Observations on the Manifesto
Posted: September 3rd, 2009 | Author: Manuel Lima | Filed under: Uncategorized |* These are a series of observations on the original Information Visualization Manifesto. I will occasionally reiterate my points, from the manifesto and subsequent comments, for the sake of argument *
It has been great reading all the different reactions to the manifesto and witnessing the open debate within our community. I feel it’s extremely healthy and invigorating. As said before, if nothing else, I hope this will result in a positive introspection of our discipline.
The manifesto has been circulating a lot over the past few days, but I was particularly pleased to read the supportive posts from Robert Kosara (eagereyes.org), and Patricia McDonald (BBH Labs).
The manifesto introduced two different propositions: (1) a list of ten guidelines or principles, that has been for the most part consented by everyone; (2) a more controversial proposal for a stronger divide between Information Visualization and Information Art. I will therefore address these two parts separately.
–
The List
The list of considerations has been praised and well received by most people, but the following points have raised some concerns:
Form Follows Function
Some questioned this old maxim and considered that in the case of Information Visualization it should be rephrased to: Form follows Data. I do oppose to this interpretation. As explained in the manifesto: “Form doesn’t follow data. Data is incongruent by nature. Form follows a purpose, and in the case of Information Visualization, Form follows Revelation.” I provided the wooden chair analogy, but in a reply to one of the comments I provided a second metaphor. I think of data as the unwrapping of a recently purchased piece of IKEA furniture. Looking at all the components scattered across the floor, you cannot avoid but feeling slightly puzzled on what to do next. You can either look at the paper or embrace your creativity and generate an alternative object. But in both cases what will determine the form/shape/layout of the final piece will be the intent. Therefore, many derivations can result from this view, Form follows Intent, Form follows Purpose, etc. I simply decided to go for the most worn out, yet explicit statement - Form follows Function.
Interactivity is Key
This principle merits the reflection of us all. Jerome Cukier and David McCandless challenged the need for interactivity in Information Visualization. In a broader definition of Visualization I would certainly agree with this notion: Information can be successfully conveyed in either static or interactive mediums. However, we have to question what really sets us apart from other parallel fields such as Information Design or Information Graphics. I do believe one of the crucial benefits of Information Visualization is interactivity – which also explains why this area emerged from Computer Science and HCI. It’s this “computer-supported, interactive” visual representation of data that truly makes us different. And this unique offering “becomes imperative as the degree of complexity of the portrayed system increases”. The representation of complex networks is just an instance where interactivity should be mandatory.
The Power of Narrative
This point in particular, should have been read as a consideration, rather than a strict guideline. Nevertheless, “the question of narrative seems to lie at the heart of this Manifesto; the need to pose a specific question of the data and to weave coherent themes and stories from it.” explains Patricia McDonald. Kim Rees and Moritz Stefaner disputed well this prerequisite on every execution, particularly the type of self-made narrative that emerges from exploratory executions. And since we’re talking about analytical tools, this will be a recurrent occurrence. I like to compare this practice to a game designer who lays out an intended context, rules and narrative for the game, but then has this moment of delight when users engender their own narrative, their own path. This is intrinsic to the conception of Information Visualization as a discovery tool.
Look for Relevancy, Aspire for Knowledge, Avoid gratuitous visualizations
As Moritz Stefaner pointed out, these three principles could have easily been merged into one, since there’s a strong overlap between them. However, I do feel they’re individually significant and assertive to merit their own independent call.
Don’t Glorify Aesthetics
This principle has been very debated and since it relates closely to the second part “The Divide”, I will address it in that context.
–
The Divide
The proposed divide between Information Visualization and Information Art was by far the most contentious issue on the manifesto. It quickly derailed to a debate on Aesthetics versus Function and Art versus Science, and we all know how slippery these domains can be. Aesthetics is not the easiest term to define, and neither is Art. I do however look at aesthetics from a functional point of view, and to that extent I also do not appreciate the occasional discredit by the scientific community. If we decompose some of its tangible elements – color, shape, composition, symmetry – we can immediately perceive how aesthetics is an integral element in the usability and legibility of any execution in the realm of Information Visualization. But it’s not the only one.
One of the greatest qualities of Information Visualization, and certainly the main reason why I became interested in the field, is its diversity. It’s able to bring in people from all sorts of disciplines and backgrounds in a remarkably cohesive manner. I look at our practice as a dense voronoi treemap (I could not avoid using this metaphor), where many branches of knowledge come together for the common goal of revelation. This setup works well, when all elements of the equation operate in a sensible way, but when one escalates in detriment of the others, then we have a problem. And lately one end of the spectrum has been pulled in a much more sturdily way. The fallacy of Information Visualization being a conveyor of “pretty pictures” is drastically threatening the field, by undermining its goals and expectations. “We have to fight that or risk the trivialization and marginalization of visualization as an analytic tool”, asserts Robert Kosara on a recent review of the manifesto.
