Discussion of Good and Bad Visualizations

From CS 294-10 Visualization Sp10

Jump to: navigation, search

Class on Jan 27, 2010

Contents

Readings

  • Chapter 3: The Power of Representation, In Things That Make Us Smart. Norman. (pdf)
  • Chapter 4: Data-Ink and Graphical Redesign, In The Visual Display of Quantitative Information. Tufte.
  • Chapter 5: Chartjunk, In The Visual Display of Quantitative Information. Tufte.
  • Chapter 6: Data-Ink Maximization and Graphical Design, In The Visual Display of Quantitative Information.

Optional Readings

  • The representation of numbers. Zhang and Norman. (pdf)

Danielle Christianson - Jan 26, 2010 05:08:00 pm

Norman reading: Great reading to think about:

1) the importance of cognitive aids (including language): I thought Norman's assertion that we only value information/data that we can measure (or represent) very important (pg. 52-53). This, along with a recent article in Wired discussing the woes of scientific research [1], make the case for a careful look at unexpected or difficult-to-explain data. Additionally, I would have liked Norman to expand on his suggestion that writen material is read without much reflection today, affirming Socrates fear of books.

2) a framework for why some representations work better than others (experiential vs. reflective): This is especially the case for Bertin's position, as it is experiential. Two side notes on this: a) I recently read a description of research done on rats in which they were able to detect neurons firing in the same spatial pattern in the brain that the rat was moving through in a maze. This suggests that we might be hard-wired to understand spatial information. Interestingly this same summary article was talking about 2 different types of navigators; one who can make mental maps and the other that uses landmarks. Not sure if the results with the rats would apply to the latter. And b) A friend was telling me about a book (name to be added later) describing the role of movement in cognitive development. The author makes the case that most cognitive processes stem from movement, which was evolutionarily used for food acquisition, danger aversion, or reproduction.

Also, Norman didn't explicitly state this but I felt he somewhat implied it in his discussion about tally's vs. numbers: substitution representations are experiential and additive representations are reflective. Is this a correct interpretation? I'm not able to think of a counter example.

3) the complexities of Arabic numbers: I think this is something of which most scientists, especially those in quantitative disciplines, are unaware. As a largely quantitative person, I had definitely overlooked this.


Tufte: I mostly agree with Tufte's assessment and can see the data-ink guidelines being a very useful baseline from which to design data graphics. I do think that Tufte goes a little overboard with the deletion of some information. One example is his rework of the French age-sex pyramid. I feel he should have keep some of the numerical information, which side is male vs. female, and the causes of various demographic features. I suspect he would argue that this could be included in accompanying text, but I think it is much harder to jump back and forth between a block of descriptive text and labels on the graphic. Another example in which I disagree with Tufte on two points is his rework of the box plot. First I think that he may be underestimating the ability of the reader to quantify blank space (the lower example on pg. 125 with points for the mean and lines for the quartile to max/min values). Secondly, I think his preferred boxplot form on pg. 124 (the offset broken line representing the mean and quartiles) is aesthetically jarring. I feel this is a general fault of Tufte's work: he asserts that consideration of the aesthetic is unnecessary. Most of the time his graphics end up being aesthetically pleasing, but I wish that he would not attribute this to only maximizing the data content and data emphasis. Aesthetics do influence the reader's interpretation of the data representation and subsequent understanding and critical use; thus, it is necessary for the designer to consider it.

Jon Barron - Jan 26, 2010 03:00:38 pm

Norman: Favorite quote: "Solving a problem simply means representing it so as to make he solution transparent". This idea has ramifications beyond visualization... it's effectively the foundation of machine learning. There were other parts where the overlap between visualization and AI-style problem solving seemed to blur. In the discussion of plane scheduling, the author shifts the discussion from "how do we visualize the data" to "how do we visualize the data such that some criteria we're trying to optimize (travel time) is easily evident". Implicit in this shift is the choice of what to optimize.

