Position Paper for Computer Science and Technology Board Workshop on Technology Literacy, Jan. 14-15, 1998
Susan L. Gerhart
Three key concepts and skills for functioning in current and new information technologies are:
All three aspects of the use of computing are, of course, informally practiced for survival as well as progress. However, these practices are not systematically taught nor provided for practice in traditional education, yet are just a step or two away from the modes of intellectual- and career-driven teaching and learning. These concepts/skills are definitely required for not only many computing jobs but also the solo professional and the educated individual who becomes dependent upon the Internet as a major information resource. The affordability, complexity, and essential role of the Internet and modern personal computers brings a new dimension of required knowledge and skills relating to systems and information.
Concept: Components of different types and acquired from different sources must function together to achieve the system’s goal, and continue so under inevitable integration of further components
Application of Concept: Integrating communication and computer systems, e.g. an ISDN connection to a home office. Subsystems include:
Engineering principles: Continuing the ISDN example:
Selection of components (services as well as equipment) with potential for functioning together, cost tradeoffs of features, expected lifetime of components and needs for system.
Defining requirements: is voice needed or just data? what speed will ISP provide, at what charge? reliability evidence for adapter? phone company repair service charges?
Modeling the system and acquiring its vocabulary: How is ISDN different from regular phones? How is a connection made and how is it represented to the system user? Are there sounds, lights, system logs or other indicators of progress and diagnosis? what’s a SPID?
Testing subsystems individually and together to determine malfunctions: Phone company certifies its lines, ISP certifies its service, now the system integrator must configure computer software, ISDN parameters and data (SPIDs, phone numbers, speeds, etc.). What’s the best integration strategy - "big bang" (put it all together and hope it works) or component testing, then pairwise, and on (which may be difficult to observe results)?
Negotiating for problem solving: who takes responsibility if the system fails or doesn’t start functioning? which is the most reliable, experienced, and intelligible representative of one subsystem who can help check out the others?
Documenting and archiving experience, settings, etc.: Once the system functions properly, what are the critical pieces of information for when it must be re-installed on another computer, when some subsystem must be repaired or change?
(Term adapted from a paper by Cathy Marshall in Hypertext 96 "Spatial Hypertext and Information Triage" discussing an experiment on how individuals using paper and hypertext systems performed a product recommendation task).
Concept: When rapid assessment is required, due to time pressures or volumes of materials, sorting by keywords or other features helps to develop strategy and provide useful preliminary processing.
Any set of materials is subject to various classification criteria, some standard, often devised on-the-fly for the task at hand.
Application: Example: an information analyst is asked to find instances of programming research that apply to the Year 2000 problem. Expert knowledge suggests "program slicing" and "program transformation" and "software testing" as terms representing fields. A desktop metasearcher yields over 1000 URLs on each term from multiple search engines (some are off-topic, of course).
Sorting strategies might be: mention of the term "year 2000" or "y2k" along with any one of the three terms; looking for a frequently linked to URL or a high-hit site (being aware of the self-reference phenomena); using Internet domains to classify commercial (. com, .net) from research (.edu, some .com, .org, .gov); separating technologies in the field (C, COBOL) from research technologies (logic, functional).
Educated adults perform information seeking tasks of different kinds: "find 1 instance" (usually not hard), "find 10 instances" (challenging in the above "Y2k and programming research" subject), up to "Due diligence" (show you’ve covered every reasonable source of information, including an audit trail). Thus, tools and processes must be flexible.
But it’s not sufficient to just find information, it’s also necessary to "qualify" the materials - the author, the citation if peer reviewed, the stability of the information (will it be there when you come back), the context and purpose for providing the information, date of publication, qualification by other individuals. Current lack of WWW-scale metadata conventions and technology support has lead to a rather sloppy state of authenticity, compounded by the rapidly growing massive amount of "gray literature" (practice but not peer-reviewed).
Engineering/Design Principles: Classification by text analysis is increasingly available from search engines. Classification technology (Library of Congress and other pre-coordination schemes), together with librarian training, illustrate the information engineering principles that must evolve to handle more diverse, much more, and changing information sets. But when it comes down to the desktop of an information broker, hobbyist, medical information seeker, etc. the requirements also include the ability to define personal criteria, use currently available and affordable tools to support triage, and sustain attention through, possibly 100s of pages. Engineering here involves integration of desktop productivity tools, plus design and evolution of a process.
