Info We Trust

by R J Andrews

  HENRY DAVID THOREAU, 1849

  This and the next chapter go together. Think of them like a pair of optical lenses. Metaphorical magnifying glasses are often employed by authors to peer at things. But here we will not employ lenses to look through. Instead, we will use them to expand and focus, as prisms transform light.

  Exploratory analysis, which we detailed in the first half of the book, expands your personal knowledge. Probing generates insights. It also produces visual artifacts that document what the data has to show. The more time you spend, the wider the field spreads. This expanding process casts a broad spectrum of ways to look at the data. As you begin to understand the data, a new sense of what is emerges. You note that certain views might also help others see what you have learned. This creative expansion and awareness is what this chapter is about. The second lens catches these products and focuses them to the reader so that they receive a meaningful data story. That is what the next chapter is about. To take a practical view of each of these broad actions, I want to step through the development of a new data story.

  Median Gothic

  I recently heard something at a reception that spiked my curiosity: Gothic cathedrals point toward Jerusalem. It startled me. Here is what that statement suggests. An arrow that begins at the main doors of a church, continues up the aisle, and through the altar would point, like a compass needle, toward Jerusalem. It claims that not just one Gothic cathedral points like this. All of them point toward Jerusalem. Does medieval architecture face worshipers toward the same city, thousands of miles away? It seemed impossible. I had to know.

  Cathedrals are deeply symbolic. From above, shorter transepts cross a long nave to make a Latin cross, what's known as a cruciform plan. The nave's name comes from the word for ship (like navy).

  Its ribbed vault is like an upside-down boat, symbolically bearing its people through the storms of life. Three front portals represent the Trinity doctrine, one god in three divine persons. We could go on, but the point is that a lot of thought went into layering meaning into these buildings. Maybe it is possible that their orientation was symbolic too.

  I always associated religious directionality with Islam, not Christianity. But maybe “cathedral compasses” are real. Did every European city clear enough space in their center to make room for a properly angled cathedral? Then, I realize: The cities were planned around their churches, not the other way around. To sort it all out, we need data.

  There are a lot of cathedrals in Europe. We are immediately faced with a deluge of possible data points. Constraints are needed to make this inquiry manageable. Gothic architecture emerged in France, where iconic cathedrals are still mobbed by tourists. Looking at the earliest French cathedrals, a few dozen sites, might be enough to satisfy curiosity. The first step is to build a list of early and high Gothic French cathedrals. A list of their names and parent cities is a good start. Right away we can locate them on a map.

  Then we can start adding a little context to each point. It seems natural to append a date to each cathedral. They each took a while to build. Many are still symbolically not completed. The year construction began, not the year construction stopped, is the moment orientation was set. So we look for the year of each cathedral's groundbreaking. Different sources provide different years for the same cathedral. Hmm. Maybe we can investigate this discrepancy further. For now, we can filter to cathedrals associated with the 1100s and 1200s.

  Next, we need to determine each cathedral's orientation. I visit them on OpenStreetMap and layer a compass over each nave's axis. You can see, for example, that Notre-Dame de Paris points 25° south of east, a value I log as -25°. I continue measuring the angle of orientation for the rest of the buildings. Seeing each additional site, I begin to sense interesting repetition and variation. An hour or two later, we have name, city, coordinates, date, and orientation for 38 French cathedrals. The data table is ready to play.

  Right away, I scatter the cathedrals according to their groundbreaking dates and angle. Each cathedral gets an orange dot. Time goes to the right. Angle data is anchored on a due-east baseline in the middle of the plot. This takes advantage of the north goes up/south goes down convention. You can see Paris at 1163, the year its cornerstone was laid, and -25°.

  Two things jump out at me from this scatter plot. First, across a possible 360°, there is a definite clustering of orientation. They all point eastward, staying away from north (+90°) and south (-90°). Not a single one points westward. This encourages me to think there might be something to this orientation idea. Second, there does not seem to be any particular pattern through time. For now, we can abandon the temporal x-axis. I also want to depart the specificity of the scatter plot to see what some grouping might reveal.

  If we bin the angles in five-degree increments then we can make a profile. The cathedral orientation histogram represents each cathedral with a blue bar. I placed it next to the scatter plot so you can see the correspondence between the plotted and binned data points.

  Now we seem to have something. Even with these few-dozen data points, a rough unimodal distribution appears, peaking around 20° south of east. Before we get too excited, we need to understand one more thing about that original statement. What does toward Jerusalem mean? Because the Earth is round and France covers a sizable amount of land, we actually need to consider directions toward Jerusalem. The angle, or bearing, changes depending on where you are in France. Could those directions be what we are seeing in the distribution?

  Here is what the direction toward Jerusalem looks like from six of our locations. Flattened on a map, the straight lines across the globe look like arcs.