Tufte: The data-ink ratio, though a useful concept, seems dangerously one-sided. Some visualizations Tufte bashes, such as the grayed-in bar plot with numbers written above on page 96, seem completely acceptable. The features of the bar that Tufte claims are independent (the left and right sides of the bar, for example) do not distract, and the number serves a distinct purpose from the bar (the exact value is often useful, like in scientific publications), and therefore doesn't seem redundant.

Similarly, Tufte's redesign of Turkey's "box plot" into these "quartile plots" all seem like huge steps backwards. Using negative space to denote the middle 50 percentile is just not intuitive. The boxes works. I do, however, like the "white grids". A lot.

One issue with Tufte's obsession with erasing ink for the sake of "efficiency of communication and production" is that now that we've all but switched to electronic displays, *white* is much more expensive to display than *black*, in terms of electricity. Perhaps we should be maximizing ink and minimizing white? Does what works in black-on-white still work in white-on-black?

Arpad Kovacs - Jan 27, 2010 01:31:09 am

Norman

Favorite quote: "It is through metarepresentations that we generate new knowledge, finding consistencies and patterns in the representations that could not readily be noticed in the world"

I thought Norman's coverage of isomorphs was excellent, especially the sum15->tic-tac-toe solution. This page has a collection of 6 similar "visual proofs", which again demonstrate the efficiency of experiential tasks over reflective ones.

Finally, the analysis of how easy it is to add roman numerals compared to arabic numerals make me wonder whether there are more efficient representations for other operations such as exponentiation or modular arithmetic. For example, exponentiation of powers of 2 is easy using a binary number system (just append zeros, eg 4decimal = 100binary, 4^5decimal = 10000000000binary), but perhaps there exists a more compact and universal notation.

Tufte

Tufte's recommendations of maximizing the data-ink ratio and removing chartjunk provide excellent advice for ensuring clean presentation. I especially appreciate his advice to chartmakers to avoid Moire patterns, which I find extremely annoying. However, when taken to the extreme, some of the minimalist final charts seem less informative and usable than their intermediate revisions. For example, the intermediate dot-dash-plot revision of the archetypical scatterplot is cleaner and conveys the distributions more intuitively than the final rugplot which appears excessively cluttered. Likewise, on the age-sex pyramid, removing all the labels detracts from the graphic; in my opinion, Tufte should have left the age scale in the center to provide context. Finally, I find the offset quartile plot harder to read than the dash-space-dot-space-dash revision, or even the original box plot, since the distinction between the quartiles is too subtle.

I also disagree with Tufte's assertion that the California Applied Irrigation Water graphic on page 101 is a "duck" that uses gratuitous art. The inclusion of the California map in the background helps to spatially organize the data (so we can observe the relative impact of latitude on water usage) and denotes the size of each basin via muted gray boundaries. The topographic relief gives an approximation of the amount of flat, arable land in each hydrologic basin area, which is an important determinant of its water use statistics. I am not convinced that this graphic could be simplified without discarding some of this potentially useful data.

Another issue with Tufte's recommendation is that the form of an unconventional/oversimplified chart could distract readers from the data. For example, most people can instantly recognize and read a bar chart, but may be bewildered by the pruned version if no explanation or context is provided. Therefore I think that Tufte's data-ink and no-chartjunk principles should be gradually phased in over time, so that people can grow accustomed to these new, more efficient styles of presentation.

Mason Smith - Jan 27, 2010 01:51:42 am

Tufte: I thought Tufte's box-plot revision was interesting, but much more so in the context of range-frames. I also think it's somewhat ironic that his final revision (line, offset line, gap, offset line, line) of the quartile plot uses more ink than the equivalent previous revision ( line, space, dot, space, line). The final revision looks too much like bad printing, essentially. I think the use of negative space is fine; it would probably be unfamiliar to most readers at first glance, but it can be explained away with a one-liner.