So what do we do at this stage? We either try to restore the balance or we acknowledge a clearer divide. I do not think we can have a convoluted multipurpose all-encompassing practice. This will be detrimental to us all.
As I stated in the manifesto, I think Information Visualization and Information Art can and should coexist, by learning from each other and cross-pollinating ideas, methods and techniques. In fact, I believe this separation is beneficial for both areas, since it frees them from inadequate concerns and aspirations. Information Art can really push the creative limits of data and in the process generate new techniques and algorithms, but also spark public discourse – one of the great qualities of Art. On the other hand, Information Visualization can mature as an analytical tool, providing a reliable and critical source of insight to many future challenges we are still to face.
We have observed a similar symbiotic process between Art and Cartography for many centuries. Several authors have written on this subject and David Woodward, in his Art & Cartography, published in 1987, describes in detail how numerous artists were influenced by cartography, and how maps themselves were hanged in walls as pieces of art. Nevertheless, both fields have always kept their independent paths and individual aspirations.
The divide between Information Visualization and Information Art is not clear-cut and there’s certainly space for a thriving middle ground. Labels can also be changed. Kosara even suggests we start using the term “Visual Analysis” as a substitute for Visualization. This is something we can certainly discuss as a community, and there are many benefits to do so. Once we all agree on what we do, it will be easier for others to recognize the goals and boundaries of our growing discipline.
–
* You’re welcomed to continue the discussion here, or add your comment to the original post on the manifesto *
Hi,
good that you don’t go into defense mode and react so thoughtfully.
However, I think you have not really resolved the issue of aesthetics. In fact, the discussion – as I experienced it – did not derail into “Aesthetics versus Function and Art versus Science”. In fact, my perception was that the strong reaction to the original manifesto was due to the fact that you explicitly introduced these dichotomies (the “pretty picture argument”).
Where I have trouble following, is the following: VC has a long tradition of presenting design approaches to network visualization. You are showcasing particular design solutions using established or new infovis techniques. So, when you publish this manifesto, I assume we are talking about design - right?
I totally agree that many of the works in the area could be improved, by e.g. talking to statisticians, scientists working in infovis, or doing usability studies. On the other hand, these areas could also really benefit by learning from designers’ ways of thinking, working and approaching a new topic, the thoughtful use of visual languages etc.
Now, the “pretty pictures argument” is one I have heard often from people outside design, who often do not really have a good grip on how design works. Simplified, these people see design as making things look good, and art is the stuff that nobody understands and is probably good for nothing anyways. For me, this came across as an undertone of the motivation part of the manifesto, and - coming back to the interdisciplinary exchange: In my view, the argument stresses a cliché that effectively deepens the gap between disciplines. That’s why I don’t like it to generally apply this argument - no doubt it does apply to individual works, but it sounded like you felt the whole field moving into this direction, which I just cannot see.
Maybe it would help to move the discussion to a more concrete level: Which, in your view, are examples that demonstrate the essential infovis characteristics in an exemplary way - and which would be examples that “over-glorify aesthetics over functionality”?
I feel uncomfortable posting negative-like comments, specifically for people whose work I particularly like, like yourself. But as you clearly seem to encourage discussion, I guess I should let you know my (hopefully constructive) disagreement.
Form Follows Data
Let’s take your IKEA furniture example. Following the instructions is not design. Not following the instructions is more interesting: when confronted with a package of undescriptive parts, a designer will typically investigate all the parts and only *then* decide what optimally can be done with them. Even when an intent is defined beforehand (e.g. design an object as high as possible that contains all the given parts, design a chair, …), the design question has not been fully solved. What will be decided to be built, and how it will be built, is mainly determined by the characteristics of the parts, as perceived by the individual designer. If not, you have accommodated engineering, not design (and the brief was so restrictive only a single solution was possible). Similarly, visualization techniques are initially determined by dataset characteristics (e.g. type, size, complexity). You can try as much as you want, but you will not be able to represent a relational dataset with a pie chart.
Close the Divide, Do Not Make One
The fact the discussion “derailed to a debate on Aesthetics versus Function and Art versus Science” is because your manifesto is exactly putting a border between these two. It is ironic that the field which only 5 years ago complained no-one was taking notice of their research contributions, is now considering running from the attention, instead of grabbing and guiding it with both hands. In fact, both information visualization and information art are not getting all the attention (and maybe we should qualify what you mean with “attention”), but the middle ground. Those projects that are innovative enough to attract the attention from lay users, while powerful enough to lead to the discovery of insights. In my opinion, it would be more beneficial to focus how this potential can be exploited for the greater good, and cross-disciplinary collaboration can occur. For example: http://bit.ly/14p2sA >>> http://bit.ly/YjUU >>> http://bit.ly/1OUa2g
What about Misleading Information Visualization?”
