Insofar as they are developed to mimic or replace the reasoning and decision-making of human experts, expert computer systems are doomed to make mistakes. The problem of expert systems fallibility and its potential consequences are discussed. Categories of expert systems fallibility are presented and evaluated.

The issue of appropriate quality control metrics is discussed in the context of the production of baseballs. Specifically, Major League Baseball players stroked record numbers of home runs in 1987, generating what is commonly known as the lively ball explanation for the unusual power output. The Commissioners office released a study proving that the 1987 baseball was not unusually lively. However, after examination of the quality control metrics used by MLB; we demonstrate that (1) it is quite possible that baseballs were indeed lively and (2) that acceptance sampling alone does not properly address the quality control problem.

No abstract

This research provides a case study and some empirical data for a small company running a virtual office today. Based on the lessons learned from thisvirtual office a look ahead is taken to what the virtual office of tomorrow will look likefor that company.COSS is a geographically dispersed computer outsourcing company located in northern Colorado that has one unit specializing in network management and another that specializes in PC technical support. COSS has 9 clients in three cities located in two counties. COSS employs 14 permanent full-time employees, 5 permanent parttime employees, and as many as 20 temporary part-time employees on an as-needed basis. All of the temporary part-time employees are students at the local universities.The initial strategy behind developing the virtual office was to leverage existing lowcost communication media into an effective virtual network that would function as efficiently as a regular office. Thus, the problem here was not to create a virtual office per se, but to create a virtual central office.The economics of COSS's virtual office structure are described. This solution costs an average of 1.2% of monthly revenues, compared to a traditional office scenario that is estimated would average 3.8% of revenues. As importantly, the virtual central office solution has proven for COSS's clients to be more responsive and efficient than the traditional office solution.The single most important difficulty with COSS's virtual office has been identified as the lack of credibility it engenders in some potential employees going through the recruiting process. However, clients report no credibility problems dealing with a company using a virtual office. A second problem is that employees report some loss of company identity.The study of COSS's operation reveals at least two major areas in which new technologies are needed if their virtual central office is going to continue as the company grows. Today's technology of pagers and cellular phones is cumbersome and does not provide consistent access to the company's knowledge base. A cellular-level WAN or LAN is one solution.The problem with cellular-level networks is that cellular costs could potentially be prohibitive. Consequently, COSS is extensively investigating the potential of the second technological area: intelligent agents. In fact, the COSS partners see intelligent agents as having the greatest potential of all "future technologies" to enhance their revenues and control their costs.