The data-ink ratio concept is useful, but Tufte seems to overlook the subjective aspects of its 'measurement'. Whether or not ink is necessary for a particular label, for instance, depends on the expected audience, the context of the visualization, and the novelty of the visualization, among other things. For instance, in his bar-graph reduction on page 128, I'd argue that, without the context of bar graph discussion, the sans-baseline version might not register as a bar graph at all. Even the version with a baseline, but without the percent y-labels, is questionable in my opinion.

Jiamin Bai - Jan 27, 2010 02:14:30 am

Norman:

What a good read. I never thought of how abstraction/representation can have such profound effects. With his examples, he has clearly demonstrated and conveyed his ideas of abstractions/representation by using excellent abstractions/representations themselves. After reading this article, I feel that humans really communicate best visually. Yet, not many people are exploiting this, resulting in many inefficient and perhaps misleading modes/representations for communicating ideas. Everyone who wants to communicate effectively should at least read this.

Subhransu Maji - Jan 27, 2010 06:33:10 am

Tufte : Though maximizing the data ink ratio is an useful concept, it suffers from two problems. First the notion that data ink is equal to the amount of ink seems incorrect. A better notion should be the entropy of the ink. Uniform patches of color in the bar plots, negative spaces etc though use a lot of ink, convery only a small amount of information.

The other problem is that it seems to ignore the interpretablity of the data. Cues like lines and shading often aid visual grouping, making it easier for the viewer to percive and reason about the data. For example, if the viever cares about the exact values of data, then the actual numbers or grids are very useful. In Tufte's redesign of the box-plot, the use negative space to denote the percentile in Page 125, causes an unwanted visual effect of grouping the dots together.

Chetan - Jan 27, 2010 09:37:36 am

Tufte - Ch. 4-6

I liked Tufte's treatment of the material. His insights are quite sensible and meaningful. I agree with him that too much of graphic design is relegated to the art dept.'s of newspapers and magazines. and he seems to think that an honest representation of the data is sufficient to get people to be interested. because interesting data speaks for itself.

However, i believe that humans are not purely information creatures and art goes a long way in making a piece of information appealing. even if data is interesting on its own, the aesthetics of a chart or graphic can be very effective at drawing someone's gaze and even keeping them interested in the piece. The aesthetics, I would argue, is not just a component but a major component. This is true especially in this age of information overload where skimming is taking the place of reading, and since aesthetics can unconsciously draw our attention, it is a major component of any graphic illustration.

Norman

I appreciated Norman's breakdown of representation and how each representation has different properties, thereby making it appropriate in different contexts. His point about representations being more effective when they're experiential rather than reflective was also important. However, in line with Socrates, I think at times reflective interfaces that engage the user are more appropriate than experiential ones. This would be appropriate especially in classroom situations. I'd be interested to examine examples where this might be true -- where the reflective representation is preferred over the obvious, experiential one.

Jeffrey Patzer - Jan 27, 2010 09:11:21 am

Tufte: What I find most intriguing about these three chapters are two specific points. The first point is Tufte's insistence on making graphics, not art. He is disgusted with the way data is manipulated to distort and lie about the message it conveys using flourishes and deceptive techniques in graphics. He insists that if you cut away the junk and lies, provided your data is interesting, that your graphic will be not just useful, but far more interesting. I completely agree that you should not have to include extra distractions within your graphic to make it interesting, if that's the case then you need to find better data which will make your graphic interesting. The second point is Tufte's argument that although a graphic can may be made more visually appropriate, the content may still be complex, and one should not mistake the ease with which content can be understood with the idea that a graphic is easy to view and decode. I think this is important because while you may make a graphic that is hard to understand, it is because you value your audience's intelligence, not because your graphic is of poor design.

