrum, 5/91.]
Nathaniel Borenstein reports that multimedia email will likely depend upon "bottom-up" solutions instead of a universal format. Email systems will invoke site-specific modules to display embedded images, or will report to the user that no display capability is available. [CACM, 4/91.]
International Data Corp. predicts explosive growth in the European software market -- for software conforming to EEC-sanctioned standards. The market is larger than that in the U.S., so it may influence U.S. software development practices. The Open Systems Interconnect model is becoming particularly popular. [Edelstein, Fuji, Guerdat, and Sullo; Computer, 3/91.] (See Computer (4/91) for more on the proposed European standards.)
Advanced Computing Environment (ACE) will be an improved PC operating system incorporating OS/2 and Unix running on Intel and MIPS chips. Leaders of a 21-company consortium are Compaq, DEC, Microsoft, MIPS, and The Santa Cruz Operation. [Spectrum, 5/91.]
MCC's Object Manager software permits a heterogeneous network of computers to function as a parallel computer. The system will be further developed in an object-oriented simulation study with Sandia National Labs and the Institute for Advanced Technology at UTexas Austin. [Computer, 4/91.]
If you're touting object-oriented technology, you may want to buy the Object Management Architecture Guide. It's a new "standards" report by the Object Management Group (OMG) -- a consortium of over 90 companies. Contact OMG, 492 Old Connecticut Path, Framingham, MA 01701. [Lance Eliot, AI Expert, 5/91.]
Another hot buzzword these days is case-based reasoning (CBR). It's sort of a nearest-neighbor classification system for object-structured data -- as opposed to statistical or neural classification of vectors -- with arbitrary amounts of expert- system code, logical reasoning, or constraint-based reasoning thrown in. (Many of the proponents are from the planning or legal-reasoning communities.) A claimed advantage over expert-system rules is that addition of new cases requires less restructuring of the entire system. Inference Corp. has been selling CBR tools in its ART-IM shell and now in a less-technical product called CBR Express. Cognitive Systems has recently won Phase II SBIR support for evaluation of CBR, continuing work from a 1987 DARPA contract. (DARPA has been the main supporter of CBR.) Cognitive is beta-testing its CBR-based expert system shell at more than a dozen sites, and will have at least 10 completed applications when it releases a commercial version -- including content-based financial telex routing. [Harvey Newquist III and Lance Eliot, AI Expert, 5/91.]
NSF will award up to $10.6M to Paul Lauterbur for a new Science and Technology Center at UIllinois, for development of magnetic-resonance technology (MRI) to map the human brain. (Hmmm, there could be some interesting data analysis and display problems there.) [The Institute, May/June '91.]
Carnegie Group will receive $225,000 from NSF for continued integration of neural-network signal recognition with knowledge- based diagnostic and repair-planning technologies for troubleshooting mechanical systems. [Computer, 3/91.]
UPennsylvania is advertising a new MSE program, Intelligent Sensor Technologies (IST). "An integrated approach to information acquisition, information processing, and information cognition." Contact Dr. Kenneth Laker, (215) 898-5340.
Toshiba has reported good success at handwritten character recognition with a neural network trained on 230,000 samples of digits and katakana. The network is invoked for characters that cannot be reliably identified with traditional statistical classification. Contextual analysis will also be added before commercialization within the next five years. [Computer, 3/91.]
If your work involves scheduling or route planning, you should be aware that an exact solution has been found for some types of traveling salesman problems. See the February issue of Science. [CACM, 4/91.]
Jeff Pepper (formerly with Carnegie Group) has launched Serviceware, a new company to serve the service industry. [Intelligent Systems Report.]
Lumigenic Media is the new name of Dolphin Design Associates (Felton, CA). The company has pioneered multimedia applications on personal computers. [San Jose Mercury.]
The April issue of CACM has a good article on the JPEG image- compression standard. Gregory Wallace mentions that floppy disks containing original, compressed, and reconstructed data are available from Eric Hamilton (C-Cube Microsystems, San Jose). If digital video interests you, check out the article by Michael Liebhold and Eric Hoffert. They summarize the limitations of video, then suggest what may be available in future multimedia telecommunication systems.
The International Telecommunications Union [ITU] is implementing email, conferencing, database access, and other services for member companies. One of their databases contains 30,000 telecommunications terms in English, French, and Spanish. [Computer, 4/91.]
