by Johanna Drucker
Read the second chapter: Interpreting Visualization (and vice versa).
Use the tag “R2” when you post your assessment of the text’s message and the questions it raises.
by Johanna Drucker
Read the second chapter: Interpreting Visualization (and vice versa).
Use the tag “R2” when you post your assessment of the text’s message and the questions it raises.
In this chapter Drucker expands the distinction that she made in the first pages of the book by comparing Athanasius Kircher's visualization with Ramon Lull's "great art of knowing". For Dr. Druker, I believe, the main argument will be that, even if we don't realize it anymore (mainly because by their continuous usage across time), the majority of visual (or data) representations are acts of interpretation.
To prove her point she provides us with different kinds of examples such as maps or charts generated in spread sheets.
There is a point that she makes about Snow's chart of deaths from cholera and the way it could be done again in order to express another dimension that includes some emotional landscape. While this might be highly criticized I would agree completely.
I kept thinking about the many conflicts that exist around the world and the thousands of people that are being affected by them and the way they are represented in graphs and charts across universities, governments and media outlets. Those numbers fail to represent "the point of view of a mother of six young children"."
It might seem imposible, or even, ludicrous to attempt a re examination of data visualizations from a humanistic perspective but I believe it is worth trying.
On Interpreting visualization, notes:
Representational versus knowledge generating and the ways in which calculations and dynamic information create new relationship and thus new interpretations as opposed to static information.
Driving influencers, industry, record keeping and observation.
What is the most humanistic visual form?
Mapping and charting of scientific and philosophical "belief systems" during the late Rennaissance and early Enlightenment periods are indicators of how
visual forms become essential to intellectual inquiry.
Most interesting are the explorations of mapping hierarchy and how structure an dform lend themselves to either establish an organic holistic one, or the ways in which structure and slection create transformative hierchies based on data sets and behavior.
Structure, style, form, calculation and transformation, lend themselves to the final interpretation of visual forms, of which both creator and observer are active participants.
Continuity implies and organic transition whereas graphical linear interfaces- of any medium- create contrasting sensations that infers itself on the information being observed. The implicit binary relationship of contrasting forms can also be interpreted as holistic and complementary, a ying yang of elements or balancing of the scales. Which both informs the work, and the observer.
How to map and chart, holistically, continually, dynamic changing systems, whether from time, spatial, or user-generated data, and account for possibility and uncertainty within prediction?
*mapping dynamic data, interpreting uncertainty
I was very excited by the last two sections of the chapter discussing the constructed nature of data and thereby it's visualizations. In my mind the question of a humanistic approach to data viz must reflect the fuzzy nature of our senses and the reality we construct. Appealing to a viewer's imagination, how they picture abstract content and attempting to replicate something that reflects the natural means of computation and virtualization we all possess. I've often thought about what a phone UI would look like if it weren't bound to the screen size and resolution. If this extension of our intellect and communicative ability was accessible in an organic alternative. If the needs of graphic differentiation between two subjects didn't need to be explicit but was intuited through some other means.
The images of the amalgamated real estate photos and productions of Google's DeepMind understanding of what a dog or a house looked like swirl about my head. A visual representation of uncertainty.
I wonder too if by incorporating the user themselves to define graphically the elements of a visualization, through association and freehand drawing, could we assemble a final product, made in part by the user themselves that introduces the concept of the fragility of what is being presented. It's ability to be distorted or skewed not only by the data but by the audience themselves.
As Flavio pointed out in his presentation on Giorgia Lupi, there's a line to walk between embellishing a visualization with flourishes that can further draw in the user, or distract and upset the objectivity of a data point. I think the same holds true if transferred to the idea of appealing to the minds eye, producing graphics that feel organic and familiar at the expense of pinpoint accuracy.
In this chapter, Drucker builds on a selection of ideas she presented in her Introduction and Chapter One. She begins the Introduction by comparing two images, one which is a “representation of knowledge” and another which is a “knowledge generator.” She begins Chapter Two by discussing this distinction in more detail. Toward that end, she provides numerous historical examples of “knowledge generators” and explains how each accomplishes that as opposed to simply displaying information.
