Section: ednmag.Comment
We're so used to designing for product life
cycles of about a year that we too often forget that end users--the
folks who buy these products--tend to keep them for a longer time. That
period may be a few years for a tech toy, five to 10 years for a TV or
PC, or even 15 to 20 years for an industrial or household appliance. As
you know, the reliability of your designs depends on many factors.
These factors include the design margins you use, any fault-tolerant
design and error-correction techniques you employ, the quality of
components, any temperature and other stress factors, and even the law
of averages. This law can work against you when you have millions of
gates, and each one has to work correctly for thousands of hours.
But what if your product manager told you
that the user specification was that the product had to function for
thousands of years? For that requirement, many of the normal techniques
you'd use are simply not meaningful or applicable. You would need to
rethink every fundamental assumption you use in designing for long-life
operation.
If you think this goal is crazy, think again.
A group has been working over the last few years to design "the clock
of the long now," which will run for 10,000 years (www.longnow.com).
The clock will tick once a year, it will have a century hand that
advances once every 100 years, and its cuckoo will come out every 1000
years. To add to the challenge, it will have to account for leap years
and similar eccentricities. The project began in what the "long-now"
proponents call the year 01998. (That new leading digit is truly
disconcerting, accustomed as we are to the year numeral having just
four digits.) You can see that such a clock is a challenge of a
different order from what you normally think about.
This 10,000-year-clock idea is not coming
from dreamers with no sense of reality or experience. One of the
efforts' leaders is Stewart Brand of the Whole Earth Handbook and The
WELL (Whole Earth 'Lectronic Link). Another leader is Daniel Hillis, a
leading designer of supercomputers, who worked at the now-defunct
Thinking Machines Corp and who is comfortable dealing with nanoseconds
and picoseconds. One of their goals is to make us stop and think with a
time baseline that is radically different from the hyped-up one we now
use. Their proposed design will use a mechanical serial-bit adder based
on levers, because gears will wear out.
The principles that the clock designers are
using are of great interest and some relevance to design engineers who
are working on more conventional projects. According to the Web site,
the designers have five principles:
- Longevity: With some maintenance, the clock should display the correct time for 10,000 years.
- Maintainability: The clock should be maintainable with bronze-age technology.
- Transparency: You should be able to determine the principles of operation just by close inspection.
- Evolvability: You should be able to improve the clock with time.
- Scalability: You should be able to build working models from tabletop to monumental size, using the same design.
Although your products' end users will
undoubtedly be happy to get a useful lifetime that is much less than
that of the long-now clock, some of these principles, or variants on
them, are probably worth your consideration in the context of
validated, elegant, frill-free engineering design. For example, does a
user need special tools just to perform first-level investigation of a
problem such as a dead unit, which may be due to a blown fuse or dead
battery? Are the functions of some buttons or keys obscure? Does basic
maintenance require special tools, fixtures, and know-how?
Engineers will find a lot of irony and
perhaps some humility in the plan for the clock. We often smugly make
it clear to others that our products, with their general absence of
moving parts, will last longer than mechanical products, which suffer
from wear, friction, and other maladies of the real world. Yet, if you
want your design to last for centuries, the mechanical approach will
probably be the way to go. Your electronics products depend on
longevity and stability of basic materials and compounds at their
molecular level, which is unlikely to be the situation.
I've always been fascinated by those scenes
in Star Trek, where an Away Team comes upon the technological remains
of a now-vanished civilization or even an early Federation starship,
and everything is still in perfect working order. Will our ubiquitous
LCDs still be functioning in 500 years? Or will a mechanical readout be
a better choice? If you visit a museum of scientific instruments, note
how many of the items from 200 and 300 years ago are still in perfect
working order, and think about why that is.
PHOTO (COLOR): Bill Schweber
~~~~~~~~
By Bill Scweber, Executive Editor
You can reach Executive Editor Bill Schweber at 1-617-558-4484, fax 1-617-558-4470, e-mail bill.schweber@cahners.com.
© 2001, EDN, Reed
Business Information, a division of Reed Elsevier, Inc. All
Rights Reserved. Copyright of EDN is the property of Reed Business
Information and its content may not be copied or emailed to multiple
sites or posted to a listserv without the copyright holder's express
written permission. However, users may print, download, or email
articles for individual use.