Killer Trait for Vis Experts #2: Can’t Get No Satisfaction a.k.a. Exploring Alternatives

no satisfaction… but you cannot design a visualization without thinking!

This is the kind of reaction I expected from my last post on the necessary urge to make a first draft. But nobody sent anything along these lines. Who knows … maybe it is just in your heads?

In my last post I argued that the number one killer trait for a visualization experts is to have a urge to create a first draft and I didn’t change my mind in the meantime. The fact is more that as usual every rule has its own counterparts, the necessary weight to put on the other side of the balance to make it work as a fine tuned instrument.

If from the one hand it is of paramount importance to see realized on your screen the images you have in your head (it’s worth repeating it: you cannot hold a visualization in your head), of equal importance is the practice of creating alternatives to your design.

It’s way to easy to jump into the first design that comes into your mind when you craft a visualization and believe this is the best one just because it’s the only one you came up with. I have seen this flaw too many times. In myself and in others as well.

And it’s hard work, psychologically speaking. The main problem is that we tend to follow our intuition and become attached to it. I am not saying that intuition is bad of course! It may well be your strongest weapon, but keeping in mind this idea of alternatives exploration is the skill that will make this weapon create great results. This whole mindset is the trick: NOT being satisfied with the first idea that comes into your mind.

But how do you combine the need to have a first draft with the one of reflecting more on alternatives?

The true answer to this question is: I don’t know. I mean, I don’t have a ready-made formula to offer to you. It’s more about intuition and mastery. As I said at the beginning the trick is to find the sweet spot and stay there. It’s not something static, you have to play with it.

But let me be a little be more concrete and offer some few classic alternatives that I often encounter in my own work. This is very far from being complete, but I don’t know … maybe you have more to suggest? I’d be happy to hear.

Some classic dilemmas I encounter often.

  • Map or abstraction? When I deal with data with an inherent geographical component it is always the same. Do I use a map or I’d rather abstract away from geography in order to have more freedom in the layout? Hard to tell. It depends on a million factors.
  • All in one single view or multiple views? Sometimes I tend to squeeze everything in one single view but there’s no way, it’s too much information. So the alternative is to create several views and try to tell a story through them. I am personally biased towards trying to keep everything on one view as much as possible but sometimes it is not possible.
  • What do I map to what (color, size, transparency, etc.)? Oh this is the most classic I can think of! It’s there all the time. You have a number of data fields and a number of visual features. How do you decide which one goes to which feature? Despite the million words spent to find a rule there is no rule, only some few constraints.
  • Links or matrices? Another big classic! You have a dataset of relationships between entities and you have to decide: link-node diagram or matrix? Mmm … disoriented? I’d bet many of you would choose link-node diagrams by default right? But there are alternatives guys. Many alternatives.

Ok that’s all folks! And what dilemmas do you have? Don’t tell me you don’t have some, I am sure you have. C’mon share them with us, leave a comment below.

If you like this post please don’t forget to retweet it. And if you are here for the firts time don#t forget to subscribe to my feeds and/or to my twitter channel. Thanks!

7 thoughts on “Killer Trait for Vis Experts #2: Can’t Get No Satisfaction a.k.a. Exploring Alternatives

  1. Jan Willem Tulp

    A few dilemmas that come to mind:

    Do I make the most effective visualization, or do I sacrifice some of the effectiveness for aesthetics. This also has effect on the story you are communicating. For instance, a dense network graph may be visualized effectively, perhaps in multiple views to get information from it in a relative effective way, but if your message is that the network is indeed a dense network, than showing the actual density of the network may be a better way to visualize, even though getting information from it may be harder. So, what exactly is the message, story, purpose I want to communicate is another way of saying this.

    The visual encoding dilemma you mention is indeed a good one.

    Sorting could be a dilemma as well. Unless you provide some interactivity to allow the user to create his sorting, this may require some thought as well. For instance, say you have a list of countries with a value for each country, for example, life expectancy. If you choose to create a bar chart, do you sort visually on bar size? Ascending or descending (again, what’s important, longest or shortest life expectancy?). But it may also be the case that users are more interested in there own country, and so their way of searching may be alphabetically. And of course, there are numerous variations: visually sorting the bars, and highlighting the user’s country.

