Skip to main content
photo
 
Dynamic Model of Opening Data

Public managers face challenges when implementing open data initiatives. These challenges arise due to the multiple interactions between actors, information flows, technologies, and interests. By defining the problem of opening government data dynamically, we are placing emphasis on how processes and relationships change over time. The two open data cases clearly demonstrated how rapid and unpredictable technological developments and shifting relationships in the social and organizational environments change over time. This section illustrates how modeling the non-linear dynamics of opening government data systems supports decision making, learning, and understanding in a complex, unpredictable world.

We begin by describing a very simple mental model of opening government data as it is described in both cases. Open data initiatives are frequently described as virtuous cycles, or reinforcing loops. The logic of a virtuous cycle is that, if left unimpeded, it can generate exponential growth or decay. In the case of opening government data, advocates assume that simply supplying more and more data sets freely and in more formats will lead to more and more use. In such a mental model data use leads to value creation, which in turn will motivate government to make more data open and accessible. This reinforcing loop leads to some form of exponential growth in supply and use represented by the solid lines in both graphs in Figure 7. What we saw in our cases, is a different mental model. In our complex model, the expansion of freely available data sets and use are constrained by agency and user capabilities, data management practices of agencies, agency effort, politics, interactions between citizens and data that create meaning confl ict, and relationships with citizens and other stakeholders. Over time, these constraints are activated and the result is a set of negative or balancing feedback loops that tend to slow the supply of data and use or reduce it all together.

Although each of our illustrative cases involves intense use of the open data by citizens and the involvement of several stakeholders, anecdotal evidence suggests that the majority of available open data initiatives do not enjoy such success. Most open data strategies look for ‘quick wins’ in the first few years, but over time, the available set of data that is easily opened will diminish, reducing the number of open data sets folks are looking to use, which will result in a loss of interest in use, less stories of valuable use, and the virtuous cycle slows. In fact, it is more common that data use follows a pattern of behavior more similar to the dashed line in Figure 7 (a). Our two examples may be the exception rather than the norm when considering the value creation of opening government data. Likewise, as we saw in the restaurant inspection case, meaning confl ict among citizens attempting to use the information counterbalanced the virtuous cycle and actually shut down (for a brief time period) the release of data (Figure 7 (b) shows this impact). The model that we describe in the following paragraphs is a conceptual attempt to explain these patterns of behavior using stocks and flows.

Fig. 7
Figure 7. Dynamics of opening government data.
Fig.7b. Dynamics of opne government data

A Causal Map to Frame Open Data Initiatives.

System dynamics is one modeling approach that can assist in uncovering the complexity of open data initiatives. This approach uses causal maps to visualize a systems structure and behavior. The basic building blocks of a causal map are stocks, flows and feedback loops. Stocks, represented by ‘boxes’, are any entity that accumulates or depletes over time. Flows, represented by ‘valves’, are the rate at which the stocks change. A variety of factors contribute to the rate at which a stock changes over time. A feedback loop exists when information resulting from some action within the system (endogenous) travels through the system and eventually returns in some form to its point of origin and potentially influences future action. A loop can be reinforcing or balancing. If the tendency of the loop is to reinforce the initial action, the loop is called a positive or reinforcing feedback loop. Reinforcing loops are sources of exponential growth or collapse. When positive, they are considered a virtuous cycle. If the tendency of the loop is to oppose the initial action, it can be thought of counteracting or constraining the reinforcing loop which balances or prevents change from happening. The model presented is a partial explanation of both cases and presents only a small set of important casual relationships and feedback processes. It is not a fully developed simulation mode

Making government information available.The basic story in both cases starts with opening government data—making available to the public the information about restaurant inspections in New York City and street construction projects data in Edmonton. Figure 8 shows a conceptual representation of this process. The box Government Informationrepresents the accumulation of government records created from government activity

Fig. 8
Figure 8. Making government information available.
All data in this accumulation becomes candidate data to be opened to the public. The second box in the fi gure—Open government information—represents the accumulation of all open data available to the public and the valve ‘Opening information’ represents the activities necessary to make available such information. Opening informationadds to the accumulation of available open government information over time. To make this happen, governments need to allocate some effort to opening information. Every (person*hour) of effort varies on how effective the person is, which refl ects that the most experienced people will be able to open more information with the same effort. In our two cases, agencies are putting effort into restaurant inspection and street construction projects data in order to move Government Informationinto the box Open government information.

