A case study of Nevada DMV system

ABSTRACT

If an information system serves a large population, any delays or poor quality in the service can be a politically sensitive issue enraging citizens. This paper summarizes the failure of change management practices during the implementation of the Information System used to run the operations of Department of Motor Vehicle in the state of Nevada, USA. The Nevada DMV was in the process of moving from old-style mainframe-based computing environment to the state-of-art client-server architecture. However the new Nevada DMV information system resulted in confusion and inconvenience to the citizenry of the state by creating problems such as 5-hour waiting time for driver's license renewal among others. This paper posits that the failure of adopting the adequate change management policy was the root cause of this problem. This study analyzes the development and implementation stages of the system using only publicly available information. The analysis of the public record supports the proposition that the change management using the method called "direct cutover" was the wrong choice.

1. INTRODUCTION

Hirschhorn (2000) identifies that the success of change management programs has more to do with execution than with conceptualization and successful change programs have one thing in common: They employ 3 distinct linked campaigns--political, marketing, and military. The execution part should involve political support gaining from customers, marketing to obtain employee's support, and military-style precise and well-planned execution of the change. All three aspects should be considered at the same time to achieve successful change. In changing an information system of an organization, this principle also applies. To gain political support, it is imperative for customers who receive service through the information system to be satisfied.

To gain supports from employees who operate the information system, it is critical to provide good educational program to ease the burden of learning a new system. To implement the new information system well, it is essential to have a detailed plan on how to conduct the system transition. The subject organization of this paper failed to gain political support due to poorer performance of the new system. In addition it didn't sufficiently educate employees on how to run the new system and military campaign to execute the change was careless and the wrong method of system transition was used. Converting from one IT environment to another needs careful strategy since many precious data and critical operations are affected.

The conversion, often called "cutover" can be done in several different ways. Palvia et al. (1991) summarized cutover strategies into four methods as following:

* Direct Cutover: cutting over at once from the existing system to the new system abandoning the old system completely right away.

* Parallel: running the existing system in parallel with the new system until complete conversion is accomplished.

* Phased: cutting over from the existing system to the new system in phases; one module system at a time; partial conversion happens gradually until all modules are replaced.

* Pilot: cutting over to a new system in one department of the organization and only after obtaining confidence in the new system, conversion is applied to other departments of the organization.

They suggested that the determination of which strategy was adequate for the given organization's conversion depended on three factors: system size, timing, and target organization size.

For example, when the organization is very large and all departments do not need to do conversion simultaneously, pilot cutover could be a good strategy since the organization can have enough time and opportunity to take care of any problem that might be caused by the transition. However if the entire organization needs to use the new system at the same time, pilot strategy is not desirable. If the system could be implemented in module by module style, Phased strategy can be used to ensure smooth transition from the old system to the new one, which gives enough time and opportunity to fix any problems originated from cutting over process as in the pilot strategy case. When the system should be implemented altogether at the same time in the entire organization, the risk of running into unexpected bugs during transition period is very high. If the organization doesn't have to put the information system in service 24 hours everyday, parallel strategy is the method to adopt since when the unexpected bug is found in the new system, the organization always can resort to the old one which is still running. This backup ability makes sure that system-down or slowing the service of the information system to clients should not happen. Unfortunately some organizations find maintaining the old system together with the new system impossible due to technical reasons such as incompatibility of different databases as well as extra cost of maintaining the old system.

These organizations choose to go for "direct cutover" which is the most risk-prone method and requires extremely careful preparation before the new system goes live. In other words, direct cutover should be the last option for any organization which is concerned about the safe transition to the new system with minimal disruption in customer service. Even after careful preparation, any unexpected bug could shut the new system down where there is no backup system available. Thus when other cutover method is not out of the question, direct cutover should not be the choice of transition especially for information systems that provide critical services such as banking, military operation, and law enforcement-related operation such as driver's license and motor vehicle registration processing.

Therefore I would like to posit the following:

Proposition 1: Characteristics of service rendered by the information system should be considered as another important factor in determining which cutover strategy should be adopted.

