72 Identifying Requirements for an Organizing System
The two parts of the definition of an organizing system explicitly suggest two categories of requirements, those that specify the intentional arrangement of the resources and those that specify the interactions with the resources. These categories of requirements both depend on resource descriptions, which are implied by but not explicitly called out in the definition of an organizing system.
Because description, arrangement, and interaction are interrelated it is impossible to describe them separately without some redundancy. Nevertheless, in this book we have done that on purpose because taking different perspectives on organizing systems in Chapters 2-10 has enabled us to introduce a broad range of concepts, issues, and methods:
Every organizing system must enable users to interact with its collection of resources (Chapters 3 and 10);
The possible interactions depend primarily on the nature and extent of the descriptions associated with the resources (Chapters 4, 5 and 6);
Intentional arrangement emerges when one or more resource descriptions are used by organizing principles (Chapters 7 and 8);
Different implementations of the same organizing principle can determine the efficiency or effectiveness of the interactions it enables. (Chapter 10).
If you are creating a personal organizing system or otherwise small-scale one with only a small number of users, you might think there is little reason to think explicitly about requirements. However, any project benefits from the discipline of being more systematic about its purposes and their priorities. In addition, being explicit about requirements enables traceability and impact analysis. Traceability means being able to relate an interaction or feature of a system to the requirement it satisfies; impact analysis runs the causal link between requirements and features in the opposite direction to assess what or who will be affected if requirements change.
Requirements for Interactions
When we describe interactions in a generic or broad way as we did in Activities in Organizing Systems we see that all organizing systems have some common interactions, but most of the time we want to pay attention to the more specific interactions that are designed to create value in a particular organizing system because of the kind of resources it contains (“Interaction and Value Creation”). The domain, scope, and scale of the organizing system determines which interactions are possible and which ones must be explicitly supported, but the priorities of different interactions are more often determined by decisions about intended users. (See “Number and Nature of Users”.)
For most organizing systems other than personal ones, the set of interactions that are implemented in an organizing system is strongly determined by business model considerations, funding levels, or other economic factors. For-profit firms often differentiate themselves by the number and quality of the interactions they support with their resources, some by supporting many of them and some by supporting a minimal number. This differentiation is strongly shaped by and also shapes user preferences; some people prefer self-service or unmediated interactions, while others prefer full service and mediated interactions. Non-profit institutions like public libraries and museums are also subject to these constraints, but unfortunately they have fewer options for adjusting service levels or changing their targeted user populations when their funding is reduced.
Some requirements for interactions come along with technology requirements, to have resources in a particular format, to conform to a particular specification or standard in order to operate in some technology environment, or to interoperate with other parties or their organizing systems.
An essential requirement in every organizing system is ensuring that the supported interactions can be discovered and invoked by their intended users. In organizing systems with physical resources, good designers enhance the inherent affordances of resources with navigation and orientation aids that direct users to points of interactions (“Affordance and Capability”). With digital resources and information-intensive organizing systems, interactions are not immediately perceivable, and poor design can create overly complicated user interfaces in which many interactions are never discovered and thus never used.
It is tricky to compare the overall capabilities of organizing systems in terms of the number or variety of their interactions because what matters more is how much value they create. Organizing systems with active resources can create value on their own without an explicit user interaction (“Active or Operant Resources”). Other organizing systems exploit stored, computed, or contextual information to create value by eliminating the need for user interactions, such as location-based smartphone apps that push information to you when you are near some particular location or some person you know (“Affordance and Capability”).
About the Nature and Extent of Resource Description
Interactions with resources within an organizing system often depend on descriptions of individual resources or descriptions of the collections that contain them. In the bibliographic domain, generic or common interactions make use of descriptions that can be associated with almost any type of resource, such as the name, creator, and creation date.
For example, any resource with a sortable name or identifier can be arranged alphabetically to enable it to be easily found, and any resource with a creation date can be discovered by a “what’s new” query to a resource collection.
Different types of resources must have differentiating properties, otherwise there would be no reason to distinguish them as different types. These resource properties can be recorded in the terms of a description language to support one or more interactions or to answer one or more questions. Simply put, choices about the nature and extent of resource description depend on which interactions or questions are most frequent or important (“Describing Instances or Describing Collections”). If a particular property of a resource has no interactions that depend on it, there is no need to describe it. However, if an interaction depends on a description of a particular resource property, a missing description or one of inadequate precision and granularity means that the interaction will be impossible or inefficient to carry out because the resource will need to be further analyzed to create or extract the required description. An ISBN is a sufficient description to find a book in a directory, but if the ISBN is the only description associated with the book you will not be able to tell who wrote it. The tradeoffs imposed by the extent and timing of resource description have been a recurring theme in this book, with the tradeoff between recall and precision being the most salient (“When Is It Being Organized?”, “Affordance and Capability”, “Category Abstraction and Granularity”, “The Recall / Precision Tradeoff”).