At the core of her idea of a “knowledge generator,” however, is the act of interpretation, or, at the very least, some sort of “reader” involvement. She explains, for example, how an object like a train time table (or a map, for the same reason) generates knowledge because it makes combinatoric calculation possible, it displays information that allows a reader to chart an individualized itinerary, thereby creating new knowledge not explicitly displayed. She also explains how even these static objects, which see, to display objective data, are really just objects of interpretation of their creators. A world map, for example, necessitates the projection of the three-dimensional spherical world onto a two-dimensional flat plane. Such a mapping necessarily creates some sort of aberration in the final two-dimensional product, and the choice of aberration is the interpretive act of its creator. She mentions the commonly used Mercator projection, which preserves the relative arrangement of landmasses but distorts their size near the Earth’s poles. Reading such a map as an objective depiction of the world without considering the interpretation involved in its origin yields an ultimately false, distorted view of the world.
Taking this idea of interpretation further, Drucker explicates the common conflation of observed data and the phenomenon observed to generate that data. Describing data as an objective depiction of its generating phenomenon erases the idea that there is always a process of observation in collecting that data. Observation involves human intervention, and as such, it is necessarily an act of interpretation — the resulting data, then, cannot be objective; it is a result of (inherently subjective) interpretation. To remedy this issue, and to create perhaps more powerful visualizations, Drucker begins describing a method of creating visualizations that make the involvement of interpretation — or at least its related uncertainty — evident. She advocates for the removal of discreet, continuous, clear-cut coordinate system on a graph, for example, in favor of a system that has discontinuities and expresses underlying uncertainty. She sees the act of interpretation not as something that is completely incomprehensible in logical terms: describing it as “stochastic and probabilistic,” she claims that I can be modeled by the same (complicated) “mathematical and computational models as other complex systems.”
"We make the world by structuring our experience of it" (74). - Joanna Drucker, Graphesis (Interpreting Visualization: Visualizing Interpretation)
Druckers' chapter on Interpreting Visualization: Visualizing Interpretation brought to mind the many mechanisms and tools we may take for granted to structure our navigation of time and space.
Clocks, planes, coordinates. We overlay these abstract conceptions across our daily lives and they make sense.
To re-examine what has been established and standardized can be a struggle. I am led to ask: To revolutionize for the sake of revolutionizing, or to identify an evolutionary expansion? What paradigm shift can re-adjust how we perceive humankind for the better?
To date, you could argue that administrative or logistical necessities have shaped our current realities. Maps to understand where we live, sure. But proliferation for military strategic advantage, standardized time and currencies to run orderly empires and to trade with others for gain and betterment.
I am hopeful by the thoughts prompted in reading this chapter that we are entering an age where pragmatism + compassion continues the efforts that have come before us. Further aided by the knowledge that mechanization and understanding do not go hand in hand, perhaps we can also standardize a globally understood vocabulary to see from multiple angles, all the time, and not be overwhelmed.
This extensive chapter looks at the history of specific types of graphs and charts can explain to us how they function and organize the data at hand. How we interpret the visuals and how visuals are interpreted. She begins by focusing on the historical origins on each type of visualization.
Timekeeping; It is based in the solar and lunar cycles and encouraged abstraction in mapping. The aim was to create a relation to the world. Hours were created out of habits because of recurring performance. However, it can be understood in both knowledge, humanistic and scientific.
Spacemaking; It is understood through the recognition of space around someone. It is based in navigation of abstracted shapes and then later forms. It pushed us to thinking about order, grouping, size, placement, proportion, that allowed us to be more analytical with later types of graphs and charts.