  I calculate the individual bearing from each city to Jerusalem. Then, I find the difference between cathedral orientation and true bearing to Jerusalem. The result is disappointing. Factoring in the unique directions toward Jerusalem seems to refute the Jerusalem hypothesis. This is something we can really appreciate by plotting each cathedral's orientation on a map, in orange. A few of the cathedrals seem roughly aligned with the yellow bearings toward Jerusalem. Many more are not. I feel a little deflated. Cathedrals do not point toward Jerusalem after all. Undeterred, I remember my earlier confusion with groundbreaking dates and return to learn more about when these orientations were set. Maybe looking deeper into the architecture of the buildings can help. I begin looking at top-view plans and entrance facades. Perhaps an aesthetic grouping could help explain what is going on? Maybe based on size or style? I start making a pictorial multiple to compare cathedral height, length, and architectural style. Once again, it seems like a dead end. Size was a function of the wealth bishops could muster. Style changed across hundreds of years of construction, famously producing mismatched spires in some cathedrals. Neither size nor style is useful for making any more sense of cathedral orientation.

  You might notice that I did not illustrate all the cathedrals before pulling the plug on this path. Knowing when to quit a certain investigation can sometimes be a fuzzy choice. You do not have to carry every idea to its conclusion to know it is the wrong way. Sometimes going partially down a path tells you all it has to offer. And even if it is not what you expect, it might still be interesting.

  While reading up on cathedral plans I learned a little about their groundbreaking dates. Many medieval churches have deep roots. Some were built on religious sites that go back to the 400s. Some may even be on the sites of older pagan temples. It would require a bigger effort than I am capable of mounting to say exactly when the angles of these buildings were set. Perhaps dates are another dead end? That is O.K. Rejecting certain paths can help us focus on others.

  Reading about cathedral history yields one more discovery. While researching history and design I stumble into a research paper that somehow eluded my initial search. Its title, “The Solar Orientation of the Gothic Cathedrals of France,” is full of promise. Its author, Amelia Sparavigna, seems to provide an answer. Cathedrals do not point towards Jerusalem. Cathedrals point to
ward the sun! And not just any location of the sun, but sunrise on a day important to the cathedral. Examples include feast days that honor the saint the cathedral is dedicated to.

  I return to the original histogram and add in a band of yellow to represent the range of sunrise locations across the year in France. The sun, of course, rises at a different point on the horizon every day. Almost all of the cathedrals fit into this solar arc. Metz, Chartres, and Le Mans all lay outside of the sun's band. They can be explained by history: All three are known to be built on the site of an ancient church. Their orientation was set hundreds of years before it became fashionable to face worship towards sunrise.

  As I look back across this visual investigation, I am satisfied with the maps, diagrams, and charts that helped me see. My initial disappointment, about Jerusalem, has already faded. I savor the knowledge that bogus statement led me to. What is more incredible: that all of these monuments face eastward, or that you could never figure it out if you were standing right in front of one? Looking at the data gives an insight impossible to gain from being there in real life.

  The deeper I got into the question, the more I thought about how I might share this insight with others. What is particularly interesting about this? How can I make it resonate in an unforgettable way that you cannot ignore?

  Content and Form

  Content is the heart of story. Content is the actual events that happened. Form is the way that those events are retold. If numbers are content, how you arrange them on a number line is the form. The fact that something happened is the content; the way a film portrays that event is the form. It is the form that makes the content engaging. Our data is the core of our content. But remember, the raw data is no way to inform anyone. The content has to be arranged into a form so that others may appreciate it.

  If you spend time with data, and are a little lucky, you discover something worth sharing. Perhaps it is a specific piece of content, like a surprising correlation or interesting outlier data point. Maybe it is a new form, such as a better way to orient a comparison. Sometimes, the thing you want to share is a map that gives people the context and tools necessary to find their own meaning. Before you consider why other people are going to appreciate the content you have to offer, consider why you care. Why are you ready to put in all the work necessary to craft a data story?

  The cathedral-orientation story could be summarized with a terse cocktail party opener. Did you know that all Gothic cathedrals face the sun? Oh really, I hear you respond, that's nice. This exchange does not get at why my new understanding is so spellbinding to me.

  When I was 14 years old I stood before Notre-Dame in Paris. I remember looking up in wonder at its facade of stone and glass, squinting to see the famous gargoyles. Two decades later, I now know that someone had designed the direction of my gaze. And that realization came about from an exciting experience of actually seeing it in a new way. I was amazed as I loaded each individual city map, unable to find a single cathedral that faced west. It was almost dreamlike, as if France's greatest medieval monuments lined-up, one-by-one, in neat rows before me. I now imagine bishops and builders, a thousand years ago, waiting in the cold of night for first dawn so they could determine where to begin building a monument they knew they would never see completed. What an epic story.