Norman: I think the most interesting idea here is that when using different representations of things, whether they are numbers (arabic vs roman) or tables vs graphs, you are trading off certain advantages in each. The most important idea is to figure out what type of display is most important and applicable to the data that you are trying to interpret and display for the viewer. Understanding how the human psyche and visual system works is important in figuring out the visual representation that helps to increase the memory and brain power of the person making use of the visual.

Yotam Mann - Jan 27, 2010 12:05:44 pm

I had never considered how much having a visual aid helps us comprehend the world much better and faster than just keeping it in our heads. I knew it was true from experience: doing math in my head vs on paper, but Norman provides some interesting examples, like the drastic difference between putting the pill schedule in matrix rather than sentence form. He also covered advantages and disadvantages of different visualizations; I think that this was an interesting article that covered the understanding of visualizations which is very significant to creating good visualizations.

Aaron Hong - Jan 27, 2010 12:11:44 pm

The reading for Norman is interesting. He talks about "The Power of Representation," and how often times we can be talking about the same thing (logically), but just because of the representation make it either hard or easy for a person to understand. The two cases are reflective and experiential representation, reflective being the slower one. Experiential representations are often lines, tallies, and other things that are spacial and perceptual. For example, we look at these things and their size for comparison, instead of thinking in our mind of what they represent and how they compare.

Also he goes over some history of representation which comes to an interesting conclusion: "it is things that make us smart." He talks also a bit about how what's hard for human, is easy for computer, and vice versa. And finally he gives lots of examples of how "no single format can ever be correct for all purposes." For example Roman numerals are better for adding, but Arabic numerals are better for multiplication and other tasks. How effectively we can represent things is actually very crucial to our development even as a culture and civilization.

Akshay Kannan - Jan 29, 2010 09:00:32 am

Data-Ink

I found Tufte's discussion on data-ink quite interesting. While ink conservation is important in an economic sense for large-scale prints, I never realized the effect it has on effective visualization until I saw the dramatic comparison between Playfair's two graphs at the beginning of this section. Just from these two visuals, it is readily apparent how much more effective the revised version, using less ink, is at conveying the point of balances moving in favor of England. While minimizing ink helps to keep a visual clean and make trends in data more apparent, additional ink may be needed to make a visual easier to read. Consider the case of a data map of the US. The vast majority of visualizations showing will have the states uniquely colored to show the distinction between them. While this may not carry any additional data with it, it helps to make the graph easier to read and makes trends much more readily visible.

Norman

Norman's article was quite interesting. I found it particularly interesting that humans have successfully exploited external cognitive aids. While traditionally, it easy to assume the egocentric view of humans being the "smartest" or the most "intelligent," I found it quite interesting to realize that the majority of our intellect comes from the use of external aids. With the age of the internet and data being readily available, these external aids have reached a new level of power, giving humans resources to tap into any ideas/concepts imaginable in a fraction of a second, and it is interesting to see the impact that this will have on our long-term technological growth.

Stephen Chu - Jan 29, 2010 11:46:53 pm

Tufte

I found Tufte's data-ink ratio metric to be intuitive and useful when comparing visualizations with higher/lower ratios. His examples clearly showed how more data can be shown with less ink, and with less ink, the data is more quickly accessible to the viewer of the graphic. I particularly enjoyed how Tufte demonstrated that you can add numbers and data by erasing/removing ink (instead of ticks, one graphic used erasure marks). I was surprised how a no frame, no vertical axis, no ticks bar graph was easier to understand than the more conventional bar graph seen every day.

Norman

In last session's readings I wondered why most of us resorted to mental multiplication over pencil and paper. The intro of this reading suggests that we place an inordinate amount of importance on "unnecessary" skills such as mental calculation because mastering such skills is rare. Pen and paper calculation and visually-aided cognition is so much easier that it is overlooked.

This was a great reading to let us take a look from the reader perspective as opposed to the statistical designer perspective. Norman reminds us that readers need to find critical comparisons and relevant variables to be compared, remember the ranking of conditions, and compare the different conditions.