The statements “one [information art] escalates in detriment of the others [information visualization]” and “the people who truly believe in the objective aim of Information Visualization are becoming disappointed, discouraged and generally saturated” are provocative as they are completely one-sided. It would be fair if you would also consider the other side, and in particular the danger in abusing (consciously or unconsciously, http://bit.ly/stbqk) so-called “real” information visualization to obscure data insights by ignoring good design practice.
First, I didn’t create a divide. Information Visualization and Information Art already exist today as separate fields of practice. And I don’t see any reason why they shouldn’t. What I said was for projects that don’t follow any of the proposed infovis considerations, a different classification should be devised - therefore the suggestion of Information Art.
Second, I’m a designer by training, trade and heart. I don’t apply the “pretty picture” argument lightly and I’m well aware of the benefits of good design practice. I still remember going to IEEE InfoVis conferences a few years back and be disgruntled with the lack of designers in the audience. Still my background doesn’t bind me from delivering a constructive criticism of our growing discipline.
Third, the aim of the manifesto was to convey a pragmatic wake up call by providing a set of 10 high-level considerations and regaining a sense of purpose that seems to have been lost. It also displayed some discontentment for the mislabeling of art-oriented works as Information Visualization, in part due to the broad overuse of the term “visualization”. In addition, it was meant to be provocative, inspire reflection and encourage discussion.
Now having said that, I would like to take something productive out of this discussion. I think Andrew’s point on the middle ground is very interesting, particularly because it’s associated with a new generation of practitioners, which many of us are part of. I’m not sure if it’s a middle ground or a significant evolution, a “second wave” of Infovis that Moritz alludes to. However, there’s clearly something different about the current state of affairs. I think this could be a good opportunity for us to get together, with anyone else who wants to join, and help define the current landscape. As the field expands, new labels and sub-divisions will inevitably occur. And as many new businesses are being shaped around this emerging new wave, it only makes sense to discuss what are the characteristics, goals and boundaries of our practice. Baby steps towards a theoretical framework of the field. Are you in?
Lastly, on the topic of Form follows Function. I’m afraid we won’t be able to agree on this one Andrew. You certainly cannot build a functional chair from two nails, unless you are as committed as Willard Wigan - http://bit.ly/vwgR9. In the same manner as you’re not able to represent a relational dataset with a pie chart. Data imposes limitations on any intended form, to the same extent that a tool can be a constraint to one’s creativity. However, the final outcome should always be driven by the desired purpose - to reveal and disclose. As Jacques Bertin recognizes on his Semiology of Graphics, for a graphic to successfully fulfill its goals it should be constructed “according to the intended function, that is, according to the nature of the preferred questions.”
By the way, this recent article on Vizworld is quite interesting. An Interview with Edward Tufte: http://bit.ly/T94HK
Hi Manuel,
I’ve been sneakily lurking instead of taking part here…mostly because I just moved from a full time job to a contract position (so more time for thinking about blog posts and getting more involved!)
weirdly enough, I agree with you on some levels, but am a little confused.
for instance, “The fallacy of Information Visualization being a conveyor of “pretty pictures” is drastically threatening the field” — but isn’t the ‘conveyer of pretty pictures’ visualisation art, or information aesthetic visualisation? infovis, to me, always has and always will be led by function, data, and accuracy. if you want to take that comment to the extreme, I have hardly ever even seen a pretty information visualisation!! really! (by the strict definition).
also, I think we’re all a little confused by what ‘data’ means. not to put words in his mouth, but I think what Andrew is trying to say relates to BOTH levels of what ‘data’ means.
- the first is the lowest level of data - the values, the fields, the data types - which determine what mapping techniques you can use.
- then, as you also say - there’s the higher level of data - the insights that can be gained from it, the meaningful message that can be conveyed.
however, you can’t start with a message and mould your data to it. it ALWAYS has to start from the data, and that is why form follows data. would you agree?
in the end, like everyone has said, people need to talk more; communication is key!
Even though a few months have past since the last comment to this post, I couldn’t help sharing three quotes I found during the research for my forthcoming book. They highlight a unified notion of function and aesthetics, which can be useful for this debate.
-
“The care for the beautiful leads us to the same selection as the care for the useful… The buildings we admire are those in which the architect has succeeded in proportioning the means to the end, in which the columns seem to carry the burdens imposed on them lightly and without effort.”
Henry Poincaré (1908)
-
“Moreover, the perfection of mathematical beauty is such, that whatsoever is most beautiful and regular is also found to be the most useful and excellent.”
D’Arcy Thompson - On Growth and Form (1917)
-
“Above all else, a visualization is beautiful when it is useful. A visualization is beautiful when it elegantly and appropriately makes the viewer think about the information organized in the visual display. A visualization is beautiful when it allows the viewer to gain insight and understanding into the information, especially when that information was not appreciated in some other form, such as written prose. A visualization is beautiful when the information is thoughtfully arranged in such a way that patterns and structures are revealed.”
David J. Staley - Computers, Visualization, and History (2002)
-