Administration & Record Keeping; It was used for highly organized graphical grid. Tabular forms with groupings were created for more administrative rolls. Later on, more graphical properties added a more systemic set of relations depending on color, order, arrangement. More importantly, we can clearly see the difference between fixed elements and interchangeable ones, timetable. The concept of cross-referencing made the basic table carry out more intersections. Tabular forms were still popularized even though other forms were proven to be more efficient to the point that Graph paper was produced. The concept of visualization depended on the process of plotting observational data and then analyzing the graph. There is more responsibility when designing the format of these bar graphs.
Trees of Knowledge; They are a fixed structure but spatial relations carry meaning by using hierarchy, distance, proximity etc... The concept of generation was the main focus in these types of graphs. While the image of the tree wasn’t the main symbol, it developed into pure lines. The notion of projection, relationships, leaps, and lineage is also expressed. Trees can be also morphed into networks where multiple starting points. It is a different type of branching nature that allows the different elements link to each other. There is a more special distribution amongst these elements.
Knowledge Generators; combinatorics calculation because they support both static and mobile organization. It is a combination of fixed values resulting in a generated value. The diagrammatic form can result multiple outcomes through different readings even though the variables are static. This form became popular with medieval cosmologists. They provide a performance of probable interpretations once the variables were graphed. This triggered a use of analytical geometry as a means of plotting information on axis to express behaviors. With both types of graphs there was the creation of tools. They were the answers to set theory, calculus, topology, network theory, vectors, and other fields. This advanced the mathematical reasoning and used to prove algorithms.
Dynamic Systems; It is generative to present a process rather than a product. Usually it’s an open ended, flux system that incorporate different knowledge generators. It is not meant to product an outcome to be repeated but to analyze events in a process. Because of that, it doesn’t lend into a specific graphical format. Usually used to show activities of the same element transforming over time. This developed with the emergence of meteorological observational instruments. Graphing intangible and invisible phenomena into a graphical language. However, a type of nonlinear data was difficult to map out without the use of motion graphics possible through digital computers.
Most visualization adopted by humanist were adopted from other disciplines. Through the translation, the concept of interpretation was blurred by creating a more subjective image. We need to create expressions that allow appropriate interpretations by preserving observer-independent reality. Visualizing interpretation can be divided into four basic levels; phenomenological experience, relations among humanities, representations of temporality and spatiality, and interpretation. For all four levels, it’s crucial to balance the humanistic knowledge and the graphical display.
In general, I believe that her approach in detailed analysis is incredibly solid and supported with historical reference. Her grouping helped me remember the different types of visualizations and how they differ in their function. However, the chapter ends suddenly without allowing room for a conversation. She presents her opinion without asking the need for humanistic visualization versus scientific visualization.
Dr. Drucker's early comments on the visualisation and subsequent interpretation of temporality strike me as primarily rooted in a phenomenological account of the relevant ontology and humanism- particularly as such tenets pertain to and come into conflict with the more empirical interpretations of time found in the natural sciences that she begins the discussion with.
She characterises the divide in terms of "interpretative knowledge", stemming from a kind of cultural or subjectivist relativism. The empirical account she presents (James Allen's and George Ferguson's discrete chart of intervals) is termed deficient in mapping out "recollection and regret ... retrospection and interpretation" and more generally lacking in encapsulating notions of temporality as historically, culturally, and relationally lensed- i.e., as the ontological ground for humanistic care and subjectivity.
The final upshot of this, to my mind, principally concerns the final question with which she ends her discussion on temporality: "How to find the right graphical language to communicate this language..while being flexible enough to inscribe the inflections which characterise subjective experience?"
But if indeed these conceptions are as intersubjective, elusive to structure, and hermeneutically based as this passage implies, isn't it reasonable to ask whether or not attempting to formalise a structured language to even account for such conceptions is possible in the first place? And if so, would such pursuits even be desirable, considering the reduction a formal visualisation might impose on the meaning therein? This ultimately feels like a reformulation of the general schema mapped out in chapter 1, with a similar lack of resolution.