Concept: every major action has known points of failure as well as unknown factors that can influence the success of the action and its effects on other parts of the system within which it operates. Identifying these factors, the chances of their failing, the overall likelihood of gain from the action versus costs of failure, and expectation of continued operation and effectiveness—the essence of progress—are often performed either blindly or intuitively, sometimes for the sake of learning, sometimes for sheer sense of taking the risk.
Application: The classical risk situation that faces all system managers (and that includes the home PC owner) is "upgrading". Consider the example of upgrading to the recent versions of the behemoth browsers supplied by Netscape and Microsoft. Will all the mail and news features work with current ones or get in the way? Will web pages still view the same (considering Javascript and Java)? Will it be possible to run both the former and the current browsers (e.g. to test web pages or to revert)? Is it possible the auxiliary modules to be installed by the new version will disable any other application? Will the resource consumption for extensive browsing sessions be compatible with other applications running concurrently or suspended? What do newsgroups tell about installation and with the new versions? Are the stories in the newsgroups credible and the fixes authoritative? .
Engineering/Design: Technology progress is risky not only for the product developer but also for the user. How is that risk distributed? It’s difficult to test a program like a browser in so many conceivable configurations of other applications, hence the practice of "beta testing". The essence of our example’s problem is "when to take the risk" vs. "the risk of not upgrading", forcing educated users of newer information technology to develop a personal risk strategy including the sources of information, the nature of the decision, and dealing with the consequences. Is there a way to recast larger system risk experience into smaller, but important to the single or small group users, terms? Is there an applicable body of risk engineering principles?
Age and Mode of Introduction
Skills and knowledge that used to be the province of professionals, such as librarians and engineers, are now being imposed upon, and required of, other professionals, students, hobbyists, and personal information seekers. For example, the solo professional (and just about everybody is on their own sooner or later) is also a system manager.
Elementary and intermediate school children exposed to electronic entertainment are highly motivated to install software games, identify needs for memory to save games, and integrate new components into existing systems, e.g. PlayStation and PCs. What children lack is the formalization of their experience: how did they get it to work? who did they consult and did that person know what they were talking about? Is the system at its max or the end of its lifetime (failures, outdated, outgrown)? Parents and other family members are, therefore, continually exposed to opportunities to instruct in the principles of system integration and risk taking. Likewise, when a child finds a topic of enough interest to scour the Internet for, e.g. "everything about Antarctica", then there are opportunities for teachers and family to present various ways of organizing, archiving, and presenting the results.
To what extent is this practical instruction a matter for formal schooling or for family or community training? Trips to the library to see how material is classified, scout troop projects re-assembling older computer systems, translation of risks in games to risks in real life—updates of current alter-education modes seem feasible. The problem, of course, is the education of adults both to see the need for and to be able to perform these skill-raising exercises.
The practice of "information triage" (as described above) is currently a largely individual and unformalized activity. Very little is known about how Internet users use browsers and other web utilities—how does the user organize his/her work, plan, apply tools, develop techniques, etc.?
At the college and continuing formal education levels, system integration appears to be either the domain of system administrators or a silent laboratory requirement. The opportunity for system integration experience is valuable but may not often be recognized as real experience in deference to the content being learned.
For adults, the introduction problem may be more painful. "Time is money" deters system integration attempts. Since so many information technologies are so new, where does the adult get needed advice? In the area of technical knowledge, the "Technology X for dummies and idiots" series of books often provide the model and vocabulary to interact with the various subsystem providers and make technology decisions. Is there a formula in the "X for Dummies" model that can help educators, information and system providers, and researchers to better prepare users for new technologies? Is the key vocabulary, the tone of the book, the assembly of practical knowledge? Is it breadth vs. depth that helps "dummies" the most?
Our System Integrator example emphasized the role of service in determining success of integration efforts. The problem for the system integrator is that of distributed, partial, and often inaccurate information, especially for the early adopter. The system integration problem is not just physical or logical components but also human communication, training, and desire to provide service. Video and animated explanations are often provided, but, due to production costs, increase the lag over written or verbal communication.
The "Help" industry is rapidly expanding as the need for technology intermediaries increases. How much of this change is actual increase in need for technology assistance and how much is a shift in the industry structure? Where once there were manuals, now there’s the after-market of a dozen book versions and "advisor" magazines. In one sense, this allows customization of the information to provide more than documentation, including application-specific examples, pointers to related products, instantiations of processes. This frees the established product developer from the technical documentation chore to focus on online help and the next version of the system, but loses the developer interaction with the user’s view of the system. Then there’s the self-help movement applied to computing. Newsgroups provide essential experience and information on technology problems. Risks in using this genre include consulting a wider group of confused individuals with different systems and goals communicating in a brief and text-based medium.
