INSS737 - Management of Information Systems - Fall 2000

Project Management

Lecture notes

This is out third lecture in IT management processes, related to the decision to keep IT in-house. The issue becomes how to manage the systems development process - project management.

Project management is an old discipline, inherited of other fields particularly engineering with large construction projects to plan and control. There are three main stages in project management: project definition, planning and control. Project definition includes risk assessment, scope, deliverables and feasibility. Risk assessment in IT is summarized in figure 1 and discussed extensively in textbook pages 451-466. Scope refers to what parts, subsystems, will be included in the system project. This is done by enumerating functions, activities to be considered in the system, and by defining the boundaries of the system - what is not part of the system project. Deliverables are the outputs of the system, what the project will generate for an organization in terms of results, reports, etc - when the project is finished what we should expect to result from it. Finally, feasibility includes operational, technical and economic feasibility, as shown in figure 2.

Project planning is more an art than science in IT. Milestones are the first notion that help as plan. Figure 3 illustrates this concept. Human beings have the natural propensity to procrastinate. If let to their own they will postpone doing anything up to very close to a deadline. The black line in figure 3 exemplifies this behavior. Now, suppose we create intermediary deadlines (milestones). What will happen? Again, procrastination will take effect and people will wait close to the first milestone to start working. As soon as the milestone passes, everybody relaxes and stop working, but not completely. Then come again another milestone and hush-hush there they go. The red line represents this behavior in figure 3. If you compare both lines you will see that the people using the red line (milestones) behavior worked much more and have a better chance of meeting the final deadline.

Project estimates are also very important in project planning. Two other concepts help us in this task the mythical man-month and the Brook's rule of thumb. Figure 4 summarizes the concept of mythical man-month. The top, left image shows what people think. If you add more people you will have jobs done in a shorter time. This is only true under certain circumstances. It is obvious that a project that requires 1,000 man-months cannot be done for 30,000 persons in one day! But in many other situations the problem is still the same but it does not seem so obvious. All depends on how the tasks can be partitioned, or divided in sub-tasks. In some cases, see image in the top, right, the tasks cannot be partitioned, where the man-month notion does not work at all. Most of the tasks in IT project management are either partitionable requiring communications, or tasks with complex interrelationships. To estimate man-months necessary in a project we first need to understand the interrelations between the various project tasks. Brook's rule of thumb are loose guidelines taking into consideration the mythical man-month concept, as shown in figure 5.

Project control is associated with a series of tools used both for project planning and control. Figure 6 illustrates a Gantt Chart, while figure 7 illustrates a Critical Path Method (CPM) graphical and table chart.

Case instructions

We will use the standard framework to analyze the case: (a) what is the problem, (b) what are the alternatives, and (c) what are the recommendations, like we have done before. We will discuss the Providian Trust case on the Web and in class.

1. What is the problem?

Why is it a problem? Is there a decision involved? If it is a problem, what are the pros and cons that makes it a problem?

2. What are the alternatives?

What can be done? There is always the do nothing alternative. If everything went well (rosy scenario) what we would like to do? If everything went wrong (doomsday scenario) what should we do? What can we do in between the rosy and the doomsday scenarios? What are pros and cons of each alternative? Be sure to list not only alternatives, but pros and cons.

3. What is you recommendation?

What criteria should we use to select the most suitable alternative in the case? Why you are recommending one alternative versus the others? How this alternative satisfies your criteria? How to implement the alternative you selected?