In this chapter, Drucker continues her investigation into the historical origins of various graphical forms. She does so in order to provide the reader with a greater understanding of the context in which forms arose and the ways in which they may have implicit biases or allegorical meanings embedded within them. By recognizing these forms as not inherently neutral or empirical, the reader (and by extension, practitioners in the field of data visualization) may be able to “break the literalism of representational strategies and engage with innovations… that augment human cognition” (p.71) — Drucker’s ultimate rhetorical goal in producing the book.
Early in the chapter, Drucker defines two basic divisions in the functions that data visualizations can serve for their creators and audiences. First, there are representations of information that is already known, which she posits as having a static relationship to what they show. Secondly, “knowledge generators,” which create new information through their use, and thus have dynamic and open-ended qualities. Drucker seems to have a clear affinity for the second type, as they more closely serve the needs of her theorized “humanistic” practice.
After running through a historical deep dive into the uses which visualizations have served (timekeeping, space-making, administration and record keeping, trees of knowledge, and knowledge generators) and elaborating on the forms that evolved to suit these particular needs, Drucker ventures back into more theoretical territory. Under the heading of “visualizing uncertainty and interpretive cartography,” she questions the very assumption of empirical observation and data collection. In her words, data are “capta”— taken rather than given, with parameters that are decided by the observer as they are recorded. Thus, the biases (and historical/social context) of the observer are embedded in the very nature of what is recorded, and the presentation of data for comparison erases the original ambiguity of the phenomena (human existence?) that were observed.
In order to restore the ambiguity of human existence and the natural world to the field of data visualization, Drucker suggests a few strategies that break significantly from the standards of the field. The use of unequal measurements or standards, disjointed plotting, and “fuzzy” or geometrically undefined forms could serve as visual markers of complexity and uncertainty. She fails, however, to provide much in the way of concrete visual examples of these practices for the reader to gauge the effectiveness of these posited strategies. In this way, it seems that Drucker’s project could be better served by the inclusion of practitioners in the field who are also interested in “humanist” representation. Where she now seems content to remain in the realm of theory and art-historical analysis, she could instead move further into the realm of experimental practice.
Chapter 2: Interpreting Visualization, Visualizing Interpretation
This chapter is a long tour through through several histories of information graphics. It begins with a tour through different graphical formats, some common, some less so. Many of the styles and format share certain principles, that function together in a systematic way. Thus the examples chosen are meant to bring order to information. I thought this sentence was quite relevant: "The challenge is to break the literalism of representational strategies and engage with innovations in interpretative and inferential modes that augment human cognition." This made me think twice about my approach to the clock exercise; perhaps there are additional ways to actually aid our understanding then just re-creating the already familiar?
The next section on what conventions we use is also interesting. For example, we all experience day and night and have clear markers for these things; but there is no clear delineation for say an hour or a minute, yet we accept these as familiar cultural norms. Also the notion of time "slow book", "fast movie" can also be very much one of individual experience and variable. I think it is very much Drucker's point that "data are capta" - they represent an observer's point of view.
I do like how Drucker keeps coming back to her definition of knowledge generators. The comparison to statistical graphs on pg. 89 is helpful in understanding the difference. Knowledge generators can give rise instead to "multiple interpretations or analysis." I think I see the point; statistical graphics are usually just representations of data; knowledge generators product the knowledge themselves. I think the diagrams on pg. 115 are the best example of this - as Drucker says the diagrams "perform the act of reasoning". I was less impressed with the idea that a list of numbers is a knowledge generator - meaning because they can be added they produce knowledge. "Combinatoric Calculation" seems like an overly broad definition.
I think Drucker's take on the John Snow map is telling. "Each dot represents a life, and no life is identical." I'm not sure if she is suggesting the dots should reflect different variables. I could imagine the map becoming overly and needlessly complex. After all the map was a tool to solve a problem, which it did help.
The idea of turning the map into an explorable 3-D type world is interesting, but to what end? I think this is where Drucker is looking for a "story". I'm not sure if this visualization benefits from being turned into a 3-D exploratory world.