Given the complexity, newness, rapid change of much information technology, it’s important for the educated adult to know how to exploit this industry - which books and authors to consult, how to utilize and interpret newsgroup advice, generally how to cope with products through third-party information.
Backlash to rapid change, often driven by commercial competition rather than user needs, may slow the computing industry or disaffect certain classes of users. "Browsers" as video and advertising tools don’t help the library, education, and other information-intensive professionals who need better ways to practice "information triage" and improve web content. By definition, technology lag is axiomatic in rapid change, but it also raises the question of industry integrity. If users are encouraged, or forced, to always move on to new versions which contain features they don’t need, but never receive improvements or fixes for the features they do need, then is there a point of purchaser and user resistance ("no, I won’t download that new version" or even a point of melt-down (Windows 95 applications that won’t run on other Windows 95 systems because the .dlls are inconsistent)? Is change always progress? Is new always better?
Biology and ecology are taught in terms of systems, but how are computing and information technology taught as systems? This is as much an attitude as a subject. Gaining the skills of system integration or information triage seems not to fit well within a curriculum but these skills are essential for professional life.
Library and management science curricula complement computer science courses in exposing students to system (including costs, personnel, regulations) thinking. How are skills taught - in project courses, computer-based training, individually or in teams?
Just to relate one piece of personal experience (based in the Wang Institute Masters of Software Engineering Program), working in a "role" on a project is a key set of skills. For teams of experienced professionals with recent course work (e.g. design, testing, databases), the most valuable experience was often not the knowledge application but the chance to play a well-defined role different from past experience. Designers could experience QA, and vice versa. Extrapolating from the industrial and education environments, it’s increasingly the "guild" that expands the roles of the individual, who must play almost all roles alone. The Internet provides many opportunities for exposure to job situations as well as self-help, e.g. to learn what web designers do you subscribe to "web-consultants". How is all this informal, distributed, impersonal education going to be balanced with formal education?
Learning a new technology at a discontinuous point is a very interesting challenge. There are enthusiasts and early adopters who will learn just because of the challenge and intrinsic interest. It’s when tools change a person’s way of working that technology introduction takes time, and that’s the essence of IT. One approach, of course, is the stepwise demonstration of new capabilities, e.g. introducing offline browsing by downloading a website for offline classroom demonstration, leading to indexing materials, to embedded browsers, then to processes like information triage. Our experience is that, especially for Internet tools, it’s hard to even find a common vocabulary to describe needs for and capabilities of new tools.
Another approach are the product evaluation publications, increased even more by the use of the Internet. Product reviews, press releases, demonstrations, lab tests all help a user or organization to make a decision to move into a new technology. Aside from bias, the major problem with this model is that the, necessarily, haphazard mode of product evaluation, once again driven by rapid publication cycles. Some reviewers, such as Web Review’s Web Addict, dig into the technology but most build feature charts. Again, this is a problem of information assessment - "how does one get the best evaluations to help decide if and when to move into this new technology?"
Of course, the best way to learn a limit is to cross one. In that sense, current IT provides many opportunities and the situation is often inescapable. Peter Neumann’s Risks Digest provides ample instances of limits, risks, failures, and foibles, but how is this information transmitted to the wider IT user population?
In this paper, we’ve largely ignored the privacy, commerce, education, and other social aspects of computing to concentrate on the perspective of an Internet system developer and solo professional. Our experience says that the biggest problem is deciding when to make a change in our operating environment since that change not only takes our time but also propagates to other parts of the system. Of course, each individual develops his/her own risk-taking strategy. The role of producing educated users of IT is not just to inform about technologies and their uses but also to support the learning of system integration, information triage, and risk management techniques.
Susan Gerhart is currently a software developer, Internet consultant, independent researcher, information broker, and website content designer. Her model of "browsing in context" is implemented in a suite of tools that provide a desktop alternative to current browser technology. Further information is available at the Research Outlet and Integration website, operated by ROI JV.
Affiliation: Applied Formal Methods, Inc., and ROI Joint Venture (with Ralston Research Associates), Houston Texas. 281-486-8480, susan@twurl.com