    And another dilemma could be: which information is relevant and should be shown or not. Often you can calculate averages, mean, minimum, maximum, standard deviation, and other statistics. Sometimes it may be interesting to show, but sometimes it could be redundant information. You should not always show something just because it is possible, but it should have a purpose.

    My final dilemma for now: various tweaks, like alignment, padding, margin. Especially if you create multiple views or a more complex visualization (although it may be true for some of the simple visualizations as well) I always spent quite some time on positioning everything. Which view do I show first, what is the order of my views, etc.? How much space between the views? Do I lay out my views horizontally or vertically? What if it’s dynamic, will it still fit for the maximum number of views? Should I wrap? How? When? How do I align the views?

    I think that I can come up with more dilemmas, but for now this is a good addition I guess ;-)

  2. Alberto Gonzalez

    Thanks for the post. As usual, very inspiring!! I agree with most of the ideas/dilemmas you pointed out. Mi view is that despite the common aesthetics & basic principles (minimum ink, small multiples..) each of us has his/her own “tech solution” for every visualization and it depends, at least in my case, of testing the first idea and see if it’s able to tell the story, look nice and fill the space. If the first idea doesn’t match those criteria, re-think and test again. So, it’s simple and complex at the same time, we share common dilemmas but we solve them in a very personal way. I am focused on business information visualization, so I try to keep the “old school” graphs as first choice, followed by new comers as the bullet. I guess for others who are more focused on different information topics have more in mind best suitable graphs for their needs.
    Thanks again for sharing your knowledge Enrico/Jan!!

  3. Enrico Stano


    thank you Enrico!

    One dilemma is now in my mind, the “links Vs matrices” one. I’m working in these days on a social map with 5.000+ 1:1 relationships between almost 500 elements. It would be great to explore the “alternatives” that you was mentioning respect link-node diagrams.

    Bye and thank you,


    1. Enrico Post author

      Sorry Enrico, I forgot to answer to your comment (how could I forget one named Enrico!!! :-))

      Maybe I should write an entire post on this subject. Relationships are notoriously hard to represent. I have a slight preference for matrices because they are less noisy. The point is that in both cases there are tens (or hundreds) of alternatives.

      The whole filed of graph drawing is devoted to finding appropriate layouts for a given problem and there are some many that it’s really hard to know all of them and know when one is appropriate.

      With matrices you have a huge scalability problem and a huge reordering problem. You start seeing interesting patterns only with certain reorderings but there’s no way to say one reordering is better than another.

      I’ll think about writing a post about it with more details ok?

  4. Enrico Post author

    Great comments! Thanks guys.

    @Jan: Great list! You reminded to me of a myriad of other dilemmas I always have, thanks. I agree, this list could go on forever :-)

    @Alberto: I resonate very much with you when you say that the representation has to tell a story. That’s totally true for me, great point. You actually reminded me I wanted to write a post on that too! Maybe another “killer trait” in the list? Thanks.

  5. Sarah G

    Here’s one of my dilemmas.

    Adherence to established graphical standards vs. (re)inventing new ways of representing.

    There are all kinds of visual icons used in data visualization/graphing/mapping/etc. They carry a lot of baggage and established and recognizable connotation and denotations. How far can you push an established icon before you render it meaningless or confusing? On the other hand, if the wheel already exists and there is an established history so people know what to do with it, shouldn’t you use it? How inaccurate or ineffective does the existing method need to be to warrant reinvention?

    1. Enrico Post author

      Thanks Sarah for the question. I think you should always be guided, as a first approach, by clarity and effectiveness. If a standard approach works because it IS effective, use it. If you want to customize it to your specific need do it, but be sure to see real improvements and not just “beautifications”. I think everything is possible: full adherence to the standards as well as far departures as long as everything is done taking the basic principles of visual perception in mind.


Leave a Reply

Your email address will not be published. Required fields are marked *