Issue
Article
Vol.30 No.2, April 1998
Article
Issue
Issue |
Article |
Vol.30 No.2, April 1998 |
Article |
Issue |
The Knowledge Management Technologies was a collection of groups working toward a common goal: create the next generation of tools to give Mac users the ability to access and manipulate ever larger and more sophisticated kinds of information.
The lab had five areas, each with a distinct mission. In its structure, the lab took on a variety of ways to approach the issues of knowledge management.
As happens in many modern industries, although perhaps with more frequency in Silicon Valley, a re-organization led to the formation of the Knowledge Management Technologies Lab. In the spring re-organization of 1997, there was an opportunity to bring together several groups that had been working separately, but ones that could clearly be synergistic within a common charter of knowledge management.
Beginning that March, five groups came together to work on the idea that there would be an interesting set of technical advances for the Macintosh with respect to knowledge management.
The theme of knowledge management, as taken by KMT, was the ability of individual users to access, use and manipulate the increasingly large quantity of information available in their workspace.
To this end, KMT was working on the entire problem from user input (Spoken Language, Intelligent Systems) to overall systems design (User Experience Research, Interaction Design), to underlying data access (Information Access, Intelligent Systems).
It was our belief that only a coordinated effort that bridged the entire process from user requirements through underlying infrastructural support would be able to make effective headway in this area.
(Blake Ward, manager)
UER focused on understanding the user-centered issues of computation. Areas of specific work included:
The members of UER spanned the spectrum from system builders to visual designers to cognitive scientists. From this overtly multidisciplinary approach they focused on the basic issues of creating new kinds of UE -- that is, a User Experience could be designed that incorporates lessons from the user interface, but also takes into account many other factors as well -- the physical design of the system, settings of use, metaphor design, and expectations about the system.
(Jim Miller, manager)
The ISP worked on a variety of topics, each centered around automatic, intelligent support of users through the interface. The topics of special interest for ISP were:
(See Jim's article on ISP for more details.)
(Sally Grisedale, manager)
Interaction Design was a small group that explored new kinds of embedded computers. How could such machines be placed into desktop surfaces effectively (physical design) and once in place, how would users interact with the device?
(Dan Rose, manager)
Information Access has been working for some time on a full-text indexing engine for the Macintosh OS, and has been exploring ways such an integrated tool could be used:
(Kim Silverman, manager)
The Spoken Language group has been working on delivering high-quality text-to-speech and speech recognition services within the OS. At this phase of their work, research and development is centered on:
Every research lab has a style, a way of approaching the content of their work and a strategy aimed at pleasing their customers.
KMT's approach was a "portfolio management" style, one where results of consequence to the customers were scheduled to emerge every few months. In theory, projects were organized so that no long dry periods would result.
In practice, this means constantly juggling and reworking project schedules as some work succeeds, and other projects get hung up in their progress.
Additionally, each of the groups within KMT was charged with maintaining some kind of working relationship with a product organization. In three out of the five groups, there were ongoing research-to-product transfers in place, with a history of many others during the past few years.
Working in the research groups at Apple meant that a project had a kind of protection from the normal delivery schedules and production requirements: it was a place where projects always had a "clear path to product," and yet could still innovate and explore with some security.
As a consequence of this imperative, the working style of KMT was primarily pragmatic, while keeping a close watch on where theory could help.
KMT contributors typically studied real users in real use settings (as opposed to artificial tasks in usability labs), built deployable pieces of software for use throughout the Apple community, and tried to emphasize the tangible aspects of their work.
At the same time, theoretical issues were of importance when it seemed that some benefit would hold for the group or the larger community. This is particularly important in fields where small stepwise refinement cannot anticipate larger improvements in methods. A theoretical stance is particularly useful when it provides a way to see farther into future work. In this way, building models of the phonology, linguistics or statistical properties of text, could help in the fairly near term.
While KMT existed for only a relatively short time (from March, 1997 until ATG's dissolution in September, 1997), the groups within KMT had been around in some form for quite a while beforehand.
In these few months, the lab had begun to identify areas where one group could work productively with another. The Spoken Language, for instance, worked with the Information Access team on identifying word boundaries. Their earlier work on phonological indicators for reading email turned out to have useful consequences for arbitrary text scanning in the full-text indexing system.
Similarly, the Intelligent Systems Program worked with Spoken Language toward using speech recognition in complex dialog interfaces.
Unfortunately, the overriding vision of the laboratory -- one of providing an integrative approach to knowledge management problems -- really didn't have much time to jell. Aspects of the story were beginning to coalesce, as the area cross-fertilizations and resource sharing began: but it was not really enough time to see the larger picture.
After the dissolution of ATG in September of 1997, a few remaining research teams have been moved directly into production groups. In talking with those remaining team members three months later, they uniformly report a greatly improved sense of accessibility to the production side of the house, and a decrease in their long-term research capability. As is natural, production -- which has a relatively short-term budget cycle -- finds it difficult to put long-term work into their development plan. Such uncertainty doesn't contribute to attaining the targets they've been given, and in the current uncertain environment, short-term goals are the highest priority.
The side-effect, of course, is that qualitative improvements will be increasingly difficult to come by, and the pace of in-company innovations will be slowed.
On the other hand, it's clear that even in the best of production/research relationships (and I believe the Apple research environment was very productive), a barrier exists between the two halves of the house.
A continuing challenge for research management is how to dissolve this barrier without disrupting the work of either group.
One approach that showed promise was the practice of rotating people and groups into production from research. KMT had some experience with this practice, and it seemed very promising. A longer period of rotation practice could be very rewarding for companies concerned with the research to production transfer process.
Daniel M. Russell currently works at the Xerox Palo Alto Research Center after a spell of four happy years at Apple's Advanced Technology Group. He is once again the manager of the User Experience Research group, only this time for Xerox PARC.
Daniel M. Russell
Xerox PARC
UER
3333 Coyote Hill Ave.
Palo Alto, CA 94304 USA
email: DMRussell@acm.org
Tel: +1-650-812-4000
Issue |
Article |
Vol.30 No.2, April 1998 |
Article |
Issue |