Proposition 2: The "direct cutover" was the wrong way to go for the implementation of the Nevada DMV information system.

Namely, if the service rendered by the existing information system is politically sensitive, i.e., any malfunction or stoppage of the system can do great harm to large citizen population, direct cutover should be avoided. The system developer should try to understand the characteristics of the service of the information system that the developer is building and must assess the level of political sensitiveness of the service. And based on objective assessment of the level of political sensitiveness of the service of the information system along with factors such as system size, timing and target organization size, the developer should decide which cutover method will be used in transition.

The D consulting who took the responsibility of developing the new information system for Nevada DMV seemed not to pay attention to the unique characteristics of government information systems like the DMV system that could be often politically sensitive given the high dependence of the citizenry on such systems. Government works differently from corporations. Dealing with the citizenry requires a much more cautious approach than with the clients of corporations since the citizenry may get agitated into a fury quickly over some troubles that they perceive being caused by the government's either inaction or mishandling.

In the past the issues that invoked revolutions are food-shortage, unemployment, lack of freedom, etc. Nowadays one more item that could make the lives of the citizenry miserable is the information systems that are used by governments. Imagine that the FBI suddenly pursues an innocent citizen whose information system falsely identified him/her as a terrorist. This might be an extreme case with almost zero possibility of happening. However imagine that a citizen went to register a new car at the DMV and ended up waiting for 5 hours in line because the new information system at the DMV had a technical problem which made it either very slow or down for a long time. And imagine that it used to take less than 20 minutes to complete the same transaction with the old legacy information system. If this kind of situation goes on for more than a year and affects the entire population of a state, what would the people of the state do? They will rise up against the government and demand the explanation because the service from the DMV information system is mission-critical for the citizenry, i.e., if they don't obtain the driver's license, they can't drive; if they don't register their cars on time, they are violating the law.

To the citizenry, the DMV services are politically sensitive enough albeit the level of mission-criticalness is much lower than banking or military systems. And this aspect of the unique characteristics of the DMV information system seemed not a major concern to the D consulting, the system developer. This unlikely scenario actually happened to the state of Nevada DMV and its people. The Nevada DMV's new client-server architecture information system with GUI interfaces was put into use in September, 1999 and as of September, 2000, the system was still slower than the old mainframe-based legacy system in terms of average transaction time. D consulting approached the implementation of the DMV system as if it had been another corporate information system without considering the different nature of the information system that was supposed to serve the citizenry who were probably not as patient or understanding as the usual corporate clients. One should not attempt to install the system with direct cutover method and assume that any delays caused by run-time bugs will be accepted or tolerable by the citizenry.

This case provides a valuable lesson that system developers should go through more extensive testing stage and before the direct cutover installation, make sure to decrease the number of run-time bugs at a much lower level than in usual corporate information system delivery. It is also highly desirable to adopt the parallel cutover method in which the organization let the legacy system coexist with the new system until the new system is finally running at the acceptable rate. In other words, in change management for the new system adoption, military campaign method should have been the parallel cutover one along with better marketing campaign by giving adequate training to the employees.

2. LEGACY SYSTEM AT NEVADA DMV

Information systems built on the architectural paradigm of a mainframe and dumb terminals are often referred as legacy systems and usually become the good replacement targets with the client-server architecture paradigm. Along with the hardware change, consultants will recommend a move from text-based screens to GUI-based ones for the user interface as well as a linkage to the Internet. If the legacy information system takes 20 minutes to process a transaction while the new client/server system takes more than one hour to process the same transaction, the change must be not worthwhile. The consultants will say that the situation is due to temporary bugs and the incompetence of workers. They might add that eventually the new system will improve to the point that it will take less time than the legacy system.

Theoretically the argument by consultants must be correct and users in the conventional corporate environment will be told not to complain anymore. However what if the legacy system is the one that is used by citizens for essential public deeds such as drivers' license issuance and vehicle registration? Can the citizens remain patient? Obviously it was not the case in Nevada DMV. Some mission-critical information systems such as military applications do require thorough fixing of bugs before they are put into practice. Although only the real battles can show all the bugs of the information system, the developer will try to test the information system in every possible situation before the live-on installation. Thus it is required for the developer of these mission-critical systems to put much more effort in the testing stage than conventional corporate systems such as the personnel record management system.

