Communications of the ACM

Virtual extension

Shadow Systems: the Good, the Bad and the Ugly

By Sandy Behrens
Communications of the ACM, February 2009, Vol. 52 No. 2, Pages 124-129
10.1145/1461928.1461960
Comments

We know them as shadow systems, workaround systems, and even feral systems. Operating at the fringes of an organization, they covertly replicate the data and functionality of formally sanctioned systems. Because of their duplicative acts, they are often said to have negative consequences for their hosts: undermining official systems,¹² sapping valuable resources and corrupting organizational data and processes.⁷ But not all shadow systems live up to their bad reputations. Some shadow systems offer an effective and efficient way for users to cope with the deficiencies of formal systems.^{1, 4} This article reports on an ethnographic investigation of a shadow system used in a higher educational institution, Central Queensland University (CQU), whose findings challenge conventional views of these organizational outlaws. The study explored how the system was built, implemented, and applied with a specific interest in people's lived experiences – the good, the bad, and the ugly. During the course of the investigation I observed a number of important lessons that impact on successful organizational use of these systems (see Table 3). These lessons should encourage technology practitioners and senior managers alike to re-examine typical prejudices surrounding shadow systems, and see them for what they are. In some cases, shadow systems may be bad to the core, but in others they can be just what an organization needs – a powerful source of creativity and innovation.

Because of their informality, shadow systems are rarely obvious in organizations and this can make it difficult for investigators to get access to them. In this investigation, though, I was able to avoid this problem by concentrating on a narrow but deep investigation of a shadow system popular at my own organization. This shadow system, called Webfuse, was developed to support the teaching and learning activities of academics and general staff in a single faculty at CQU. Background details of how Webfuse, and the corresponding formal systems, developed over time in response to significant milestones in CQU's history are documented in Table 1. The basic functionality provided by Webfuse, a home-grown system, is similar in many ways to commercially available learning management systems. Unlike commercial variants, however, it provides many additional functions tailored to CQU's specific circumstances. One of its more popular functions (one that shadows the functionality of the central system) is the uploading of students' end of term results into the Enterprise System. This process is referred to as 'results upload' or 'course results upload'. Features like this have helped to ensure that Webfuse is not only popular with staff but also firmly entrenched in their working lives.

This study was primarily informed by and triangulated with interviews and documentary evidence. Most interviews were conducted face-to-face, but because of the personal preference and work commitments of some participants, other interviews were conducted via phone or email or both. Interviews were conversational in style but focused on a specific set of questions concerning the participants' perceptions of and experiences with Webfuse. While the interviews were being conducted, documentary evidence such as email archives, organizational policies, procedures and charts were examined. This helped to clarify some of the puzzling and at times contradictory statements provided by interviewees and where necessary led to further, follow-up communication. The interviews and the documentary evidence were both analysed to determine the good, the bad and the ugly sides of the Webfuse shadow system.

A total of 17 initial interviews were conducted within CQU. Some of the study participants were more familiar than others with Webfuse because of their direct and regular interactions with the system. These interviewees included developers, users and sponsors of Webfuse. Other persons who had dealings with Webfuse but weren't as familiar with the system included developers and middle managers in the Information Technology Division (ITD). Additionally, persons in senior management (Chancellery) were interviewed.

Back to Top

The Good, the Bad and the Ugly

Hollywood movies tend to present the world in black and white, clearly distinguishing the goodies from the baddies. So, too, in the area of shadow systems, many researchers have yielded to a similar desire and cast shadow systems as the baddies. While this investigation reveals that shadow systems suffer their fair share of problems, it also identifies many positive characteristics. The most troubling afflictions uncovered were not inherent traits of the system but centred instead on more complex social issues, such as politics and the stigma associated with these systems. I have termed these types of afflictions the bad and the ugly. The good characteristics associated with the Webfuse system included its capacity to foster creativity, its innovative qualities, and its ability to bring about stability and order. The results presented here highlight the need for organizations to avoid casting shadow systems as the baddies. By doing so they may be destroying the one thing they need most: a mechanism that ensures their adaptability in an increasingly uncertain but competitive environment.