A new CD-ROM tutorial package for spoken Japanese is getting excellent reviews. It's NihongoWare 1 from Ariadne Language Link, distributed by Marubeni America, (212) 599-3750. Ten animated lessons and drills (requiring 80 to 100 hours) cover travel conversations and cultural material. Mouse clicks open additional helps such as phrase translations, phonetic notation, and slowed pronunciation. Reiko Etoh, an executive secretary at Fuji Xerox, developed this as an intrepreneurial project. A Japanese firm did the HyperCard coding using material from Waseda University. $623, for the Macintosh. [James Mitchell, San Jose Mercury.] (If you've got a good idea, you, too, could raise capital and hire others to develop it! Only the investors will get rich, though.)
DAK (Canoga Park, (800) 325-0800) is offering CD-ROM bonuses in order to sell PC-based CD ROM players. For $499 (or $699 external), you can get a BSR player and six discs containing the 33,000-entry Academic American Encyclopedia; Webster's New World dictionary, thesaurus, and quotations; NY Public Library Desk Reference; Dictionary of 20th Century History; Concise Guide to Writing; 1990 National Directory of Addresses and Phone Numbers; 320-map world and U.S. atlases (with economic/geographic databases); 17 bilingual/multilingual dictionaries; and the full text of 450 literary works. For an additional $149, you can get five more discs with 9 versions of the Bible; 20 Bible dictionaries and commentaries; 107 U.S. history books; 220 small-business publications; Time Magazine's Compact Almanac with 5,000 historical articles; and 110,000 articles and abstracts from 1990 issues of 342 other magazines.
If you teach a course in human-computer interfaces (HCI), scan the inexpensive resources that Larry Press lists at the end of his April 1991 CACM column on personal computing. These include an interesting videotape (Brad Meyer's "All the Widgets"), state-of- the-art collaboratory research systems, and several book recommendations.
Computer scientists aren't noted for running rigorous statistical experiments. For those who need stochastic modeling, though, the March 1991 issue of Computer has a review of several leading programs for designing experiments and analyzing data.
Need to print some advertising T-shirts for your local group? For $89.99, you can get a 350-print cartridge of black or colored transfer toner for your laser printer. Contact Blacklightning Inc., (800) 252-2599, for samples, supplies, or transfer services. With their professional equipment, you can even print on ceramics, metal, wood, cardboard, etc. [Craig Crossman, Knight-Ridder.]
E. Gordon Bell has a new book out, "High Tech Ventures: The Guide for Entrepreneurial Success." Bell is currently chief scientist at Stardent Computer.
If you'd like a good reference book on applied database design, check out Leszek Maciaszek's "Database Design and Implementation," 1990. Pradip Srimani reviews it very favorably, except for its lack of problem sets. [Computer, 3/91.]
Another good reference book is "Computer Graphics Handbook: Geometry and Mathematics," by Michael Mortenson, 1990. It's reviewed favorably as a "quick reference guide" by Leon Tabak in the April issue of Computer.
Johns Hopkins University is offering a $10,000 prize in its NSF-supported search for aids for the handicapped. Write to Personal Computing to Assist Persons with Disabilities, P.O. Box 1200, Laurel, MD 20723. [Computer, 4/91.] (This is obviously a worthy cause, but I have to question whether developing aids for the disabled is a good career move. It's a rather small and cash-starved market. If you're after grants, though, this is an area that NSF loves to fund.)
The tax break for donating old technical journals can be considerable. Call Lou Scheffer at (408) 944-8675 to see if your back issues are needed by the six Black universities in South Africa. (Five-year spans, 90% complete, are desired.) [IEEE Grid, 5/91.]
The essence of most new applications is automation of a previously manual or computer-assisted data-analysis task. William Sands (UUtah) has had success with a Prolog-based expert system to monitor athlete's training responses. The system detects symptoms of overtraining, undertraining, diseases, growth spurts, and even psychological problems. If you were looking for a job developing an interesting new application, would you think to contact a coach? [Justin Kestelyn, AI Expert, 4/91.]
Another expert system is used by Alamo Rent-A-Car to set prices for every car, in every city, every day. Linear programming wins out over expert systems in many applications, but the models are difficult to build and maintain. For any such application now in use, there may be a market for expert systems to replace the linear model or to build a user-friendly interface to the existing package. [AI Expert, 4/91.]
Newsweek (5/6/91) mentions a computer program from Specialized Data Systems (Jenkinstown, PA) that generates form letters with variations in the phrasing -- thus concealing orchestrated letter- writing campaigns. Will NL scientists contribute to an identity- concealment measure-countermeasure war? Is there a market for translating text into assorted jargons and educational levels, with parametrized emotional content?