Unlike military applications, the DMV information system does not affect the outcome of life or death matter. It might seem to consultants that the DMV information system is just another conventional corporate information system. For example, personnel record management system has a striking similarity to vehicle registration system as well as drivers' license issuing system. Both are essentially data base management systems that either vehicle identification number or social security number is a key field in a record. Other fields will share the same values such as address, phone number, date of birth etc. Both personnel record management system and vehicle/driver registration system will utilize the identical procedure for the database query processing. For example, the DMV system will look up a record of vehicle registration of a car accessing the record with the key field value of a vehicle identification number while a social security number is used to access a personnel record. Creating the data base schema as well as writing the data base query for DMV system must have looked easy and familiar to the developers who were used to developing corporate data base systems. They neglected the important characteristics unique to the DMV system; the DMV system was politically sensitive.

3. RELEVANT LITERATURE

There are numerous examples of system transition for mission-critical systems using parallel cutover methods. Messmer (2000) reports about the transition of EDGAR (Electronic Gathering, Analysis and Retrieval) system that is the US government's online document system for corporate business data such as earnings statements. Obviously the system is mission-critical, thus when the government decided to move from OS/2-based network to Windows-NT-based network, it chose to go ahead with parallel cutover method by running the old system together with the new system. Visa International Inc.'s old mainframes which contained and processed critical financial information on people's credit card authorization moved to Unix-based authorization servers that connected 600,000 point-of-sale terminals with Visa's large data center (Vijayan, 1998). Visa chose to do the transition by adopting parallel cutover method in which cutover was done in multiyear periods while both systems have been running.

Grover (1999) addressed the fact that when the system change happens, not only technical factors but also non-technology issues such as strategy and people should be important concerns in change management. Especially mission-critical systems require careful management of non-technology issues such as people's willingness to stand the hardship during the cutover installation stage. The situation at Nevada DMV offices was worsened by the fact that workers who had to go through training for the new system didn't have the adequate technical knowledge to explain the problem in a reasonably understandable way to the citizenry. As Fisher (1999) pointed out, the technical communicator could play a beneficial role in dissolving disputes that arose due to lack of technical knowledge. However there was no technical communicator who could play a role in calming people down in Nevada DMV case. The citizens of Nevada went to their politicians who, subsequently, held numerous public hearings over the fiasco. In hearings, the developers simply reiterated their position that the bugs in the new system were not abnormal; what happened in the DMV system usually happened to other new systems during cutover installation stage.

Due to the nature of being a government information system, the Nevada DMV system was built for a hierarchical organization that lacked flexibility when facing fatal errors. It was noted that improvisation could alleviate the pains during the change period (Macredie and Sandom, 1999). For example, if local workers at DMV offices had the legacy system on for particularly busy hours the complaints could have been reduced remarkably. In other words, from the beginning the parallel cutover method should have been used instead of direct cutover method. The DMV leaders as well as developers who held on to the unrelentingly rosy view on the system did not allow this kind of improvisation. The overly optimistic view on the IS development project often led to serious disasters (Glass, 1999; Grimson et al., 2000; Vise, 1999). The number of bugs in the new system could have been lowered if a few high-defect modules had been identified and treated soon after the installation as Tian and Troster (1998) found from a comparison study of software defects with a legacy system and a new one. Tian and Troster also found that defects in the new system were closely related with various design features. Thus the fact that it took nearly a year to reach a somehow stable level of bugs in the new DMV system indicates that developers should have done a better job of taking care of software defects.

