Communications of the ACM
Technical opinion
Time-Asymmetry in Business Processes
Bill Gates, in his recent book Business@the Speed of Thought, argued that a digital nervous system (DNS) would enhance the efficiency and effectiveness of business processes.
To support this theory, the DNS is required to supply "accurate, immediate, and rich" information to different parts of the firm, and maximize "the insight and collaboration [thus] made possible." Though highly suggestive and pregnant with wide applicability, Gates' analogy between the human nervous system and the DNS does not capture a fundamental characteristic of business processes. The impulses transmitted by human nerves assume physical form in activities that move forward and backward in space. Since these positions can be routinely reversed with little effort on the part of the individual, in effect the real time involved goes both ways. In contrast, business processes are not readily reversible, so that real time only goes one way for the firm.
The applicability of Gates' idea would be enhanced, if the (real) time-asymmetry of business processes is recognized when designing and evaluating DNS and the underlying information technology and information systems. At any moment in time, individual ideas, expectations, assumptions, and information, which span past and future and propagate with "the speed of thought" over the firm's DNS, are reified in terms of decisions to implement certain business processes. As time passes, these processes remain optimal if events turn out as expected. (In dynamic programming, a similar observation is embodied in Bellman's principle of optimality.)
In business, what is optimal looking forward today need not be optimal looking forward tomorrow. Suppose expectations six months ago turn out to be unjustified. Say sales assumptions were too optimistic, and now it is necessary to retrench on production and marketing. If the firm is able to return to its original position and start afresh at zero cost, then effectively time can run backward in business processes. However, many of the actions decided upon have been implemented and cannot be undone without cost (abandoning a half-built factory or leaving it idle is costly), and interest over the six months would be difficult to recover. In this case, time goes forward, while changes in the firm's organization and operations must be partly or wholly reversed. A decision problem then emerges, to optimally run business processes backward over an upcoming period of time. This would require efficient support from the firm's DNS and underlying information technology and information systems.
Implications for DNS Design and Evaluation
The problem going backward is generally different from going forward. Since the decision environment is different, the static information required and its value would be different. More importantly, the task of coordinating, directing, and evaluating information flow in the active Dertouzosian sense shifts in dimensionality and complexity. Different demands are therefore placed on the firm's DNS and supporting information technology and systems when going forward and going backward.
Specialization leads to higher productivity if plans are realized. However, under time-asymmetry, a DNS specialized to process and select "actionable" (static and active) information going forward and optimize its rate of flow and use-value are unlikely to be equally efficient going backward. In the former situation, the firm seeks to maximize performance subject to constraints deriving mainly from expectations and ex-ante variables. In the latter, the decision changes to one of minimizing adjustment cost subject to constraints in which ex-post considerations feature prominently. Correspondingly, forward information would be associated with expectations and assumptions and is therefore more readily susceptible to numerical estimation and treatment, while backward information would tend to derive from softer, more organic and multidimensional effects (for example, those arising from managing human resource changes). Since different constraints apply, max/min duality does not follow: a DNS specialized to business processes going forward would generally fail to be as efficient going backward (and vice versa). Firms tend to discoverwhen they can least afford tothat specialization in information-based functions also works one way.
It is imperative for the firm to recognize that, under time-asymmetry, information processing and selection capabilities going backward are as important as going forward. Systems specializing in one direction are unlikely to function as well in the opposite direction. A DNS optimized with respect to both forward and backward business profiles therefore requires a compromise in information processing and selection capabilities along any one direction. On the supply side, this raises a challenge to architects and engineers to create information technologies and systems that are dual-performing. Though the resulting DNS may not be the most efficient in any specific direction, overall it would be superior.
Another systems engineering implication can be noted. Under time-asymmetry, the firm optimally thinks at different speeds going forward and going backward. Though its DNS should support business "at the speed of thought," the criterion of simultaneity must be modified to take this qualification into account. (A facility reminding users to pause and ponder at each decision node in a business process would be relatively more useful in the latter situation.)
Investment Evaluation
Standard project evaluation indices such as return on investment (ROI), that are calculated on the basis of planned scenarios, are likely to overstate net returns. Productivity gains from forward DNS and information technology and information systems specialization would be, if expectations and assumptions turn out to be misplaced, offset by unplanned retrenchment costs. Instead, a proper calculation of ROI should take into account expected net cash flows from going backward as well as from going forward. The increasing volatility of stock indexes may be due in part to the fact that when expectations turned sour, unplanned adjustment costs emerged to further damage profit prospects, or it was discovered with each application that firms' DNS and information systems and technologies were inefficient with respect to backward business processes. A recent consultancy survey of 451 companies in 22 countries reported many cases of growth- or forward-oriented systems being abandoned as unsuitable for management in the economic downturn.
Business planners are generally bullish, which often leads to a bias toward forward processes. However, it is equally important for backward business processes to be efficiently supported. The firm's DNS and underlying information technology and information systems must be designed and evaluated with due attention to time-asymmetry and its consequences.
©2002 ACM 0002-0782/02/0500 $5.00
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.
The Digital Library is published by the Association for Computing Machinery. Copyright © 2002 ACM, Inc.
No entries found