Back to Top

The Good

In the course of this investigation, interviewees identified many positive attributes of the Webfuse shadow system. All interviewees made some positive comments, regardless of their position, but the most vocal and passionate were Webfuse users. The positive attributes that were most consistently raised were: creativity surrounding the system; the perceived innovativeness of the system; and the stability and order brought about by the system.

Creativity is commonly thought to be associated with an individual creative genius. In this study, however, interviewees tended to reinforce observations made in previous research that creative works are actually collaborative systems.² Creative works are said to be not only novel (i.e., unique, original, unexpected) but also appropriate (i.e., useful and valuable)¹¹ – and it is the appropriateness of a creative work that enforces a collaborative system view.² In this investigation, Webfuse could be seen as a novel product in its own right, given its origins as a Web-based system developed in 1996 for the use of teaching and learning. Yet of much greater importance to the study participants than its novelty was that Webfuse fostered collaboration. As one administrative employee commented,

I remember talking to [a Webfuse developer] and saying how I was having these problems with uploading our final results into [the Enterprise Resource Planning (ERP) system] for the faculty. He basically said, "No problem, we can get our system to handle that"... and 'Hey presto!' there was this new piece of functionality added to the system. You felt really involved. You didn't feel as though you had to jump through hoops to get something done.

Although the creative value of Webfuse for the study participants can be attributed to many factors, two of the most significant were that it was appropriate and timely. Table 1 shows the drastic operational changes undergone by CQU. These changes have had many repercussions for workers at the coalface (academics and administrative staff) trying to manage and administer courses. Formal organizational system responses to these operational changes were, on the whole, very slow. Additionally, these responses were not tailored to the rapidly changing needs of workers. On the other hand, Webfuse responded quickly and appropriately to the teaching and learning needs of those involved with course management. One of the many reasons for this appropriateness and timeliness was the involvement of staff in the creative process of Webfuse. Indeed, the initial creator of Webfuse was fully aware of many of the new needs of coalface workers because he was an academic himself. As the system continued to develop, people's individual working needs were promptly integrated into the system.

This responsiveness was in stark contrast to the lack of responsiveness of the formal systems. As one of the managers directly responsible for both the implementation and the continued support of the central Enterprise Resource Planning (ERP) system said,

We just can't react in the same way that the Webfuse system can, we are dealing with a really large and complex ERP system. We also have to keep any changes to a minimum because of the fact that it is an ERP. I can see why users get so frustrated with the central system and our support of it. Sometimes, with all the processes we deal with it can take weeks, months, years and sometimes never to get a response back to the user.

Perceived innovative qualities. When interviewees explained why they preferred the Webfuse system over formal system solutions (see Table 1), their explanations tended to centre on what previous research has termed its "innovative characteristics"⁸. These characteristics as perceived by the interviewees are listed in Table 2 (below).

Past research also shows that when people believe an innovation has these qualities, they are more likely to adopt it.⁸ Accordingly, the perception that the Webfuse system was innovative might explain its exponential rate of adoption during the years 2001 through 2006 (refer Table 1).

Stability and order. Organizational researchers have suggested that social forms of shadow systems (those not incorporating technology) are a "self-organising network,"¹⁰ essential in bringing order out of chaotic conditions. Many interviewees confirmed these observations when discussing Webfuse. For example, one academic user elaborated on how Webfuse had helped reduce her anxiety at a very stressful time of year by enabling her to meet the deadline for the course results upload:

The main benefits for me as an academic was that it saved me so much time. Because I know that with my spreadsheet it's two clicks of a button and I look at the errors and if there's nothing there I don't have to do anything about it any more, whereas previously I had to go through a report and check each one [line] separately... agggghhhhh!