Drucker seems to provide a deeper explanation of 'humanistic' on page 130. In my reading, her point is 'humanistic' is basically the opposite of objectivity. Scales contain breaks, repetitions, edges of shapes may be "permeable" and graphics represent a point of view. Her sentence "recognizing that such methods are anathema to the empirically minded makes even more clear they are essential..." I find myself agreeing with some of her views but shaking my head at the poor argumentation for them.
The preamble for this chapter starts off with a much-needed (and welcomed) clarification of terms followed by a survey format similar to the last. Drucker wastes no time in solidifying the distinction between the two types of visual forms mentioned in chapter one. These are roughly visual forms as rhetorical assertions vs. knowledge generators.
Relative to the questions raised from my last entry, and regarding her use of loaded terms like "knowledge", this was a particularly important section as I can now decipher much of what left me baffled previously.
With respect to representations and knowledge generators Drucker defines the former as being static to what they visualize while the latter have "combinatoric" qualities. That is, using her example of the train schedule, there are many assertions within a train time table and everyone who experiences the table does so in a different way determined by their needs. She would assert initially that table lends itself to an algorithmic process while suggesting later that a knowledge generator can "perform" the act of reasoning. While I would agree with the former, the later remains unparsable.
Of importance to me is that while perhaps sharing similarities, I (now) would not imagine that she is referring same sort of Knowledge as understood by a philosopher or social scientists. With this context articulated and reasonably defined, the rest largely fell into place.
Again the bulk of this chapter belies a studied historical acumen. Drucker once again organized a trove of (mostly) relevant examples of graphical forms, not only to flesh out the inductive argument from the prior chapter, but to serve as exemplars of positive and negative use cases. For each, she does her best to disambiguate specific types of similar visualizations and contextualize them historically. Each is defined and introspected into for what their "graphical relations" can tell us about the practitioners, their disciplinary roots, how they perceived various units (time, space, administration, etc.), the data they found relevant, and the abstractions their cultures were capable of making.
Towards the end of her review, she starts to consider what I would regard as a more active types of visualization. That is, those which are also physical mechanisms or embodied in a GUI. This is important because she implies that to the degree that any visualization is combinatorially a knowledge generator, it will possess more pursuant complexity. With this, Drucker begins a summation I both agree with, and am conflicted by in equal measure.
The distinction between data and capta is relevant. I support the idea that everything is relative/co-constitutive, firm belief in the Truth of simple graphical forms needs to be challenged, we rush to visualizations, and (to a degree) richer / more ambiguous data sets could be useful in some cases. To make a simple example, I am skeptical that a handful of x-y graphs should be considered a fair handling of a complex issue like gender pay inequality.
That said, I'm not yet convinced that better graphical forms are the best place to put our efforts. I understand the issues she outlined foremost pedagogical/cultural issues. Before we create new visual forms and ambiguous visualizations, it would be useful if academics would be trained to understand that the forms should be used cautiously and always in a context of healthy skepticism. Transparency can not just be assumed.
As if to support my conclusion she introduces the final example of the cholera map and proposes alternate data sets like the viewpoint of an "elderly man whose son has just died" be added to make it more "expressive". The example comes replete with a mock-up. I really can't tell if she was just trying to be provocative with this prototype. It is unusable to me.
In a real sense, it seems like her example wants to be a crossover of a human-run state machine and an ambiguous data mining tool. My problem here is that both of these tools are still largely the domain of computers for reasons we discussed in class about colors and shapes. Humans don't have such extraordinary visual capacity in these ways. If I remember correctly, we placed the human imposed limit at around six different colors and shapes for each attribute. How would this be any different? This said, "conceiv[ing] of every metric as a factor of x" seems absurd on the face of it.
The truism that there is no technical solution to a social problem seems relevant here. As Drucker herself stated in this chapter, the visual forms we use, imply what we find important and our societal conception of the issue. Don't these precepts need to be challenged effectively for new visual forms, of the sort she is proposing, to be applicable?
interpreting visualization :: visualizing interpretation
In the second chapter of Graphesis, Johanna Drucker takes her readers through the development of a variety of specific visualizations across time, from timekeeping in the Roman empire all the way through to complex mathematical visualizations with modern digital computing tools. Most crucial in her account of these visualizations and their uses for Drucker, however, is her explanation of the societal context in which they spawned - specifically with what assumptions and interpretations are baked in.