  Once again, I am awestruck by the same old building. But this time, it is not because I looked deeper into Notre-Dame de Paris. Only by seeing the other orientations did she become more interesting. Each individual cathedral orientation is unexciting. But together, new meaning emerges. Together, they reveal a reality unseen across daily life.

  Not all data reaches into your childhood. I do not want to exaggerate the personal resonance (or importance) of this particular vision. But I am sure of one thing: Your own experience examining data makes any insights gained personally meaningful. Figuring out how to share that experience can help the content become more meaningful to others too. I do not want you to take my one-liner about cathedrals at face value. I want you to see the pattern, on your own, so you can create meaning for yourself.

  Form cannot be greater than content, it is only the manifestation, the shape of content.

  BEN SHAHN, 1957

  Sometimes, even after you have the data insight, another flurry of creative data sketching is necessary to find the right form. Expansion continues as you search for the right ways to evoke what is meaningful. As we saw with the cathedrals, different visual forms evoke different aspects. Some will highlight the content you want to deliver better than others. This book does not offer any chart-choosing decision-tree flow-map. Instead, pay attention to how well different charts illuminate your data. Depending on the content you wish to elevate, certain charts will create stronger visual metaphors. The same visual meanings of position (up is good), size (big is important), color, and everything else we have covered apply across all forms.

  There are hundreds of statistical chart types we use to visualize data. But, they are only one flavor of visual storytelling available to us. Geographic maps and pictorial diagrams also reveal. They can give us new ways to think about statistical exploration. Those curved yellow lines toward Jerusalem gave me the idea to calculate each cathedral's orientation-bearing difference. Furthermore, maps and diagrams help anchor abstract data back to the real world from which it came.

  Abstract means existing in thought or as an idea, but not having a physical or concrete existence.

  As we become familiar with something, we think about it more in terms of the use we put it to and less in terms of what it looks like and what it is made of. This transition, another staple of the cognitive psychology curriculum, is called functional fixity… because people get fixated on an object's function and forget its physical makeup.

  STEVEN PINKER, 2014

  Once you have the story content, look down. See that you now hold it atop a tower of abstraction and specialized context. For the cathedrals, we climbed and manipulated a stack of abstract ideas. We learned about archaic architecture, geographic bearing, and solar angles. Now, you must guide your audience through the same tangle of concepts, hopefully making the tower easier for them to scale.

  We use form to develop more sophisticated understanding of abstract content. And then we must take that knowledge and make it more accessible. We have to make our invisible understanding, visible. From imagination to image.

  CHAPTER

  18

  FOCUS ATTENTION

  Vision without execution is hallucination.

  WALTER ISAACSON, 2014

  The previous chapter began with a visual metaphor about two lenses. The first diverging lens expanded creative exploration. It set loose a search for insights, and better forms for conveying those insights. The second lens is a converging lens. It catches the products of our creative exploration and focuses them to the audience. That is what this chapter is about.

  There is an endless variety of ways to package a data story, depending on purpose, audience, medium, and distribution channel. You are the one who knows these best, and must keep your own particulars in mind as you focus your story.

  Workshopping

  There are two ways to draw a dragon. In one approach, first block out the general shapes with light lines so it is easy to correct errors. As you become satisfied with the form, begin adding detail. Gradually, darken lines. The dragon emerges through a conversation between the productive hand and the critical eye.

  As the designer quickly creates a conceptual design sketch an ongoing perceptual critique is occurring. This can be thought of as a form of meta-seeing in that it is critical and analytic … It involves the interpretation and visual analysis of the marks on the paper that have just been put down.

  COLIN WARE, 2008

  The other way starts from the head of the dragon and includes every detail in one continuous flow toward its tail. Some master artists are able to draw like this. They may claim that the creative planning occurs in their mind's eye before p
en hits the page. I cannot draw anything by making only a single detailed attack. I also do not know how to produce a successful data story without many cycles of creative iteration. Editing is one way of thinking about creative iteration. Its association with deletion can cast a negative connotation. Throw out the bad and keep the good. This attitude portrays editing as a purely subtractive process, like carving a statue from stone. Editing is more like working with clay, which allows you to add as you remove. As you critique existing forms, new forms may emerge. You might call editing revision, the work of seeing something anew.

  Editing is not taking out, it's putting together. It's taking a story, which has been photographed from many different angles and, very often, in many different takes, and making it play in the best possible way that it can.

  DEDE ALLEN, 1923–2010

  Instead of editing, I call the process of revision workshopping. When we workshop a project, we scrutinize it in pursuit of making it better, just as a musical is refined through its rehearsals. Here is the cast of characters who help the creator workshop a project.

  Whenever you feel an impulse to perpetrate a piece of exceptionally fine writing, obey it—whole-heartedly—and delete it before sending your manuscript to press.

  Murder your darlings.

  ARTHUR QUILLERCOUCH, 1914