|Volume 3: No. 16|
The Japanese Science and Technology Agency has conducted a delphi-style survey of 3,000 technologists. Predictions, released in 11/92, include the following R&D goals: 1GB commercial ICs (2002); improved air traffic control systems, telecommuting for clerical workers, and commercialization of health-care robots (2003); 10-TFLOPS computers [high priority], commercialization of micromachines (2004); multi-day warning of volcanic explosions (2006); development of a translating telephone, commercialization of 3D home-entertainment systems [without special glasses], diffusion of small-scale earthquake forecasting (2008); multi-day warning of major earthquakes, modeling of human memory and learning, "intelligent material" incorporating sensors, programming, and effectors (2010); completion of a manned observatory on the moon (2015). [Dave Farber. Bruce Shriver (email@example.com), firstname.lastname@example.org, 4/15/93. Tim Finin.] Bruce is soliciting other predictions.
Arthur D. Little consultants have told the Boston Airport Authority that airport traffic may be 20% less than expected by 2010 due to videoconferencing. [Bill Park, 4/10/93.]
A recent editorial in Barrons said, in effect, that NREN, NII, etc., do not matter. When the public and the technology are ready, government programs will be left in the dust. The US government will not prohibit commercial networks (as it did private postal services), although it may tax them and subsidize alternatives. Government-run services are likely to be sluggish, inept, and incompetent. [Rob Horn (email@example.com), com-priv, 4/16/93.]
Lettre de l'IA (10/92) estimated the neural-fuzzy market at $300M in 1991 and $580M in 1992. A 1992 Spang-Robinson report estimated the market for KBS tools at $270M in 1991, expected to reach $370M in 1992. OVUM estimates the total European KBS market at 500MEcus ($600M) for 1992 and 1500MEcus ($1.8B) for 1995, providing that the technology continues to improve. Intelligent Software Strategies (3/93) is much more pessimistic about market growth. [Alexandre Wallyn (firstname.lastname@example.org), Neuron Digest, 4/17/93.]
Software (or neural-network) patent analysis may be an important consulting area. One analyst charges $10-$25 for each patent in an analysis, or about $15K-$35K for a competitive patent-trend analysis of something broad (such as OCR, with 1700 US patents). [Jim Carroll (email@example.com), Online, 4/5/93.] Contact Jim if you need the analyst's address. (He gets a finder's fee.)