Shifting from legacy systems should not be a hurried road; this is especially true for politically sensitive systems. As Bergey et al. (1999) suggested, moving toward a new system from a legacy system should be a disciplined evolution since the global environment rather than a single line of software products has to be considered; you never know what kind of side-effects the new system might create. By choosing to close down the legacy system all together and shifting to a new system in the next day, the developers of Nevada DMV systems took the undisciplined road that led to unbearable inconveniences for people of Nevada. Developers must have taken the operational viewpoint on the legacy system of Nevada DMV among viewpoints on legacy systems identified by Alderson and Shah (1999). They considered the DMV system as the system that supported mechanical operations and could be replaced with the new one without a big fuss as if a machine part were to be replaced. However the politically sensitive system like the DMV system cannot be treated as a machine composed of readily interchangeable parts; the DMV system influences the lives of the citizenry as well as the users who are government workers.

Some politicians who spoke at the public hearing over the DMV matter questioned the wisdom of moving to client/server system itself. As Duchessi and Chengalur-Smith (1998) reported, management's primary motives for moving toward client/server technology should include non-technical issue such as improved customer service. In the Nevada DMV case it looks like only technical concerns such as avoiding outdated architectures were reasons for moving to client/server systems. If the citizens are not helped by the new system, there seems to be no point of having the state-of-art architecture. The developers of Nevada DMV systems not only failed at the change management of the legacy system but also failed at the adoption management of the client/server system.

4. IN THEIR OWN WORDS: ANALYSIS OF PUBLIC HEARING

During the installation stage, a decision was made to bring down the existing legacy system all together on the day of launching the new system. In other words, they decided to adopt "direct cutover" method in transition. In addition some of bugs in the new system were considered as the ones that should be solved after installation, or "Go Live" stage. Thus management of DMV and the system developers went ahead with the plan to launch the new system in 9/2000 and ended up with serious bugs that made the new system down for many hours of the day. Due to the decision to retire the legacy system all together, DMV workers had no choice but to be idle during the down hours, which caused long lines of waiting in DMV offices (Morrison, 1999).

The management of DMV as well as D Consulting considered this kind of difficulty as normal. Following are the comments by Donna West, head of DMV Project Genesis team (Anonymous, 1999).

"The department is anticipating long fines through September and October as our technicians learn the ins and outs of the system and as final bugs discovered and fixed."

The citizenry who were affected by the new system unleashed an enormous political firestorm that resulted in a public hearing by the state politicians and all parties involved with project Genesis. Following are the comments by Ginny Lewis, Deputy Director of DMV in the hearing (Minutes of the meeting, 1999).

"Our progress is steady, perhaps not as rapid as we would like, but I continue to be confident that we are headed in the right direction. There is nothing that I view as alarming to jeopardize our success. In retrospect, the department would have made the same decision, knowing what we now know, when we went "live". What has occurred over the past four weeks could only have been learned in a five environment. We knew that there would be "bugs" pushed out of the system when we went "live" but we also knew that we had to be responsive as problems were detected and that is what we have done."

Management as well as developers maintained the position that there was nothing wrong about going "live" with anticipation of delays due to run-time bugs. However this attitude is not appropriate for mission-critical systems such as the DMV system where the citizenry could be arrested for not being able to register their vehicles. Following are the transcribed comments by Jon Lemelin, a representative of D Consulting in the same hearing.

"Mr. Lemelin said he wished to reiterate that it was D Consulting firm's commitment to do everything possible to help alleviate the long lines and that D Consulting would continue to look at system performance and continue to make modifications to the software to speed up the transaction time. He also pointed out that the screens were currently operating every 3 to 5 seconds, which he thought was well within the industry's standards."

It is clear from the transcribed comments that D Consulting maintained that all the suffering by the citizenry was normal and in accordance with "industry standards". However "industry standards" are designed for corporate clients, not for an angry citizenry. It is obvious that D Consulting didn't understand the need to choose different cutover method such as parallel cutover one toward mission-critical systems like the DMV system since it did not give a careful consideration on the characteristics of the service rendered by the existing information system. After several months of uproar, the bug problem was considered as contained around 1/2000 (Whaley, 2000). In 3/2000 due to political pressure from politicians as well as angry citizenry, the governor removed the head of DMV (Morrison, 2000). However at the time of writing this paper which is 10/2000, some customers still wait for an hour to get their vehicles registered. D Consulting insisted that eventually the processing time should be reduced more as DMV workers became proficient with the new system as well as further fixing of remaining bugs were performed. Nevertheless the bad publicity about the performance of D Consulting in the Nevada DMV system case is still being generated and spreading.