As it is shown in Figure 8, on the one hand, agencies’ efforts to make information available may be increased or decreased by political or legal requirements. On the other, new technical developments will contribute to people’s effectiveness in making this information available. In the restaurant inspection case, for example, the Internet as a new technology was an important trigger for making the information available in the fi rst place. In the Edmonton case, it is clearer that new policy related to open government constitutes the main motivation to make the street construction projects data set available. Of course, the expectation of governments is to create public value by making information available. That is to say, making restaurant inspection information available to the public creates value by informing the public about their health safety when eating in a restaurant by ensuring basic requirements of hygiene in each establishment. Making street construction projects data available, on the other hand, creates value by helping people better plan their routes when driving from one place to another or for taking into consideration increased commuting times

Making government information 'fit-for-reuse.'

Government information has been available to citizens long before the Internet or open data initiatives. However, the effort needed by citizens to physically get this data has been reduced over time. First, the Internet made it easier to post and access and second, recent platforms and format changes make machine-readable data more fi t for re-use in different applications

Fig. 9
Figure 9. Making open information fit for re-use.
Figure 9 (page 23) represents these changes in technological ease over time by adding a second set of stocks, Government information fi tness to re-use, and Open information fi tness to re-use. The boxes in the fi gure represent the way in which the characteristics of information have been changing over time. For example, in the case of restaurant inspection information, prior to the Internet, it was not very easy to re-use the signs that displayed the information at each restaurant location. However, by placing a variety of related information together in HTML or PDF formats in a single location on a website reduced the effort of gathering this information. But, a citizen would still need to print, re-type it or pre-process it before being able to re-use it. Today’s tools make machine-readable formats quite easy to re-use and as a result, new applications are developed to encourage mobile use of the information.

The valves Archiving informationand Making information easy to re-useare fed by the agency activities needed to make such information available in any format. The release of street construction projects data offers insight into agency processes for archiving and making it fit for reuse. In Edmonton, there are a variety of candidate data sets to be made available to the public. When trying to balance resources, time, and effort, choosing which data to pay attention to was not an easy task. Interviewees from the case commented that this particular data set was made available, at least partially, because of the commitment of the data set owners. They considered commitment from data owners as a key factor for success. Additionally, the current data management practices of the department that owned the street construction projects data enhanced the commitment from data owners and increased the effectiveness of making data more fit to re-use. Good data management practices will reduce the cost and effort of making information available and increase the probability the data will be opened and easy to re-use. On the other hand, poor data management practices will increase the cost and effort required to open data and make it available in machinereadable formats.

Another important aspect is the quality of data management practices. Good practices involve providing excellent metadata suitable for the purposes of opening government data. The developer of the road construction application in Edmonton described how the excellent quality of DOT’s metadata for this particular data made it easier for him to, first, imagine what kind of application he could build and second, to make a quick assessment that the development of this app would take approximately 30-40 person hours. These decision points were very important in his analysis of whether or not to build an application.

Making information more fit to re-use requires agencies to allocate some effort to the process and it is likely that agencies will vary in their levels of effectiveness in trying to accomplish this task. As it is shown in Figure 9, the amount of effort to prepare the open information also depends, at least partially, on political and legal requirements. The agency’s effectiveness in providing open information fit for re-use also depends on technical developments. The development of XML, for example, makes it easier to prepare information to be machine readable and US President Obama’s Open Government Directive(a political and legal requirement) has clearly signaled to US agencies that they must increase effort to open information and make it more fit to re-use.