The properties of resources that are easiest to describe are not always the most useful ones, especially for information resources. Anyone can determine the number of pages in a book, but often only a skilled cataloger can accurately describe what the book is about, a far more important property. (““Description” as an Inclusive Term” and “The Limits of Property-Based Categorization”) For non-text information resources this problem is magnified because the content is often in a semantically opaque format that it optimized for the devices that creates and processes it but which cannot usefully be analyzed by people. (“The Semantic Gap” and “Describing Non-text Resources”)
Business strategy and economics strongly influence the extent of resource description. In many museums and archives there are not enough trained people and time to describe every pottery fragment or document, and many resources are described only at an aggregate level. In contrast, some people argue that the explosion of content in physical and digital form mandates significant investment in descriptions that facilitate resource discovery in a crowded marketplace.
Automated and computerized processes can create the resource descriptions in an organizing system and their use is primarily driven by scale (“Automated and Computational Resource Description”). Search engines index web pages and analyze their link structures because it would be impossible to treat the web as a traditional library and organize it by human effort. The benefits of digital cameras, video recorders, and similar devices would be far fewer if people had to manually identify and describe each resource when creating it. Instead, these devices can automatically assign some contextual metadata. Similarly, competitive pressures on vendors to provide real-time and context-sensitive information services mandate automated collection of contextual information like location from mobile phones, portable book readers and tablet computers.
We might seek some optimal degree of description given some set of requirements or purposes for an organizing system and some estimate of the organizing effort that could be applied; in practice this is elusive for two reasons, both relating to scope and scale. First, as the number of users of an organizing system increases, it becomes more difficult to identify and anticipate all its possible purposes and constraints it must satisfy. Even if most users share the goals for the organizing system, any particular user might have some additional specialized use for some attributes or relationships that would require more description to satisfy.
Second, even if it were possible to implement some optimal degree of description in a particular organizing system, we would still encounter problems when multiple organizing systems exist in the same domain or in domains that intersect across context, device, or application boundaries. Since organizing systems are designed and evolve to satisfy the specific requirements of their particular context, companies will often describe the same resources differently, which creates integration and interoperability problems when companies need to exchange and combine information resources (“Transforming Resources for Interactions”).
About Intentional Arrangement
Organizing principles depend on resource descriptions, so requirements for the former are always intertwined with those for the latter. Specifying requirements for the intentional arrangement of resources is analogous to specifying why and how resource categories can be created (“Principles for Creating Categories”). In turn, the creation of resource categories often becomes a question about the number and kind of resource properties that might be analyzed and exploited by organizing principles.
We noted that there is a continuum of category formation that ranges from minimal use of resource properties to more rigorous use of multiple properties, and finally to statistical or composite use of multiple properties, some of which are induced or inferred rather than explicit. The simplest principle for defining a category is by enumeration, just putting the resources into a set without any specification of any properties they might share. The enumerated resources might very well have common properties, but the principle of enumeration ignores them; the only property that matters for that principle is that the resources are in the same set. This corresponds to the simplest principle of intentional arrangement, that of collocation, just putting the resources in the same location without any additional organization.
Collocated resources often acquire some additional arrangement as a result of their use; consider how the books, papers, or other resources gathered for some writing project often end up in piles in your office or on your desk close to your work area. For a small collection, the proximity-to-use organizing principle is the easiest way to satisfy a requirement to minimize the time to find frequently used resources.
As we have often seen, the scope and scale of the organizing system is a dominant design consideration and it applies to principles of resource arrangement too. The collocation principle of arrangement is sufficient for small resource collections because it is not necessary to define the optimal organization if the time to find any particular resource is short even for an inefficient search method of scanning the entire collection. Using the extrinsic property of frequency of use makes search slightly more efficient, but only in organizing systems where the user population is small or interacts with the resources in similar ways. Otherwise, arranging resources to facilitate their frequent access for some users would hinder other users who never use them. Imagine if you shared your office desk with someone who works all night on other writing projects and leaves his frequently used resources in piles close to his work area—which becomes your work area in the morning.