5. CONCLUSION

The Nevada DMV system that caused 5-hour waiting lines for driver's license renewal is a valuable case in which we can learn about how to manage the change of politically sensitive legacy systems. Without considering political ramifications that could be touched upon by angry citizens who have to put up with long delay, conducting direct cutover to the new system with the assumption that the new system might produce bearable number of bugs should have not been attempted. The information system developers should have been more concerned with the characteristics of services provided by legacy systems that they tried to replace. Therefore, both proposition #1 and #2 outlined at the outset of this paper hold true in this particular case study.

Lessons from the Nevada DMV System Case

The IS developers as well as IS project managers should be aware of the following lessons which were drawn from the experience in the Nevada DMV system case.

1. Check to see if the system is politically sensitive. If the system is a government information system and the system's operation affects general public, it could be a politically sensitive system.

2. If the system is politically sensitive, pay extra attention to the cutover method choice. Especially when considering "direct cutover" installation, try to have more time and effort to spend in minimizing the number of bugs.

3. If possible, choose "parallel cutover" method. Don't throw the legacy system away all together. During the installation phase it is useful to back up things with the existing legacy system particularly when a bug runs amuck inside the system.

4. Try to educate public on the inevitability of delays during installation phase. Do not maintain the attitude that we didn't do any wrong and you people just don't understand.

APPENDIX--PROJECT BACKGROUND

The Nevada DMV is one of the state of Nevada's largest departments, employing almost 2,200 people in 36 offices statewide. The responsibilities of the department include enforcing statues regarding vehicles and watercraft, licensing motor vehicles and being accountable for all functions of the Public Service Commission, the Driver's License division and the Nevada Highway Patrol. The department is organized into the following divisions: motor vehicle, motor carrier, driver's license, Nevada Highway Patrol, gasoline and special tax, administrative services, investigation and training.

Currently the DMV handles 131,000 vehicle registration renewals and 10,000 new vehicle registrations a month. It also handles 6,000 new driver's licenses and 30,000 license renewals per month (Macy, 1999). In the past decade the state of Nevada has seen phenomenal growth of population and the DMV was flooded with people applying for new driver's licenses and vehicle registrations. As a part of improving and modernizing government functions, the state government launched a project named Genesis to create the state-of-art information system in replacement of the existing system that was built in 70s (Edwards, 1999). However the idea of moving toward a new system was not brought up by any visible troubles of the existing system in terms of serving people of Nevada; there were no long lines of customers in DMV offices.

It was more based on the idea of creating the new information system architecture for the future in which more population growth is expected. D Consulting (pseudo name is used to protect the identity of the firm that is one of major management/IS consulting firms in US) was chosen as the main developer of the new system. The responsibilities of D Consulting included application design, coding and unit testing, debugging with 3-moth warranty, pilot and "Go Live" assistance, and end-user training. The original contract ran from 4/1998 to 11/1999 with the installation, or "Go Live" date in 9/1999. Right after the installation of the new system, bugs forced the system down frequently, which resulted in long lines in DMV offices. Thus the state government of Nevada gave an extension of contract to receive maintenance support of D Consulting from 12/1999 to 6/2002. In total $14 million was already spent for the project Genesis that will cost $35 million eventually after 7-year project period is over (Vogel, 1999).

The existing legacy system was based on a mainframe with dumb terminals with text-based user interface and the new client/server architecture-based system was implemented with graphical user interface as well as Internet access. The existing legacy system was complex and difficult to use since the DMV workers had to access several different systems to obtain information regarding driver and vehicle information. Vehicle registration and driver's licenses were treated separately in two separate information systems that were designed in the early seventies by different vendors. One was used to handle driver's license information and the other with vehicle registration information. Each system stored much of the same information and maintaining the correctness of information at both systems was difficult and costly. With the new system, developers claimed that these problems would disappear since the new system would be a single information system with graphical user interface where duplication problem as well as usability problem was solved.

