| Past Meeting Archive | Los Angeles ACM home page | National ACM home page | Click here for More Activities this month |
| Check out the Southern California Tech Calendar | |||
|
Joint Meeting of the Wednesday, September 6, 2001 2001: A Year of Living Strangely Peter Coffee, eWEEK It's been an ironic year for the information technology
industry. Home and small-business buyers seem far less interested in technology
for its own sake, with heavily promoted products like the Pentium 4 failing to
strike sparks -- while at the same time, long-awaited breakthroughs in
nano-scale manufacture and high-speed communication are finally leaving the
laboratory. The Internet is maturing into a distributed computing platform, with
an industry standard mechanism for remote Web services interaction -- while at
the same time, Internet users say that they regard the 'Net more as a library
than as a marketplace for either goods or services. The dot-com implosion has
drastically reduced investors' interest in "pure play" Internet ventures --
while at the same time, major players in every industry are moving more
aggressively than ever to exploit the efficiencies of being on line. Make sense
of it all in an evening of analysis and comment from Peter Coffee, Technology
Editor of eWEEK, in his annual presentation to our fall kick-off meeting. |
Peter Coffee's Annual Review "2001: A Year of Living Strangely" The core of E-Week's audience is people with real business, real earnings and real customers. Their only challenger for front running technology is biotechnology. "Biotechnology itself needs a lot of electronic storage space so biotech is our friend." Storage volumes are still growing at a compound rate. One extrapolation carried to extremes predicts that sometime before 2100 ALL of the atoms in the crust of the earth will be used for storage of zeroes and ones. Desktop storage is growing the most quickly, and most of it, from 82-89%, is duplicated information. We all want our own copies and don't trust networks. People do not see the Internet as a distributed computing platform; they see it as a library. They go to the net when they want facts. Applications developers have changed; people no longer develop algorithms, they spend a lot of time determining what the job is, learning APIs and studying libraries. This is a big change in the way computing professionals do their job; we used to be much closer to the basic hardware (even drum rotation rates) and software. Now we have all become users. The core technologies have been having trouble moving forward in meaningful ways. The Pentium 4 chip is an example of a waste of usable resources. It runs at 2 gigahertz but doesn't really do 2 gigahertz of useful work. Clock rate is a lousy predictor of how fast things really are. Some cheaper Ibooks are faster at processing some graphics than MAC power books because they have a better controller. There are many systems wrapped around a CPU that affect the speed of your job. The notion of the net as a distributed computing processor is not widely appreciated. There are varying opinions on Microsoft .NET. Peter Coffee sees a lot of good in .NET. The pendulum swings back and forth from real smart servers (called mainframes) with dumb terminals to smart clients (called PCs) and dumb servers called database servers. .NET assumes either side could be smart. Things can be passed around as a text stream, not a lot of bits from within an opaque binary system. .Net will be a real revelation to people who have only been writing code for a few years. It is simpler to produce more complex code with simpler tools and methods. It has the potential to be very good but might have security problems and be very vulnerable to hacking. A database in Redmond, WA will know much about the user including his credit card number and what he has purchased. Microsoft people could not answer who owned the data they stored. They said, "Talk to our lawyers." This will cause a great deal of nervousness among users. How do people make money out of net services? Either micropayment or subscriptions could work; subscriptions are easier. Microsoft would like to sell you services on a subscription basis. (They have learned that three service calls wipe out the profit on shrink-wrapped software.) Corporate sites are starting to move away from Microsoft software because of Windows registration requirements. After reviewing their software, and how buggy it has been, many of them have been moving to Linux. The notion that the Web is the place where you build your business is supposed to have driven the HP / Compaq merger. This is a bad merger as there is nearly a one-to-one correspondence of their product lines. Dell and IBM are breaking out the champagne to celebrate; they are picturing 6 months of shooting fish in a barrel and getting former HP and Compaq customers. Peter Coffee says he does not know how the Bush Justice Department will look at the merger because they don't have a track record on new mergers. HP and Compaq are trying to figure out what parts of their product lines will survive. 64 bit platforms are the place for Web servers but HP and Compaq are throwing away their own 64 bit platforms and moving to the Intel Itanium hardware. In contrast, Sun and IBM can offer their own systems in competition and Dell looks good as a retail competitor. Peter does not know whether IBM can take marketing share away from Sun. He says it depends on hard-to-predict marketing expertise. While companies with no business model have been dying the deaths they richly deserved, Cisco was never worth 160 times earnings and could not justify its selling price. The hi tech market was the old Holland tulip boom and bust all over again. Core technology is greatly improving. Motorola has announced a gallium arsenide chip. Gallium arsenide has been too brittle, making manufacturing a problem. They have laid down gallium arsenide on top of silicon and have made a very important advance. The net step is indium phosphide, which is in the same class. Now you can do optical processing at incredible speed. Peter saw statistics that 90% of optical fiber in the ground is either unused or under used. If bandwidth gets fat and cheap; hardware problems will be the bottleneck. There is interesting work on new architectures that get good results with many defective chips. The machines are imperfect within the box but still give accurate results, and the machine can be fixed over the net while it is running. Despite what may have seemed to be a terrible year, it is an incredibly fun business to be in. Wireless is coming along very fast especially outside the United States. People want a small one-bit interface, a magic eight ball that really works. One problem with wireless is spectrum, finding efficient and legal ways to use spectrum. There are various technical ways of using the spectrum including systems dependent on very cheap computation. The problem with wireless is security and virtually everywhere security is unbelievably loose. Peter believes in strong cryptography. Government crypto control doesn't seem to be a problem any more as the U.S. Government has basically given up on control. Voice recognition is working now.
Speech is determined from context. One method is to use GPS to find out where
you are, and use your location to narrow the problem of the context in which the
speech should be interpreted. |
| Past Meeting Archive | Los Angeles ACM home page | National ACM home page | Top |
Last revision: 2001 0518 [Webmonster]