Why does the model matter?
Hickey: “When you understand the various models and you understand the differences between them, and you understand how you can mix and match them, you begin to realize that perhaps you have more options than you initially realized. As Bill was describing, these models provide that framework or that structure to evaluate the efficiency and effectiveness of your current data programs. They can also help clarify the tradeoffs in choosing one approach over another. No model is without its pros and cons and it’s good to go into those with eyes wide open… The factors that are considered in each are the cost, who shoulders the cost under each of these different models? Which might be most cost effective for which types of data layers? Authority certainly, who is the authoritative data source? Who’s funding and who has that funding? To what degree does local control impact the flow of data in these models?”
When working with a large variety of stakeholders and data, there is no one right answer for how to model a statewide data program. Various circumstances can impact what model should be used and for what purposes. Different data models will perform differently, but that’s okay; the key is matching the right model to the right data program. To find the correct model requires answering a number of questions about where authority, funding, and control should reside within the program to get the best results. Regardless of the model used, being able to communicate expectations and requirements of your model to other stakeholders is essential. The model can provide a valuable framework for local governments and state agencies to cooperate under, but only if they are properly informed.
Johnson: “In a decentralized model decisions happen locally, the authority remains local, the funding remains local. Local here doesn’t necessarily mean ‘local government.’ It could be that you’re talking about executive branch state agencies, each acting independently. Don’t confuse local with a level of government here, we’re talking about the independent autonomy of agencies working on their own. The decentralized approach is quite challenging to be successful with it, because it requires everyone on their own to do things consistently, and that generally doesn’t happen.“
The decentralized model can be thought of as the default model when there is no state effort to coordinate agencies or local government. This can occur when there is no active incentive for the state to force these actors to work cooperatively with each other, or when there are barriers that inhibit coordination. Each actor is focused on meeting its own needs, rather than working towards shared goals. Shifting away from a decentralized model can be difficult as local stakeholders may fear losing power or funding if their local control is altered. However, when there is sufficient economic incentive, states can shift from decentralized towards federated or centralized statewide data models, as they have begun to do with addresses and parcels.
While decentralized models can work, they are reliant on high levels of communication and coordination between local stakeholders. Most of the time, this cooperation is extremely difficult to achieve. Different stakeholders have different levels of funding, expertise, training, and ability which can complicate cooperation and result in significant gaps in work between stakeholders.
Hickey: “In a centralized model there is a single data source that’s been identified and that is the benefit there to ensure consistent authoritative data. Kind of like the centralized factory and that revolutionary assembly line of the Ford Motor days. It’s really an economies of scale model, it’s all about efficiency here. In terms of authority and command structure, this is absolutely a top down model that is required to make this work. The authority really cannot be shared, not effectively at least. A GIO might get input from others but the decision and the operation resides in that one entity. And of course, not surprisingly, the funding determines who has that authority… Often what we see in these centralized approaches is that there is a strong, sometimes urgent business driver. It might be something like Next-Gen 9-1-1 or broadband mapping or even more recently the COVID case mapping throughout a state.”
The main advantage of a centralized approach is the guarantee of coordination and standardization of the data produced. This coordination can vastly increase efficiency by reducing duplication of effort and streamlining the process of data management. However, centralized models can be extremely difficult to create. Compared to a decentralized model, which can be seen as a default, the top down approach of the centralized model only results from a dedicated and intentional effort. To succeed, business drivers must provide incentives to convince local stakeholders to replace their existing systems.
The issue for local stakeholders is that centralized models may involve giving up control or even funding to a higher authority. Offering incentives to counterbalance these losses can be a powerful tool, but the incentives that are attractive to local actors can vary wildly. Some actors might want a replacement for the funding they might lose, additional training for their staff, or other kinds of assistance. Luckily, incentives can come from a variety of places such as federal grants or mandates, non profit organizations, political prioritization for a centralized project, or even general economic development in a local area. Determining what local stakeholders need requires a working relationship and constant communication.