It is in this chapter, Drucker extensively lays out her skepticism and caution regarding the conflation of visualization and objective knowledge. She argues that all data are capta, which I interpret as observations or interpretations of one's surroundings rather than hard pieces of fact, and all information is interpretation. She emphasizes this point in this critical chapter, retorting:
...the rendering of statistical information into graphical form gives it a simplicity and legibility that hides every aspect of the original interpretative framework on which the statistical data were constructed (128)
Her standpoint is morally sound and well backed by examples of misuse - most interestingly to me in her comparison of network diagrams/topic maps to tree diagrams. While order and spacing carry analytic value in tree diagrams (you cannot adjust the spacing or hierarchy without changing the meaning), while variability in the configuration of network diagrams may not signify a substantially different relationship between the data points represented. Readers may easily still interpret aspects of this configuration as having semantic value (e.g. spatial differences), potentially even without realizing it, creating a very real window for misinformation.
I find Drucker's case for caution in conflating visualization as fact very compelling, and I am curious to what extent she sees analytical and intellectual merit for the use of visualization in fields outside of humanism, if she sees any at all. Her passionate mistrust of visualization is well documented in this section with flowery language (which I found overly obtuse and dense at times):
These graphical tools are a kind of intellectual Trojan horse, a vehicle through which assumptions about what constitutes information swarm with potent force (125)
She mentions that "crudely conceived numeric statistics are useful only in the most reductive circumstances," (133) but I find it hard to believe that fields like Mathematics and Science do not benefit from the use of data visualization that drive these disciplines towards the discovery and communication of universal truths that hold real intellectual and societal merit. Drucker herself cites Snow's visualization that helped identify the origin of Cholera outbreak, which saved many lives. Her additions add unmistakably important societal context about who Snow's data points are and how they feel as human beings, but I am curious to hear the class's opinion on how these additions that make this a "more complex statistical view" influences its effectiveness and/or importance within society.
In this chapter, Drucker lays out the core of her argument in this book: That all forms of data representation/visualization are actually acts of interpretation, even if we don't see them as such anymore. She makes this point very compellingly by pointing out that much of our thoughts about concepts of time, mapping and mathematics come from (sometimes arbitrary) graphical conventions that are so common place we don't notice them anymore (like breaking time into discrete 'day' chunks or taking for granted the implicit chart form underlying basic arithmetic operations).
"So naturalized are the maps and bar charts generated from spread sheets that they pass as unquestioned representations of "what is."" (p. 125, emphasis added)
She then argues that we need to develop methods of visualization that acknowledge this fluid aspect of interpretation to bring to the forefront the fact that all data already includes some level of interpretation.
"Rendering observation (the act of creating a statistical, empirical, or subjective account or image) as if it were the same as the phenomena observed collapses the critical distance between the phenomenal world and its interpretation" (p. 125)
Then, to make her argument more tangible, she then asks us to imagine if Snow's famous chart of deaths from cholera could be redrawn "to express the emotional landscape" or "from the point of view of a mother of six young children." This is where I start to disagree with her position. While I do agree that it is important to keep in mind the social conventions that come pre-baked, so to speak, in our representations of data, it is also exactly this ability to abstract from an individual's viewpoint that gives strength and generalizable force to data visualizations.
She even touches on this in the section about cartography,
"We cannot "see" the land's shape, its contours, or outlines ... But abstracting this into a topographic view requires understanding the rationalization of surface and its ordered schemes" (p. 77-78)
Overall, I think Drucker's critical analysis of the interpretive nature of data is incredibly strong and compelling as she astutely points out graphical conventions that are taken for granted and add implicit meaning to representations. However, her call to action for re-examining data visualizations from a humanistic perspective goes too far and seems naive in underplaying the importance of abstraction.