– Aligning Supply Chain Design for Boosting Resilience

I approached the literature review from five viewpoints. First, I considered how humans interpret the objects in their environment. The personal significance an object may hold was the second point of view. I then reflected why humans sometimes perceive objects as interacting with each other. For the fourth viewpoint, I deliberated how physical objects can represent, contain, and manipulate data. Finally, I studied systems in which physical objects are used to construct programs.

1.11.1 Objects and their meaning

Objects surround us and the study of how we interpret them is the research field called Semiotics.

Peirce (1935) developed a three-component model to describe the relation between the object, the individual, and the meaning. He referred to these as the representamen, interpreter, and semiotic object respectively. Saussure (1959) proposed an alternative model consisting of a signifier and the signified. He called the object being observed the signifier and the meaning that results he called the signified. In contrast to Peirce, Saussure did not explicitly link the resulting meaning to an individual.

Peirce’s model is useful when the individual’s background is relevant whereas Saussure’s is more compact.

Peirce puts it that an object does not carry meaning in itself but meaning is instead constructed in the observer’s subconscious mind based on his past experiences (Chandler 2007; Martin & Ringham 2000), experiences with the object (Fiske 1990) or experience with the object type (Palmer 1999).

Yet the meaning can change (Souza 2005) over time. The individual’s interpretation can also differ from other members of the same cultural group (Barthes 1982). The result is that even if the object has a common meaning within a cultural group, the individual may associate a different meaning with the object. This was my motivation for exploring how the user can make his objects and use them for programming.

Saussure’s model is useful when there is little risk that meaning will differ between individuals. For example, when used to describe the C (Kernighan, Ritchie & Ejeklint 1988) programming language where a word such as ‘while’ is very well defined and no ambiguity can exist. In contrast, Peirce’s model is useful when the research topic relates to how the meaning of words and objects can vary from one person to another. In this study, I used Saussure’s model to illustrate that a single concept can be expressed in more than one way whereas Peirce’s model was useful to show that the meaning conveyed by an object can vary across individuals.

Underkoffler and Ishii (1999) demonstrated that an object’s meaning can include attributes, nouns, and verbs. Oh, Deshmane, Li, Han, Stewart, Tsai, Xu and Oakley’s (2013) Digital Dream Lab objects

are examples of nouns, adjectives, and verbs. I identified that objects can also serve to mean a quantity, adjective, and an adverb. Finally, objects can represent discreet quantities or continuous values as in Dietz and Eidelson‘s (2009) glass.

1.11.2 Personally meaningful objects

Objects are most often designed by persons other than the user and consequently the meaning objects hold can vary according to the user. Design methodologies such as Druin’s (2002) do attempt to personalise the designed object but the result remains a compromise of all the participants’

inputs. Another result is that the meaning of an object may not be obvious and Patten (2005) reports on an instance where the user modified the appearance of an object to better represent a given concept for him. Krippendorff’s (1989) model illustrates the problem by highlighting that the user and the designer are often distinct persons. I addressed this problem by letting the user design personalised programming objects. To this end I found McCloud’s (1994) comment useful. He states that a simple object design is better suited than a feature-rich option when the objective is for the user to identify with it. From this, I deduced that personally meaningful objects are best made using materials that hold little intrinsic meaning. I therefore based my final artefacts on wooden blocks, dowels, clay, and paper.

Published research describes systems in which the user designs his objects. For example, Sanders (2000) created tools with which the user may fashion tangible representations of ideas while Sherman, Druin, Montemayor, Farber, Platner, Simms, Porteous, Alborzi, Best, Hammer, Kruskal, Matthews, Rhodes, Cosans and Lal (2001) report on having users create objects with which to interact with data. Story Room (Alborzi, Druin, Montemayor, Platner, Porteous, Sherman, Boltman, Taxén, Best, Hammer, Kruskal, Lal, Schwenn, Sumida, Wagner, et al. 2000), Quilt Snaps (Buechley n.d.), and Diorama Table (Oizumi, Mikami, Sasada & Ubukata 2007) do the same for programming environments. My research differs in that I explicitly incorporate Gestalt principles into the programming environment.

1.11.3 Gestalt

What we perceive is sometimes different to what our senses detect and Gestalt principles help describe this phenomenon. For example, when we observe two objects close together we tend to group them together (principle of grouping by proximity). In addition, when we drive along a long road and there appears a junction to the right, we do not consider the path to the right as a continuation of the current one. Gestaltists call this the principle of good continuation and I incorporated this principle in my artefact designs.

Objects are not always observed separate from their environment, including other objects. The way they are viewed along with other objects can be explained by Gestalt principles. We often consider the placement of one object relative to another as significant, for example in board games including chess. The arrangement can also affect the way a system behaves. Some HCI systems are based on the notion that the meaning an object holds depends on their positions relative to other objects (Gorbet & Orth 1997b; Marco 2011). Beckmann and Dey’s (2003) SiteView include examples of objects that do not hold individual value. The distance between objects can also hold meaning and Patten, Recht and Ishii (2006) use the distance between objects to represent a numerical value. Yet some objects can hold value independent of others. For example, Dietz and Eidelson‘s (2009) SurfaceWare drinking glass indicates a numerical value and Mazalek’s (2001) genie bottles can function on their own or be combined with others.