Jaeyoung Choi - Jan 31, 2010 07:21:17 pm

Norman

On pg52, author wrote that, 'things left out are mostly things we do not know how to represent, which is not the same as things of little importance'. Is this really true? I think we tend to leave out things in our representation when they are of less importance. Within limited resources(time, space to represent, or ink maybe), don't we try to organize our representation based on the priorities of things to convey?

Tufte

When I look at several reports that I've written in the past, I could see why principles of data-ink ratio are needed. In some graphs, it seems that I tried to include as much information as I could, which might have made the reader to be felt bombarded with too much info. I guess we need to keep the balance between two goals, maximizing data into an image, and maximizing data-ink ratio. In the conclusion of chap 6, the author claims that there is nothing lost to the readers who are puzzled from frame of dashes in the new design. I disagree with this claim as our reader may get confused and think that newly designed frame has some special scale or info which leads to different interpretation of data points.

Pauloppenheim - Jan 31, 2010 05:26:11 pm

Tufte ch4-6: So many valuable tips, this is where the book becomes "The Elements of Style" for data graphs.

  • data-ink ratio is a recipe for minimalism, i already hear the detroit techno in my head
  • Chartjunk - "grids as garbage" is a total brainshock. It's amazing.
  • OH SNAP dot-dash graphs are like nothing I've seen before, yet completely familiar. (This is what a should have done at work!)

once again, for more.

Stephen Jayanathan - Feb 01, 2010 10:53:55 am

I think a few of the other students have already touched on this, but the usage of white gridlines absolutely looked fantastic. Also the elimination of redundant/unnecessary data looked amazing for some of the images, but others (the raised quartile lines) it wasn't immediately obvious.

On a more practical note, how are we supposed to try to imitate these types of plots? As a common scientist/engineer, even if we know these principles are important, I still find myself struggling to apply them further. I don't exactly have tons of Illustrator skills, or a ton of experience making quality visualizations using other programs. I think we're still limited by the most commonly available plotting apps (Excel, Matlab, etc.) which don't seem to allow the level of customization that is truly necessary to produce a sexy image.

Getting the data-ink ratio to the point where it makes a better visualization at this point still seems to be easiest only if it is a hand-sketch.

Priyanka Reddy - Feb 02, 2010 10:08:29 pm

Norman: I really liked this reading. I thought his point about things making us smarter was very interesting - it's not something I had ever really thought about. I think the other examples he gave to demonstrate how the representation of information affects our understanding of it are all things we do subconsciously, but it was nice to be able to understand some of the theory and concepts behind it.

Tufte: I think the idea of maximizing the data-ink ratio is simple and a good rule of thumb to follow. However, I'm not as convinced about the erasing principles. I think erasing the obvious non-data ink makes sense, but he takes it to the extreme in some of his redesigns. For example, the parallel schematic plot on page 125 that he redesigns certainly looks more sleek, but I think it's harder to read and slightly harder to compare between the box plots since you now only have points to compare as opposed to the horizontal line segments we had before. In addition, his redesigns of the scatterplot were clever in cases where you want to display all that extra information (min, max, average, median, etc) but might be distracting in most cases. Overall though, I think his idea of redesigning some of the classic graphing methods is a good one and deservers further consideration.

Ebby Amirebrahimi - Feb 02, 2010 01:20:13 am

Norman: I loved Norman's high level approach to visualizing data. I thought his historical introduction to the topic set a unique tone which allowed him to examine the further-reaching goals of data representation. The discussion on problem isomorphs was great, and I thought the incremental improvement of representing the flight schedules demonstrated how settling for a poor visual can easily lead to confusion.

Tufte: I was glad to see the data-ink ratio clearly explained. This seems like one of the most important rules of thumb in data visualization, especially when it comes to simplifying visuals to aid understanding.