An administrative employee confirmed that the potential for error in using the central system alone was very high: "There is no easy way to make any sense of the reports that you get, no summaries of the statistics of how many passed or failed or what the errors are". She went on to outline the uncertainty introduced by checking for errors "by eye" in the central system reports:

You're pretty much reduced to going through the report line by line... checking for errors that can be rather unassuming... You might not think that a particular output on the report means that a student's result hasn't uploaded.

This process is time intensive and further complicated by the issue that some courses are very large – "2000 results are not uncommon." With Webfuse, anxiety is reduced because of the perception that human error is largely excluded.

Back to Top

The Bad

Interviewees from the senior management level (senior executive) and middle management within ITD raised various concerns about shadow systems in general. The majority of their concerns tended to centre on what the interviewees called the Hit by a Bus scenario. They explained this scenario as an event whereby something disastrous (such as being run over by a bus) happens to the key person or persons developing or maintaining the shadow system. Because of their informality these systems may have very little, or no, supporting documentation. In this situation those who rely on the system are left in an unworkable situation. There is no chance of immediate recovery if a key individual is lost and only a poor to non-existent chance of medium to longer-term recovery because of the absence of system documentation. The Hit by a Bus scenario represented the one major concern about shadow systems for management because of its paralysing potential.

Although management remarked that the Hit by a Bus scenario was a problem for many shadow systems, Webfuse appeared to be largely immune. While some systems, formal and informal, may be incapacitated by the loss of a few key individuals, Webfuse was able to reduce the risk significantly. It did this by ensuring system documentation was as complete and current as possible, and by making sure that no one person in the team had system-critical knowledge or abilities that weren't duplicated across other team members. Furthermore, there had been a gradual formalising of the system. For example, in the faculty documentation there was formal recognition of the team that supported Webfuse and each person on the team had a formal position description sanctioned by the university's Human Resources department. The gradual mainstreaming of the system was accompanied by a concerted effort on behalf of faculty management to capture the strategic value of the system, incorporating it back into business practice. For instance, from the very beginning, management encouraged the simultaneous development of the system along with the type of "business renewal,... flexibility, and encourage[ment of] learning" Ciborra³ discusses. This transformation, though, was not sustained, and nor did it progress to the "whole business,"3 for reasons perhaps best attributed to the bad and ugly aspects of the system, as listed below and in the next section.

Control. Like many organizations, CQU relies on the principles of classical management theory: that management is a process of planning, organization, command, coordination, and control.⁶ In the implementation of these principles at CQU, the Information Technology (IT) work is relegated to and controlled by an IT division, as mentioned above. The IT division is then subject to command and control by senior levels of management. For those interviewed in management and ITD, the biggest concern about Webfuse, was that the system existed outside the predefined structure that regulates and controls IT work. When individuals in management and ITD were asked to describe Webfuse and their experiences with it, they frequently referred to it as a "feral system" – a system that was "wild and out of control". Because Webfuse was not owned or controlled by ITD, it was said to be a "thorn in management's side". Nonetheless, a manager in ITD confided that if the division had been in control of Webfuse, it probably would not have been as successful as it had been. The manager elaborated on the difficulties that their division had, and continues to have, in dealing with stringent, heavyweight processes (such as their development methodology), and how this affected their ability as a division to adapt to changing circumstances. The division also had predetermined goals and this affected what they could and could not work on. Given these restrictions, the possibilities for a system like Webfuse to be even thought of were low to non-existent. It all depended on what the predetermined priorities were at the time – and when academics really needed a system like Webfuse, they were "not even on the priority radar". Therefore, the implementation of the classical management principles at CQU meant that a system like Webfuse would not have arisen out of the formal side of the organization - leaving many user requirements unfilled.

Stigma. Users and support staff of the Webfuse system often discussed the severe disapproval of ITD and senior management toward it. This disapproval appeared to stem from the informality of the system. A developer who has worked with both the shadow system and one of the central systems elaborated on the "unfortunate" perception of Webfuse as a shadow system: "If you ask me Webfuse isn't really a shadow system." The interviewee agreed that it does replicate some data and functionality but argued that

There really is no other system around that can do all of the things that Webfuse does... I think it's really unfortunate that Webfuse is perceived as a shadow system - why not a supplementary system or something else a little less damaging?.