Two objects can be used together to modify data and Patten, Ishii, Hines and Pangaro (2001) and Patten et al. (2006) combine a data container with a rotating knob to modify digital data. Also, a physical object can be associated with its digital counterpart or another object by considering the position of objects relative to projected text, an area, or the other object (Oh et al. 2013; Patten et al. 2001). Associations like these are predictable using the Gestalt principle of grouping by proximity and my final artefact design is based on this property. Finally, physical gestures can also be used to group data associated with multiple objects and assign data from various sources to a single object (Merrill, Kalanithi & Maes 2007).

Gestalt principles can be identified in most tangible programming environments but none of the reviewed environments was designed explicitly with these principles is mind. Of the environments reviewed, most only incorporate one Gestalt principle. Only Story Room and Diorama table combine more than one principle with these being grouping by proximity and good continuation. My research explicitly consideres the relationships between objects and I use Gestalt principles to explain the relationships. Consequently, my work differs from others in that I make explicit reference to Gestalt principles to describe my programming environment.

1.11.4 Tangible objects and computing systems

As far as it concerns using physical objects to represent data or manipulate data, Ishii (2009) developed a model that both illustrates tangible objects as being distinct to digital data and also how one can represent the other in the digital and physical domains, respectively. According to the model, data can have both digital and physical representations. This is similar to a computer program that can take form as text on paper and also be a series of 1’s and 0’s in a computer’s electronic store. Ishii’s model therefore proposes that a dataset’s representation can vary according

to the domain in which it is applied. My research takes Ishii’s model one step further by demonstrating that a dataset’s physical representation can vary according to the individual. I also identified that Ishii’s model does not address the design of the physical representation. In contrast to Ishii’s model, my model highlights the origin of the object’s design.

Machine readable identification markings can help establish a link between the data and the object (Ullmer 2002). My literature analysis of how data and their representations are linked concluded that the identification may be classified as either electrically active or passive whereas the identification information can be transferred between the digital and physical worlds using mechanisms that are either tethered or untethered. I captured these options as a two dimensional model.

A common design approach is to develop functional objects without concern what meaning a user may attach to the object. Krippendorff (1989) proposes an alternative approach in which the form of the object also conveys meaning. My research is aligned with Krippendorff’s proposal and considers how individuals may choose diverse physical representations for data.

1.11.5 Tangible programming

Tangible programming environments rely on supporting technologies that range from active circuitry to passive sensing mechanisms. Passive sensing offer benefits that include a wider selection of materials to choose from, reduced cost, improved robustness, and more design options (Horn, Solovey & Jacob 2008). These benefits inspired me to design artefacts based on passive sensing technology.

Some tangible programming environments like SiteView (Beckmann & Dey 2003) dictate the object placement order along with their positions. These constraints are similar to those found in textual programming environments such as those developed for the C (Kernighan et al. 1988) language. In these environments the sequence in which symbols may be placed is often fixed and prescribed. To illustrate the point, consider the scenario in which a user assigns the numerical value of 10 to a symbol named “decade”. The correct sequence to do this is “decade = 10” and not “10 = decade”. I considered this an unnecessary constraint on the user and addressed it using the Gestalt principle of grouping by proximity.

SiteView places similar constraints on the user. For example, when the user wants to indicate the three conditions ”rain”, morning”, and “Monday” in a program rule he has to place each representative objects at prescribed positions. I argue that these conditions represent distinct data types (being the weather state, time of day, and day of the week). I developed software that can

make this distinction without human intervention and interpret the objects appropriately. The result is that the three conditions in this example can now equally well be placed in different sequences including the two arrangements “Monday, morning, rain” and “rain, morning, Monday”. From this, I concluded that the symbol sequence is not important; instead, it is the combination of words that should be considered. Therefore, as long as the user places the words in close proximity to each other he does not have to be concerned about their order. This is another example of the Gestalt grouping by proximity principle.

Programming language designs often include the option for the user to determine how actions and parameters are represented. Yet, with these systems, the user has to base his designs on language elements such as parameters and actions previously determined by the language designer. For example, in the C (Kernighan et al. 1988) language, the user can choose to use the sign ”decade” to represent the quantity 10 and does this using the symbol sequence “#define decade 10”. Some tangible programming environments also include this feature (along with its limitation) and examples are Story Room (Alborzi et al. 2000), Quilt Snaps (Buechley n.d.), and Diorama Table (Oizumi et al. 2007). In Story Rooms, the user is constrained to using the designer’s physical signs but has the choice to combine these with personally meaningful objects. With Quilt Snaps, the user can embellish squares yet the size and shape are determined by the system designer. Finally, when choosing his objects for the Diorama Table, the user has to keep in mind that the way the shapes will be interpreted by the system has also been predetermined by another person. It is, therefore, possible for the user to choose what programming signs look like but this must be done within constraints determined by a system designer. I considered a different approach and demonstrated a system in which the user is free to choose the physical representations of program actions and parameters.