Kerstin Keller - Feb 02, 2010 11:35:14 pm

Tufte + Data Ink Ratio:

I like the idea of increasing the data-ink ratio in general. However, in my opinion Tufte goes to far when redesigning the quartile plot (p. 124). The plot on the top of the page can still be easily percepted. However, offsetting the middle half and leaving space for the median makes the graphics very hard to read.

Norman:

The example I liked best was the visualization of how to achieve the number 15 with a combination of three number by using a Tic-Tac-Toe game. The idea to visualize the possibilities to get the 15 would have never occured to me, but it turns what seems to be a fairly hard task at first into a very easy one.

Sara Alspaugh - Feb 03, 2010 11:25:19 am

Tufte -- I disagree with some of his conclusions about what looks good, in particular in the data-ink ratio chapter. Sometimes he goes a little overboard with this principle, as with the revised box-plot using an offset line. I thought that looked terrible. However, I did like the white grid. I suppose I would conclude that visualization is as much of an art as a science, though it's not always easy to say what is art and what is science.

Timothy Wheeler - Feb 03, 2010 11:19:54 am

Tufte: Many of his redesigns are visually appealing and clearer than the original. However, his super-minimalist designs, like his modified box-plot, are so subtle that they are hard to reproduce (e.g., by a photocopier or inexpensive printer). Even in modern journals and conference proceedings I see horrible reproductions of the originals.

Mila Schultz - Feb 03, 2010 11:51:46 am

Tufte When reading Tufte, it is important to remember his goals and audience. The chapters on "chartjunk" and data-ink maximization are examples of this; it seems that the goal is to make the data as clear as possible, but this is not always the goal of visualization designers. Though Tufte sometimes mentions his assumptions about the audience, such as in the conclusion of the Data-Ink chapter, he basically discounts any discussion of their limitations when he writes "why not assume that if you understand it, most other readers will, too?" I think this assumption is questionable, as readers frequently do not understand an author who does not factor in the limitations of the reader. I would like to see Tufte discuss how to arrange information for different audiences in particular, especially as he creates charts that are unfamiliar to most audiences.

Norman on Number Representation The discussion relating to figure 7, which explains the process of multiplication in various number systems, clearly highlighted the strengths and weaknesses of each number system. The paper reinforced that the system of representation influences which ideas and concepts are formed and attempted to be represented, which is also touched on in the other Norman reading. In this case, entire civilizations developed certain areas of mathematics rather than others because of their systems of representation of numbers and algorithms, which demonstrates the power of representation.

Shimul - Feb 03, 2010 12:40:54 pm

Norman

Norman argues that reading in itself is not the ideal way to learn, for you should be able to question the words and expect to get an answer for them. It does sound plausible at first but while this may emphasize the need for visualizations, I doubt a static visualization would do any better a job at it than a piece of text. If it is for making the text more readable, a visualization is an obvious first choice; however, it does not serve the purpose of allowing for an engaged discussion either, unless the author is available. However, the article also talks of the power of visualizations in aiding one's mind and that argument definitely carries weight. The examples with the Roman vs. Arabic numbers are good instances. Even as we study computer science, the power of diagrams is prevalent everywhere. Examples include, box and point diagrams and environment diagrams (from CS61A), trees (61B) and all path algorithms (CS170). Without the aid of these diagrams, it would have been much harder to get a grasp of these concepts.

Data-Ink Maximization/Chartjunk

The chapters on Data-Ink and Chartjunk brought out the interesting notion of revisiting a visualization to remove all redundancies and non-relevant data. Decoration of a visualization can help only if the data it represents is substantial. Chartjunk can only provide an aesthetic feel, not back up the data or make it more informational. As such, the aesthetics should be the last concern while making a visualization. That being said, completely disregarding the presentation is also not advisable. But as long as the image/visual looks readable, it should be good to go. The duck example supporting the idea of decorating construction vs. constructing decoration was apt and funny.