Back to Top

The Ugly

Culture and politics form the ties that bind an organization together⁹. Therefore cultural and political forces are likely to impact on both the formal central systems and the shadow systems of an organization. Nevertheless, in this investigation shadow systems appear to be impacted more heavily by culture and politics than their formal counterparts.

Culture wars. The general stigma and misunderstanding surrounding the Webfuse system resulted in what may be described as a type of "factional war-fare."⁹ For some, using the Webfuse system was seen as a way of defying ITD and management. As one interviewee said,

Just because management somewhere decides that 'Yep, we've got this new beaut system and everyone's going to use it' – it's just rubbish. They don't understand how those decisions impact on us and the things we have to do to actually use the thing... Until these [formal central] systems do the things we need them to do we will continue to use this and other similar types of [shadow] systems to get the job done.

Although Webfuse was seen as a way to revolt against management, it also represented a weakness others could exploit in an attack. One instance of this was the events of August 2003. After receiving information that the Webfuse system was presenting inaccurate information, senior management ordered parts of the system to be shut down immediately. The Webfuse support team complied, even though many staff depended on these parts of the system for the day-to-day management of their courses. Only after senior academics in the faculties "kicked up a fuss" was it established that Web-fuse was not presenting inaccurate information; there had simply been a misunderstanding on the part of the person relaying the information to senior management. The decision was therefore revoked. One academic interviewee commented that: "We really needed that system and [neither] senior management nor ITD was going to provide us with anything like the functionality we get from Webfuse".

Politics, according to an academic user of the Webfuse system, is everything: "Everything's political... every system that's put up is political in some way". While this may be true, it appears that shadow systems are exposed to a greater depth and range of politics than formal systems. This has the effect of leaving shadow systems open to a large degree of direct and indirect manipulation.

Interviewees indicated that they were aware that certain senior management figures either actively supported or opposed Webfuse. Therefore, support for – and, indeed, the very existence of – Web-fuse depended on who was in power. In one specific instance, it was noted that when a senior person left the organization, the person who replaced them was hostile to Webfuse and immediately distanced themselves from the system. Interviewees believed that this eventually led to the transferral of the Webfuse system from the faculty to ITD.

A number of interviewees voiced their concerns about the transfer of Webfuse and its support team to ITD. The team are now viewed as being behind the "impenetrable wall" of ITD. One interviewee elaborated further, saying: "there are many more benefits that can be added to the system, but now that it is in ITD it is going to suffer from the same problems the corporate systems did in the first place: slow if any response time, and the ability for requests to go into the organizational ether". Paradoxically, while the move of Webfuse to ITD has made its existence as a shadow system questionable, its transformation into a formally sanctioned system seems even less likely. At the time of writing this article, members of the Webfuse support team have already been placed on to other projects – primarily because the teaching and learning needs of academics are not seen as a priority within ITD. Gradually, the support of Webfuse is being eroded, to a point where many believe the system will be unable to survive.

Back to Top

Conclusion

This analysis demonstrates that contrary to popular opinion, shadow systems can be positive sources of energy for organizations: a resource of creativity and innovation leading to order and stability. This confirms many of the observations made by organizational researchers of social shadow systems.^{5, 10} On the negative side, life is not necessarily easy for shadow systems or those working with them. Afflictions such as stigma and organizational politics can potentially paralyse these systems and marginalize those connected to them. From this case study, five important lessons for successful organizational use of shadow systems have been observed. They are listed in Table 3 (below).

By avoiding the desire to cast shadow systems as the baddies organizations might see these systems for what they are: potential diamonds in the rough. Shadow systems may be just what an organization needs – a mechanism that ensures their survival in an increasingly competitive and uncertain environment.

Back to Top

Back to Top

Back to Top

Back to Top

Back to top

The Digital Library is published by the Association for Computing Machinery. Copyright © 2009 ACM, Inc.