Death March

By Richard T. Dué

Four years ago, Ed Yourdon wrote his startling and influential book The Decline and Fall of the American Programmer to alert North American information technology professionals both to the deteriorating state of their productivity and to the important challenges to their livelihood from their increasingly productive competitors in Europe, Asia and South America. This summer, he is writing a new book, called Death March, that chronicles the disastrous journey of missed deadlines and overspent budgets common to most of our information technology projects.

According to productivity metrics experts like Capers Jones, Howard Rubin and Larry Putnam, the average software project in North America is now six to 12 months behind schedule and 100 percent over budget. In North America, the rate of software project failure and total abandonment is now about 25 percent for projects of 5,000 function points (roughly 500,000 lines of COBOL). This rate of complete project failure rises to about 50 percent for projects of 10,000 or more function points. Given this pathetic track record, and coupling it with today's competitive pressures, the opportunities presented by the siren lure of new technology, and the ridiculous political games of "Dilbert-esque" management, Ed Yourdon's current assessment is that the best we can hope for on our next project is merely to survive.

In my regular Information Economics column in the journal Information Systems Management, I have previously presented the problem of the information technology productivity paradox [1]. Simply stated, the paradox is that since information technology hardware and software expenditures are approaching 50 percent of the U.S. economy's spending on capital goods, and that since the cost of computer hardware and telecommunications is declining exponentially, then why isn't the economy showing any signs of this potentially fabulous productivity increase? Instead of a productivity increase, Dr. Ernst R. Berndt of MIT and Dr. Catherine J. Morrison of Tufts University have found, for example, that the U.S. manufacturing industry’s investment in computerization from 1968 through 1986 actually returned less than 80 cents of each dollar invested. Paul Strassmann, the former CIO of Xerox and of the U.S. Defense Department, has found, in his recent study of over 1,500 U.S. organizations, that there is absolutely no correlation between levels of information technology spending and the profitability of the organization.

How can we explain this paradox? One suggestion made by Strassmann is that organizations, since the 1970s, have failed to account for and manage the true extent of their expenditures for information. This lack of accountability has in turn led to a fundamental misallocation of resources. Strassmann proposes the following analysis to determine the actual return on information investment:

He first approximates the cost of information management from company financial statements as:

Cost of Information Management = Cost of Sales, General & Administrative + Cost of Research & Development

In a typical company this figure is 25 percent of revenue and is growing. This contrasts to the 2 or 3 percent of revenues actually spent on computer hardware and software, which most organizations believe is the extent of their information investment.

He then calculates the cost of capital to the company as:

Cost of Capital = (Shareholder equity + Capital surplus) x Interest rate for equity capital

In over 80 percent of U.S. organizations, the real cost of information now exceeds this cost of capital. Capital has become just another commodity. The real asset the organization now needs to manage is its own ability to turn information into profits.

Strassmann defines information productivity as the return on information management costs:

Information Productivity = Value added by information / Cost of information management

where

Value Added by Information = Net profit - (Financial capital assets x Interest rate for borrowing)

In the public sector this equation translates to:

Information Productivity = Cost of operations / Cost of information management

In either case, Strassmann's analysis shows that for the majority of organizations the overall productivity of information is now negative. His latest studies show that in certain industries, including car and truck manufacturing, metals (steel and aluminum), computer manufacturing, and telephone companies, this negative rate of return is now actually accelerating downward.

Capers Jones, the American software productivity guru, has reported some additional, important observations. In 1990, the difference in the productivity of the best and worst software development practices of organizations in his database of nearly 6,000 software projects was 4:1. Today that productivity difference is 600:1. He interprets this to mean that those organizations which have made orders of magnitude improvement in their software productivity have already put their competitors out of business. Their competitors haven't woken up to this fact yet, but even if they committed themselves today to the "best practices" of the successful organizations, it would already be too late for them to catch up.

One definition of insanity is the repetition the same action again and again with the expectation that each time there will be different results. Yet, many supposedly sane organizations that I visit are still preparing for their next death march project by looking for some "silver bullet" tool or technology or some "Rambo coder" that will somehow save the day even though every other silver bullet or coding hero they have tried in the past has failed. Sadly, for most organizations the pursuit of client/server or "object-oriented" or data warehousing or "inter-," "intra-," "extra-" nets will have the same negative outcome. Unfortunately, as Frederick Brooks states, "There are no silver bullets, but there are werewolves." The problems are frighteningly real, but there are no magic solutions.

Of course, if you are young enough you may feel that you can somehow turn a death march into a success just by your superior coding skills, intelligence, or through your willingness to put in long hours on the job. If you are old enough, you know that following this approach you will most likely end up burned out, divorced, or having sold your soul.

The successful organizations I visit around the world are those that have gone back to management basics. Successful organizations have instituted appropriate productivity measurement programs. Amazingly, over 95 percent of organizations have no measurement of the productivity of their information technology investment. Those few that do bother to measure productivity use some inappropriate and counterproductive metric like "lines of code." Scarcely any are trying to determine and mange their actual value added by information.

Successful organizations use system development methodologies which are appropriate to the type of project that they are building and which are appropriate to their level of software development maturity. Over 66 percent of organizations do not have any system development methodology. Many of the organizations that have a methodology do not bother to use it.

Successful organizations know and continue to measure their level of software development maturity as shown on, for example, the Software Engineering Institute's (SEI) five-level model. The SEI reports that nearly 80 percent of organizations are still at the initial, chaotic level of maturity. Over 66 percent of organizations still do not practice the software engineering disciplines of structured programming, analysis or design that successful organizations have been practicing for over 20 years.

Successful organizations practice effective project management. Successful organizations have established development standards and policies. Successful organizations have identified "best practices" (e.g., peer reviews, "just good enough" software, weekly builds, etc.) and use them. Successful organizations use proven patterns of software development. Successful organizations reuse their plans, analysis, design, documentation, project management, testing, frameworks, and code assets. Successful organizations employ new techniques and tools only when they can demonstrate and prove that the new techniques and tools will contribute to the overall productivity of the organization.

My best advice is to make sure that you measure and invest in your own basic skills. Then find an organization to work for which is willing to do the same. Otherwise, start packing up for your next death march.

Already, people are lining up for year 2000 conversion projects. This deadline can not be changed. We already know that there are not enough skilled people or other resources in the world to fix and test all of the affected code by the deadline. Yet, these projects must be completed on time or the organization will go out of business. Thirty percent of organizations haven't even begun to look at year 2000 conversion, much less understand the problem facing them. Combine this latest death march will all of the other projects in the backlog and with the unrealistic expectations of management and users and maybe its time to start to look for that coffee shop franchise.

Notes:

[1] Richard T. Dué, "The Productivity Paradox Revisited," Information Systems Management, Winter 1994, Vol. 11, No. 1, Auerbach Publications, NY, NY, pp. 74 -76.

Look for further information on this topic at each of the Web sites included in this article.

Death March will be published in the spring of 1997 by Prentice-Hall.

Richard T. Dué is president of Thomsen Dué and Associates, Ltd., an international Information Technology consulting and training firm based in Edmonton. He can be contacted through his homepage or by E-mail or telephone (403) 439-4627.