Boaz Avital - Feb 03, 2010 09:00:30 pm

Tufte: I love the idea of the data-ink ratio. Improving by removing is a great tool in writing but I didn't expect it to be a great tool in visualization. Though it's difficult to think of a beautiful visualization in a popular magazine following the principle too closely, it would be interesting to see it attempted. I believe what data-ink optimization accomplishes best is allowing you to put even more dimensions of information into your image. Since, with his method, a charting of 2 variables becomes unbelievably sparse, a charting of 4 or 5 or 6 variables like Napoleon's march on Russia does not become overwhelming. I do think he went a little far with the minimalism in Chapter 6 though. the box diagram that is just 2 lines and a dot makes it pretty difficult to compare the size of the upper and lower quartile, and that rug scatter plot was completely unreadable to me.

Norman: The difference between experiential and reflective tools is a good and useful example of making explicit something that was previously implicit or 'just a feeling.' Though he seems to make the case that experiential representations are always better than reflective. While experiential visualizations are decidedly better for smaller datasets and information that must be conveyed quickly, I believe for very dense, complicated data that is hard to visualize, the extra time the user puts in to understand reflective tools could be well worth it, as they are able to represent a very wide array of information. I mean, I believe by his definition that all nominal labelings, such as hue, are reflective, and you can hardly say those are never useful. Also, when he talks about one day electronic devices will allow us to instantly access information any way we want, I think he's being overly optimistic as there is no way they could ever make portable computers like that.

Prahalika Reddy - Feb 04, 2010 01:50:39 am

In Norman's article, I found the distinction between the "represented world" and the "representing world" to be very interesting. The difference between the two concepts is very important, though not explained often. Whenever objects are used to represent other objects, the focus is usually on the object being represented, not as much on the object doing the representing, as it's only a prop. However, as Norman explains, when people "get them wrong, the representation is misleading, causing people to ignore critical aspects of the event or perhaps form misguided conclusions". The actual objects that are chosen to do the representing as almost as important as the objects that are represented.

Norman's discussion about Socrates' thoughts about books was also an interesting read. I don't quite agree with Socarates' ideas that books aren't important and essential tools for thinking and reasoning about things. However, his comparison of writing with painting was noteworthy and accurate.

Tufte's chapters on data-ink and chartjunk were enlightening to read. It's remarkable how the less ink there is, the more impact a graphic may have. It's another concept that doesn't seem to be explained very often to most people and isn't followed for many graphics.

Jonyen - Feb 07, 2010 10:32:44 pm

After reading Norman's article, it seems as though a lot of information visualization is common sense, but I think that Norman goes beyond to explain why it makes sense that one representation might be better than another. The understanding that we live in a represented world and that there is a representing world tells us that representation is indeed a very powerful thing. The representation can tell us a lot of details, and it can hide details, which can help to make things easier to understand. Furthermore, Norman talks about how the choice of representations can be better for a particular task, such as Roman numerals vs. Arabic numerals, or the additive scale for mapping vs. the substitutive scale. All of these examples show how one representation might be better than another. It isn't just arbitrarily decided--rather, there is substance behind the claims that he is making. Overall though, I think that the power of representations lies in how much easier it aids in the computation process and how much time is saved in a more simplified representation of the facts that are present.

Zev Winkelman - Feb 10, 2010 03:21:00 pm

Favorite quote from Norman (46): "During the Middle Ages, readers were taught the rules of rhetoric and were implored to employ them with each sentence: mnemonics, to memorize and learn the material; allegory, to find the multiple levels of meaning hidden beneath the literal text; typology, to think in historical parallels. ... Readers in the latter part of the middle ages did with books exactly what Socrates had claimed was impossible: They questioned and debated each idea. ... Today we may have regressed to match the fears of Socrates: We read too quickly, without questioning or debating the thoughts of the author. But the fault lies not with the book, the fault lies with the reader."

Quite an indictment. Do visualization techniques mitigate this fault, or do they create yet another format for expression of thought which we will consume too quickly without question or debate ?



[add comment]
Personal tools