The problem of Emails

As information becomes more and more available, the act of managing information becomes more and more challenging. As knowledge work becomes increasingly predominant, the importance of managing this work clearly increases. Information overload occurs when a knowledge worker is unable to effectively manage the continuous stream of knowledge objects that he or she receives and is expected to process. These knowledge objects include email messages, telephone calls, paper mail, etc. The problem comes from both the sheer volume of information and the interruptive nature of the arrivals of this information. Email alone is an enormous challenge. As the use of email has gone from a novelty to a necessity, email overload has become an increasingly relevant problem. It was estimated that in 2000 the average corporate email user spent almost two hours daily processing their email (Ferris Research, 2000). International Data Corporation (IDC, 2000) estimates that on an average day, around ten billion emails are sent worldwide. IDC estimates this number to more than triple to 35 billion emails by 2005, while the number of email mailboxes is expected to grow from 505 million in 2000 to 1.2 billion by 2005. Within the House of Representatives, for example, the number of email rose from 20 million in 1998 to 48 million in 2000 (Goldschmidt, 2001). Certainly, much of this email does not need to be processed. Spam represents a challenge that has been addressed by research. Filters are able to block much of unwanted email that is received. Internal spam also represents a challenge. Messages that are sent to an organization’s mailing list often result in workers receiving information that is not applicable or is redundant. Filters have also been developed to aid in this problem by categorizing incoming messages according to user-defined rules. Our research begins with email that has made it through such filters. In other words, we assume all email is in fact in need of processing. Given this, the general goal of our research is to aid the knowledge worker or network of knowledge workers in decisions regarding how best to process these email. Decisions that we have considered include how best to schedule email work, how best to prioritize email messages while taking into consideration possible future scenarios, and how best to distribute email messages within a network of knowledge workers. Our aim is to recognize and identify unique characteristics of knowledge work, and to prescribe solutions for improved worker performance. We consider both the stochastic nature of knowledge work, and the timely nature of knowledge work. Unlike raw materials in an assembly line, the time needed to process information is not always predictable. By ‘timely nature,’ we recognize that just as with other forms of work, knowledge work may lose value with time. For example, an organization may lose business, if they wait too long with a decision in response to demand. Our strategy is to use existing modeling tools to model and analyze the unique characteristics of knowledge work. Our focus within the knowledge work realm is the processing of email messages by both an individual knowledge worker and a network of knowledge workers. 

Email has, to a great extent, changed the ways in which communication is accomplished, and work is processed. Our research has focused on modeling and analyzing the processing of knowledge work (specifically email), with the goals of understanding and improvement of this process. Our modeling efforts thus far have relied on queuing theory and simulation. Like other forms of work, knowledge work has both inputs and outputs, as well as a process in between. Knowledge workers receive inputs in the form of questions, information, requests, etc. These inputs often come in the form of email messages. The knowledge worker must process these email, and respond with some type of output. This output may be a decision, some improved form of information, advise, etc. Because we can view the knowledge worker as a server and email as a customer, we see queuing theory as an appropriate fit for the modeling of knowledge work. Performance measures offered by queuing theory have meaning within an email context: the average wait in a queue can be seen as the average wait in a knowledge worker’s in box (email response time), and the average length of a queue can be seen as the average size of a knowledge worker’s inbox. Simulation enters the picture as we attempt to model the complexities of decisions regarding the processing of email. Email, like other types of work, may be prioritized based on different criteria. We’ve used simulation to model these prioritization decisions. Finally, we have used Stochastic Programming in an attempt to optimize the utility of those email that are processed, while taking into consideration both the current state of a knowledge worker’s inbox and both future arrival scenarios and possible future processing scenarios.

Contributor: Robert A. Greve

To the Main Page