A1-TealeFristoe
From CS294-10 Visualization Fa07
[edit] Good Visualization
Explanation and discussion
This image was taken from Principles of Geology by Gilluly, Waters, and Woodford. The image features a series of photographs of a man made river, showing how a constant forward force can create a meandering stream.
The image makes good use of photography to tell a story. As the images progress to the right, we can see the river develop through time. The stark contrast between the water and soil in the photos shows an actual example of a river developing, which is a more believable and convincing change from the illustrations common throughout the rest of the book. Finally, a subtle white grid overlays the various images, allowing for more accurate comparisons of the images without overwhelming the content of the photos.
Deconstruction
The size and shape of a river as it develops over time are the subjects of this image, so its data model includes position in two dimensions and time.
Because the image is a collection of photographs, it does not have especially well defined image model or encodings for position. Roughly, the position of the river is represented by the x- and y-axes in each of the smaller pictures, though the perspective of the photographs means that the correspondence between page coordinates and actual coordinates of the river is imperfect. The overlaying grip helps to show this, but it also helps to show how the river is related across time. Time is also encoded on the x-axis, though between the different small multiples as opposed to the x-axis within each sub-image. We can't be sure that time is accurately represented according to page space either, as there is no indication of how much time has passed between each picture other than the rough "a few days" between the first image and the second two in the caption.
Using photographs means that a truly immense data set is contained in this image, but we can estimate its size by simplifying things a bit. There are approximately 30 grid squares on each of the 3 images. We can say each of the grid squares contains information about the point at which the river enters the square and the point at which the river leaves the square, for a total of 2 points in each. Therefore, the visualization constitutes a data set of 30 x 3 x 2 = 180 points of information. This estimate is of course very rough, because in actuality far more information about the river is contained within each grid.
The overlaying grid does not contain any new information, as the information about position is encoded within the photos themselves. However, the grid is so useful in showing perspective and comparing the river at one time to another that I believe it is essential for making the visualization useful.
[edit] Bad Visualization
Explanation and discussion
The following image was taken from the June 22nd, 2007 edition of New Scientist. The article it accompanies discusses how evolution, perhaps counter-intuitively, can sometimes lead to less complex organisms. The image complements the article by showing that some species that have no brains exist today despite the fact that their ancestors had brains.
This image has a number of problems. To begin, it contributes very little to the article other than flashiness. Its contents could be summarized in a single sentence: "Natural selection has caused some species to lose their brains."
However, even this meager content is lost in a collection of chart-junk. While it is definitely nice to have some data in addition to the data that delivers the message, so the viewer has some means of comparison, this chart features 19 branches that do not show the desired information versus only 5 that do. Cutting the chart off at Urbilateria, the common ancestor that had a brain, would have been a good first step, but the Ecdysozoans also fail to contribute any useful information. The scientific names of the more general classes, like Protostomes and Deuterostomes are also excessive. Finally, the images of different animals do not contribute to the message of the chart.
Perhaps worst of all is the choice of colors to show the branches that have lost their brains: putting green against and green background and red against an orange background and pink branches should have raised some eyebrows.
Deconstruction
This image is primarily concerned with different groups of organisms, their evolutionary relationships to one another, and their relative complexities. The image revolves around the phyla, subphyla, and classes that label the leaves of the tree, but also includes the more general groups labeling the colored boxes. The complexity of the organisms is defined by the ordered set of "All species have lost their brains", "Some species have lost their brains", and presumably (one of the weaknesses of the image is that the final class is not definied) "No species have lost their brains (perhaps because they never had brains)."
In the image, position along the x-axis represents different groups of organisms; position along the y-axis roughly represents time, which is used to help establish relationships between groups; and color represents complexity of organisms.
There are a total of 25 different phyla, subphyla, and classes described, which constitute all of the data points in this image.
Uninformative elements in this image include the four drawings of organisms, which are unlabeled and add no information about complexity.
Redesign
My primary goal in redesigning the image about a loss of complexity through evolution was removing much of the image which did not contribute to the message. To start, I removed all classes and phyla outside of the Bilateria group, because none of them lost their brains, and it isn't even clear if any of them even had brains to begin. I also removed the entire Ecdysozoans branch, because even though they descended from Urbilateria, none of them lost their brains, and so they do not contribute to the meaning of the image. Removing this branch did not hurt the image at all either, as the image makes no strong quantitative or statistically claims. Finally, I removed the Protostomes and Chordates labels, because these groups do not make useful distinctions.
After pruning away much of the chart junk, I chose a new color scheme, which I believe makes the image more understandable. I chose to encode complexity by darkness, so the groups with members that all have brains are dark blue, while groups where members have all lost their brains are light blue. This helps identity complexity as an ordinal data type rather than a nominal one.
By removing a lot of the image, the redesign became radically different. One aspect in particular that I regret losing is the explicit tree structure. While the tree structure of the data remains in the image by its embedded structure, by losing the explicit branching it also loses its sense of time and therefore the message that some species have lost their brains is not communicated as well. However, because the time information in the original image was rough at best, and because I wanted to make things cleaner and simpler, I decided that losing this message was worth it.