REFERENCES

Alderson, A., and Shah, H. "Viewpoints on legacy systems", Communications of the ACM, 42(3), 1999, pp. 115-116.

Anonymous "In brief", Las Vegas Review-Journal, Aug 31, 1999.

Bergey, J., Northrop, L., and Smith, D. "Enterprise framework for the disciplined of legacy systems", Information Strategy, 16(1), 1999, pp. 15-35.

Duchessi, P., and Chengalur-Smith, I. "Client/server benefits, problems, best practices", Communications of the ACM, 41(5), 1998, pp. 87-94.

Edwards, J. "Online, not in line", Las Vegas Review-Journal, Jun 20, 1999.

Fisher, J. "Improving the usability of information systems: The role of technical communicator", European Journal of Information Systems, 8(4), 1999, pp. 294-303

Glass, R. "Buzzwordism and the epic $150 million software debacle", Communications of the ACM, 42(8), 1999, pp. 17-19.

Grimson, J., Grimson, W., and Hasselbring, W. "The SI challenge in health care", Communications of the ACM, 43(6), 2000, pp. 48-55.

Grover, V. "From business reengineering to business process management: A longitudinal study of trends and practices", IEEE Transactions on Engineering Management, 46(1), 1999, pp. 36-46.

Hirschhorn, L. "Campaigning for Change", Harvard Business Review, 80(7), 2000, pp. 98-104.

Macy, R. "Drivers losing patience with DMV computer glitch", Las Vegas Review-Journal, Sep 24, 1999.

Macredie, R., and Sandom, C. "IT-enabled change: Evaluating an improvisational perspective", European Journal of Information Systems, 8(4), 1999, pp. 247-259.

Messmer, E. "Thanksgiving no holiday for EDGAR", Network World, 17(42), 2000, pp. 18-19.

Minutes of the meeting-19991005, "Minutes of the meeting of the Interim Finance Committee's Subcommittee on Project Genesis/Department of Information Technology", http://www.leg.state.nv.us/70th/Interim/Statcom /Genesis/Minutes/IM-Genesis-199910052101.html, The State of Nevada, 1999.

Morrison, J. "DMV delays bring blame", Las Vegas Review-Journal, Sep 25, 1999.

Morrison, J. "Guinn removes acting DMV director", Las Vegas Review-Journal, Mar 10, 2000.

Palvia, S., Mallach, E., and Palvia, P. "Strategies for converting from one IT environment to another", Journal of Systems Management, 42(10), 1991, pp. 23-27.

Tian, J., and Troster, J. "A comparison of measurement and defect characteristics of new and legacy software systems", The Journal of Systems and Software, 44(2), 1998, pp. 135-146.

Vijayan, J. "Visa moves some apps off hosts", Computerworld, 32(45), 1998, pp. 63-64.

Vise, D. "Computer ills may imperil patients: Aging systems, budget cuts blamed for problems at D.C. General", The Washington Post, Mar 24, 1999.

Vogel, E. "DMV to speed up plan for phone, web services", Las Vegas Review-Journal, Nov 18, 1999.

Whaley, S. "Guinn backs DMV request", Las Vegas Review-Journal, Jan 27, 2000.

Author Profile:

Dr. Ook Lee is currently a professor of MIS in the College of Information and Communications at Hanyang University in Seoul, Korea. He held professorship positions at University of Queensland in Brisbane, Australia, University of Nevada at Las Vegas, USA, and North Carolina A&T State University in Greensboro, USA. He holds a B.S. in Computer Science and Statistics from Seoul National University in Seoul, Korea, and an M.S. in Computer Science from Northwestern University in Evanston, Illinois, He also earned an M.S. and Ph.D. in Management Information Systems from Claremont Graduate University in Claremont, California.

Ook Lee, Hanyang University, Seoul, Korea

COPYRIGHT 2004 International Academy of Business and Economics
COPYRIGHT 2005 Gale Group