D-Lib Magazine
December 1995

From the Editor

Telling Time

How many times a day do you say to yourself, well, it's all a matter of timing? Or check your watch? Probably more than once. Notions of time are deeply embedded in our culture, and in the felicitous words of William Y. Arms, they are "highly context dependent." Or, as my grandmother might have said, where you sit is what you see. (She may have been talking about the movies.)

Certainly, the post-industrial West has striven to commodify time, and more than one historian of industrialization has noted the importance of accurate clocks and measurement as an enabling technology. Frederick W. Taylor built a science of management based on timing tasks and apportioning hours. Taylor's time and motion studies contributed to Henry Ford's successful implementation of assembly line organization, which was predicated in part on breaking down a manufacturing process into its replicable components and calibrating the movement of the equipment past the various stations.

In pure science, until Albert Einstein's invention of relativity in the early twentieth century, classical physics was based on Sir Isaac Newton's space and time. Newton's physics postulated time as an absolute; however, he also explored notions of historical time, compiling elaborate -- and inaccurate -- biblical chronologies. Newton's notion of time as an absolute is similar to the concepts of time embedded in Taylor's time/motion studies, but his understanding of historical time is akin to time as we experience it, that is, "psychological time," as described by Stephen Hawking, who now occupies Newton's chair at Cambridge.

Hawking talks about concepts of time that can be captured mathematically or imagined but not experienced, such as the reconstruction of a tea cup at the edge of black hole where it is unlikely any of us will stand. Yet time as a function of the speed of light is really quite different from an engineer's notion of time, which is still different from psychological time understood by historians and social scientists -- or even biologists and geophysicists. Notions of geologic time enabled Charles Darwin to imagine incremental changes that took place over tens of millions of years. In contrast, the slivers of time in which software engineers measure performance seem like completely different phenomena.

For example, a 1,000-byte packet takes approximately one thousandth of a second to be sent on an Ethernet. The machine in front of me processes a single machine instruction in approximately one hundredth of a microsecond. Expectations of and within the inanimate system are quite different and more precise from those that derive from the human end-user. Delays of less than a tenth of a second are widely accepted as imperceptible to users, but users may be willing to wait. How much is situational. When I asked some of my knowledgeable colleagues, one software engineer stated flatly that keystroke echoes are "mandatorily fast", but then said more speculatively that users' patience is a function of their perception of load on the system. A second systems designer told me that users' willingness to accept delay is a function of how hard they think the system must work to, say, return responses to a query. And at least one author has told me that I can expect D-Lib's readers to tolerate about 90 seconds for files to download. But even then, there are visual cues -- interlaced images, flashing messages -- to reassure users that the system is working and that, in fact, move them along before the transfer is complete. Yet as those of us who remember life before fax machines can verify, expectations of time and performance change, and what was acceptable yesterday may be unforgivable tomorrow.

It seems to me that digital libraries will have to deal with many of these kinds of time. Certainly, engineering time at the machine and human scale is fundamental to systems and networks that are and will be built. These embed notions of psychological time as well as time that is measurable yet not perceptible to humans. But the content side also carries notions of time. There is time as a subject access component of a metadata record, which may borrow from existing subject domain definitions (i.e., Jurassic). There is the passage of time in the sense of legacy data that must be integrated into current and future systems. And then there will be obsolesce of time-sensitive content and formats. Notions of obsolesce are likely to vary, since there will continue to be some institutions, like the Library of Congress and the National Archives, which will maintain an archival mission, and must be able to access and "read" obsolete formats; this requirement is a possible barrier to use of the digital medium as a preservation strategy. Finally, there is capturing time as a characteristic of the information itself. The simplest example is dating, which is an element of registration and authentication. More subtle are notions of capturing and representing information in which the linear, temporal sequence matters to the meaning of the discrete values -- econometric time series data, for example, or weather data. -- and which must be stored in a way that is different from data that is not time-dependent in the same way.

Digital libraries are proving a fertile area of research not just because they are collections of information for users on the electronic networks but because they compel us to look at research problems in a rich spectrum -- from response time of the processor to the patience and needs of users yesterday, today, and tomorrow. Thus, information technologies come full circle from fractions of seconds to the forward passage of psychological time.

Or, as my grandmother might have said, all things come in time.

Amy Friedlander

Copyright © 1995 Corporation for National Research Initiatives

D-Lib Home Page |  D-Lib Magazine Contents Page |  Next Story | Comments