Fig. 10
Figure 10. Contextualizing open government information and creating value.
Contextualizing open government information and creating value It is not enough to focus only on the technical components of opening government data, strategies must also consider the social aspects of information more generally, particularly providing suffi cient context for information use. The effort agencies make to contextualize the information for use among diverse audiences and users is important. The box Information Contextis a third accumulation (see Figure 10). As with the other boxes discussed so far, adding context to data requires effort by the agency and capability to be developed. Providing additional context makes the data more fit for useby various audiences and users, which contributes to public value creation by increasing the value of the information.

The dynamics of providing context are often not addressed by agencies when designing open data initiatives. Context is closely related to creating public value for specific stakeholders, purposes, and applications. Since technical developments do not help to improve context, it may partially explain why the availability of so many open government data sets has not generated the uptake of use first envisioned.

Figures 10 and 11 also show for the first time two possible reinforcing (virtuous) feedback loops labeled as ‘R1’ and ‘R2’. As noted earlier, a reinforcing loop is a virtuous cycle that contributes to exponential growth or decline in public value, but over time, constraints are engaged. A lot of agency effort is spent creating data that is machinereadable but it is not contextualized in a way that might generate value. Thus, stakeholder involvement is a way to increase the effectiveness of contextualizing information. For example, restaurant and consumer associations and citizens could participate in the process of agreeing on types of data to be opened and ways to present this data in order to create value. However, as we mentioned before, reinforcing processes can represent an initial trap. It is hard in the beginning to get stakeholder involvement because they are uncertain of the value of the information.

Fig. 11
Figure 11. Potential meaning confl ict in opening government data projects.
Conflict of meaning Some constraints even have the potential to shut down an initiative (see Figure 11). In the case of restaurant inspection data, when the information was initially released it was made available in exactly the same format to the public as its primary users (e.g., inspectors, restaurant owners). Less of an emphasis was placed on potential new users such as city visitors or citizens. The lack of information context (i.e., releasing technical language like vermin) created a conflict in meaning, or misunderstanding of the underlying or intended data element. This confl ict of meaning triggered two other feedback balancing loops labeled in the fi gure as B1 and B2. In this case, confl ict of meaning for some data elements created negative pressure to hide public data, reducing the effort by agencies to make the data available or even forcing the political/legal areas to create safeguards to eliminate easier public access to the already public data (see B1). On the other hand, the same confl ict of meaning may trigger positive pressure to contextualize the information, increasing agencies’ efforts and potentially improving the quality of the information that will lead to public value (process B2).

Fig. 12
Figure 12. Application development and value creation.
In the street construction projects case, we see what happens when a feedback loop is dormant (B1). That is to say, the agency provided the information in the same way that the primary/users see the information and did not provide any additional context. Releasing the data in this way did not create meaning conflict or create any pressure to hide the information. However, both other feedback loops remained active (R1 and B2). As noted in the case, a variety of new stakeholders are encouraging the agency to make more frequent updates to the data or provide additional data fields that will improve the value of the information for their new and intended uses

Developing apps and creating value.Finally, both cases show that public value creation from opening the data is increased by the development of mobile or Web applications as seen in the reinforcing loops R3 and R4 in Figure 12. Both loops contribute to value creation bymaking information more useful to more audiences/users. R4 describes how the quality of the data set for re-use incentivizes the application developer to create an app. In the Edmonton case, the relevance of the data, how easy it was to use, and the quality of the metadata made his decision easy. His personal motivations, including that he was new to the area and recently attended a conference of fellow civic hackers, also contributed. R3 describes the social components that support use through applications, which are not just for the few that are capable in fi guring out how to use machine-readable structures. This was evident in the positive reviews of the apps. While we do not have direct evidence of citizen use of the mobile apps in either case, it is presumed that the mobile apps provide additional flexibility and value.