Another Newsweek article mentions that public acceptance of lotteries may soon revive horse racing as well -- a sport where a serious, restrained handicapper can consistently come out ahead. The future appears to favor "fewer, smaller tracks, elegantly appointed, televising live races to other tracks and betting parlors everywhere." Such flows of information and cash are bound to create opportunities for databases and handicapping programs. (Several such programs are already on the market.)
Karl Bergerson (Neural Trading Co., Seattle) claims spectacular performance for his neural-based trading system, Neural$. A fictional training account increased 660% in two years, and the system was 89% accurate on new data. He claims that the key element is his carefully chosen training data. (One can only wonder whether the same data would yield equally powerful statistical methods.) [Justin Kestelyn, AI Expert, 5/91.]
Larry Press, in his 4/91 CACM column, mentions some significant advantages of personal computers as research tools. Since large and small schools alike use the same hardware, research progress can be distributed as software; results are thus easily demonstrated or replicated. Students can take their work with them after graduation. And conversion to commercial products is relatively painless. (Fred Brooks, though, once estimated that the cost to finish a programming system is nine times the cost of the prototype.)
I used to develop computer vision algorithms, as did most of the people I knew. Some are still at it, mapping ever-simpler algorithms onto ever-more-parallel hardware. They dream of the day that their customers will buy them pipeline machines, teraflop CPUs, or custom-VLSI neural simulators.
My own enthusiasm faded when I realized what a limited market I was serving, and how unlikely it was that I would ever buy a Connection Machine and go into business for myself. I bought a Mac instead, and have spent the last three years learning to use it -- and I'm still nowhere near exploiting its full potential. I have no regrets whatever, and I hope that Larry's words will lead others to the same career choice.
I will offer one word of warning, though. Common hardware and shared software make it easy to build your research on top of code written by others. For academic research, this is a Good Thing -- provided that significant findings are checked with independently written code. For commercial developers, this is also a Good Thing -- provided that licenses are available for all incorporated code. The potential for trouble arises if you try to move academic software into the commercial world. You may find yourself unable to license critical components. I find this unlikely, but Richard Stallman, Will Tracz, and others concerned about intellectual property rights have lately been raising the alarm. Keep it in mind.
Robert A. Rivers points out an interesting phenomenon in The IEEE Grid (5/99). It's fairly obvious that managers and industry leaders want an oversupply of engineers so that they can pick, cheaply, from the top 10%. (Well, you don't want the bottom 10% designing reactors, do you?) NSF's predicted shortages could also be self-serving, since its education and "pipeline" budgets depend upon the perceived need. (Then again, NSF could be right. And should we stop encouraging women and minorities?) Rivers points out that a third biased group, the engineering educators, are strongly represented in the engineering societies. Educators naturally push for increased enrollments. Thus the professional societies themselves tend to work against the stable employment, high salaries, high status, and creative work that would accompany a moderate shortage of engineers.
Robert W. Lucky mentions [Spectrum, 5/91] that health care entails a relatively constant amount of work, but that the need for legal help increases in proportion to the number of lawyers. It's not clear whether engineering is closer to medicine or to the law. (Probably the latter, given a competitive marketplace.)
My own view on the U.S. "shortage" is that it's all relative to our competition. If other countries are no more successful than we in training engineers, it's a level field. No great harm is done if society has to wait an extra few years for the next advance in electronics. If Japan and Germany train more or better engineers than we do, though, design and manufacturing will move to those countries. Even that is not terrible if we see it coming and invest accordingly. If we don't recognize that a competition exists, and that we are losing, we will deserve our economic woes.
Even if we do have a problem in the U.S., I suspect that it has little to do with the number of graduating engineers. Rather, it is with a system that throws away experienced engineers in favor of cheap, inexperienced ones. I'm not going to tell corporations how to run their businesses, but there should be alternative employment option where the experienced engineers could show their stuff and win revenge on their short-sighted former employers.
Consulting or starting one's own design business is such an option. Unfortunately, most engineering projects require physical resources far beyond the means of startup companies. One solution would be state sponsorship of such resources, with access awarded by need and merit, much as we now give access to telescopes, supercomputers, cyclotrons, MOSIS VLSI fabrication, and even government grant money. The Stanford CIS project to develop remote mechanical fabrication facilities is in this tradition.
The other missing link is in sharing of human expertise and effort. Modern engineering usually requires team effort, with additional teams of accountants and support personnel. Wouldn't it be great if engineers around the world could contribute expertise incrementally to an evolving design, each receiving compensation in proportion to contribution? Sort of a Xanadu project for engineers. I have no idea how compensation would be determined, but it's something to dream about.