Larger resource collections usually require multiple organizing principles to manage the complexity that emerges when more users and more varied interactions must be supported. A valet parking lot might organize cars only by size to make optimal use of limited space when parking and fetching them, but when cars are organized for sale they would be organized by price, performance, seating capacity, manufacturer, and many other properties. It is essential to establish the priority of users and interactions because these requirements determine the order in which the principles are applied to arrange the resources. This ordering creates a logical resource hierarchy that affects the efficiency of interactions and the maintenance of the organizing system over time.
Information resources are invariably challenging to arrange because their aboutness is not an easily perceived property and because of the open-ended purposes they can serve. Information collections with broad scope most often use a standard system of bibliographic classification (“Bibliographic Classification”). In contrast, special libraries have narrower collections that need to support domain-specific interactions for a relatively small set of users, and as a result they require more specialized organizational schemes. The principles for resource arrangement in large firms of every type are often required to conform with laws and regulations for accounting, taxes, human resources, data retention and non retention, access control, and other functions. (See “Governance in Business Organizing Systems”, “Mandated Classifications”)
Dealing with Conflicting Requirements
Any individual, group, or enterprise can create an organizing system that meets their specific requirements, but once this organizing system involves two or more parties with different requirements, there is a potential for conflict. Roommates or spouses sometimes argue about how to organize items in the kitchen, in the refrigerator, or in some other shared space. To a person who arranges spices alphabetically and condiment jars by size, arranging them according to cuisine or frequency of use makes no sense. Similarly, if you are the sole user of a Dropbox or other cloud storage account, you can organize it any way you want. You can use any number of folders that need only make sense to you, or you can leave everything unorganized in a single folder. However, if you share the Dropbox account with another person, they are likely to have different organizational needs or preferences. Perhaps you tend to organize resources by file type, while they prefer to organize resources by topic or project.
A small number of people can often agree on an organizing system that meets the needs of each participant through informal negotiations. The potential for conflict increases when more people are involved, and “bottom-up” ad hoc negotiations to resolve every disagreement between every pair of participants just are not feasible. In many domains conflicts are avoided or suppressed because the parties have developed or agreed to conform to standards (“Classification and Standardization”). Nevertheless, conflicts in organizing principles for large-scale organizing systems are often resolved by parties with the legal authority or economic power to impose a solution on all the participants in a “top-down” manner.
An easy to remember framework for prioritizing requirements is MoSCoW, which classifies them as Must, Should, Could, and Won’t (Desoky 2010). (Winkler and Pilgrim 2010) is a comprehensive review of academic research and best practices for requirements traceability.
“Customer segments” or “customer models” are well-established constructs in product and service marketing and operations (Batt 2000) (Zeithami, Rust, and Lemon 2001). They are key parts of strategies for acquiring customers, increasing market share, and retaining customers. Customer segments can be identified using numerous overlapping criteria, including demographic variables, product or behavior choices, and preferred interaction locations or channels. For example, an airline might segment its customers according to their age, gender, home airport, ticketing class, and travel frequency.
Funding cuts for public libraries lead to reduced staffing, reduced hours, and reduced acquisitions and many of them serve populations facing economic challenges of their own. (Johnson 2010).
Organizing systems differ in the extent they can initiate interactions or use information to make them unnecessary. In libraries the organizing systems are typically designed not to preserve user activity records longer than absolutely necessary; in commercial organizing systems, user activity records are the basis of business processes that create highly detailed user models (called “microsegments” or “microcategories”) that enable personalized product and service offerings See (Taylor and Raden 2007), (Rosen 2012).
The Dublin Core was proposed in 1995 as a small vocabulary with 15 common elements that could be broadly applied. The emergence of many specialized derivatives of the Dublin Core since then illustrates the inherent tension between the simplicity of using a small set of common descriptive elements and the precision enabled by a large or more domain specific vocabulary.
For example, we have often used a home kitchen as a setting for organizing systems. Suppose the home kitchen is to be used as the set for a cooking show and the designers want to arrange cookbooks to make the background visually pleasing. The designers would like to search for cookbooks on the basis of size and spine color, but these descriptive elements would be of little value to other users.
The category of special libraries includes law libraries, corporate libraries—both those that support the head office and the research organization—medical libraries, military libraries, museum libraries, prison libraries, and might even be stretched to include libraries of software components.
Some people call this the “Wal-Mart approach” to standardization. A firm with dominant market power does not need to negotiate standards because it can impose whatever standards it chooses on its partners as a condition of doing business with them. When there are conflicting requirements, different relationships within the set of participants trying to reach agreement, and different extents to which they are subject to the authority behind the desired agreement, it is not surprising that approaches “that require perfect coordination and altruism are of no practical interest” (Rosenthal et al. 2004, p 47).