Envisioning Information was recommended to me by a friend as a way to improve how I think about and design visualisations. Although less popular than his other work Visual Display Quantitative Information (which is still on my to-read list), this book has many interesting examples and ideas on how to present complex information. Of particular interest to me, this book gives a large discussion on how cartographers design and present geographical maps. It also gives practical rules on how to design, colour and layout data in a visualisation.
This post is part review, part write-up and part discussion about what I found interesting in this book. Take from it what you will, and if you like or disagree with anything, please leave a comment.
In this book, Tufte gives a wide definition of what a visualisation is. He describes visualisations of maps, train lines, periodic tables, planetary movements, sun spots, scientific data, memorials, calligraphy, engine construction, train signals, river lengths, user interfaces, dance step instructions and much more. The amount of things that he considers visualisations reminds me that many times I fall into the trap of having a very narrow view on what a visualisation is. I have to remind myself that there are more visualisations than just good looking charts or graphs of big data on the internet.
While reading this book it reminded me of the documentary Objectified which is about how everything is designed. Looking around has shown a constant meme in society that visualising data is seen as a worthy endeavour. For example, rather than merely stating that people in Scandinavia are more likely to be blonde, giving a visualisation to make the point is better.
After widening my view on visualisations, I started to realise how often I use them day-to-day, e.g. the metservice's daily forecast:
My folders are also a great visualisation that gives depth to my very flat and sequential hard drive.
These are just examples that are readily available to me now, but there are many examples that go beyond the computer. For example I recently had to construct furniture and the instructions looked something like this:
Goals of a Visualisation
According to Tufte, the goals when creating a visualisation is to increase the number of dimensions on a flat surface (computer or paper) while increasing the density of presented data. This is a 'cognitive art' (as described by Philip Morrison), it is a presentation that is aesthetically pleasing and rich in information.
However, as Augustus Pugin notes:
It is alright to decorate construction, but never to construct decoration
Visualisations are meant to contain information, yet often they are polluted with chart junk (meaningless decoration) that only negates from a visualisations value.
The closer you look at a visualisation, the more information you should get, not just decorated fluff.
Additionally, Tufte says that a visualisation should not be judged by how much information is shown, but how effectively it is presented. Showing to much information can be just as bad at lowing a visualisations value as chart junk.
Tufte also talks about the importance of the visualisations design for its consumers. Good graphic design, typography, object representation, layout, colour, production techniques and good visual principles are required for a visualisation. Any clutter or confusion in a visualisation are failures of design, and not the fault of the viewer. This message echoes the theme of the book The Design of Everyday Things, which is all about adapting to a users expectations and not blaming them for misunderstanding your design.
Throughout the book, Tufte described and defined the characteristics of good visualisations, I have tried to distil what he said into one sentence:
Increase data density and dimension without cluttering with superfluous information or adding unnecessarily junk to the design.
Practical Advice for Visualisation
I suck at design! That is, when I get my ideas out of my head and onto paper (or in the browser) they are not how I imagined them to look. So the main reason why I liked this book is that it had many practical guidelines to follow when designing a visualisation, and designing in general. Tufte, although he gave a lot of advice, drew much of this information from other books like Josef Alber's Interaction of Color and Eduard Imhof's Cartographic Relief Presentation . This external knowledge was distilled and added to by Tufte, which created a great combination of general design information and visualisation specific info. Some of which I will try an present here.
Colour is an important aspect of how a visualisation presents its information. Colour can be used to:
- differentiate - colour to discern between annotation and annotated
- join - colour to show relatedness
- label - colour as a noun
- measure - colour as a quantity
- imitate reality - colour as a representation
- decorate - colour as beauty
Given the importance of colour, Tufte gave many guidelines around its use.
He used the rules from cartographers when creating maps (specifically from Eduard Imhof's Cartographic Relief Presentation) and described their general applicability to all visualisation creation:
- First Rule - pure, bright or very strong colours have very loud, unbearable effects when next to each other, but they have striking effects on a dull background
- Second Rule - bright colours next to white is bad
- Third rule - background should be dull, let the foreground do the work
- Fourth rule - if the picture is divided by colour, put colours from one area intermingled in the other, and all colours should be represented in the background
Some of these rules you can see applied to make this visualisation of marshalling signals more understandable and striking. Using red arrows and yellow sticks to draw the attention, while drawn on the less important dull grey person.
Tufte also described how to layer a visualisation using similar colours. For example, if a river is visualised as light blue then labels for the river should be darker blue to link the two together. Additionally, to draw attention to specific points use saturated red as it stands out as separate from the blue and green layers. This kind of straight forward reasoning about design and colour lets me critically look at my own visualisations, and see if I have used colour to its full efficacy.
Note: I got this book for my dad, and highly recommend it
In addition to using colour, the weight of the line can be altered. In the example below, the thick black line is underlined by the fine red line:
Types of Visualisation
Tufte also described a few different types, or techniques, of visualisations. These examples are practical ways in which to think about layout and presentation of data.
This is where Tufte introduces micro/macro design, what I think of as small picture/big picture. The idea is to have the
same ink serve more than one informational purpose.
Using large quantities of data with a high density, give an overall picture using the smaller ones. This way you can immediately convey the meaning, while allowing for deeper analysis.
An interesting example that was given for this kind of visualisation was the Vietnam War Memorial in Washington D.C. This memorial lists the names of 58,000 soldiers who died in Vietnam, in order of the date they died. Tufte saw this as a good example of a Micro/Macro visualisation because each name has three functions; to memorialise the person who died, to show the sequence of when they died, and to add to the overall visual representation of the number who died. It demonstrates the big picture tragedy that 58,000 soldiers lost their lives without diminishing the small picture tragedy that each one had a name and a story. It is well thought out, powerful, emotive, yet simple.
A Small Multiple visualisation is the same display repeated, but each time with different data. Visualising data in the same manner side by side will highlight differences that are present across the data.
Such a small multiple visualisation can be seen in the Trilogy Meter by Dan Meth.
This visualisation shows the IMDB ratings per movie for many different trilogies. In this visualisation you can see how audiences enjoyed the movies between and within trilogies. An alternative way of visualising this data could have been to display it on a single bar chart. Such a visualisation would be far less striking, more complicated and without conveying any more information than this example does.
For me this book is a great read if you are stuck or in a mental block. It makes me want to go and create something beautiful and meaningful. When describing themes, Tufte often gave lists instead of trying to define something. This is a great method he used to convey something complicated. Here is the list he gave of the things a visualisation can do for you to read and take ideas from if you get stuck:
select, edit, single out, structure, highlight, group, pair, merge, harmonize, synthesize, focus, organise, condense, reduce, choose, categorise, catalogue, list, abstract, scan, idealize, isolate, sort, integrate, blend, inspect, filter, smooth, cluster, summarise
I really enjoyed this book. I spent a long time going over the examples and visualisations trying to extract every point that Tufte made in his writing. His ability to pull apart visualisations and describe why they are effective is a skill I think is worth fostering in a world that wants to visualise everything.
Tufte finished his book with the thought:
Perhaps one day high-resolution computer visualisations [...] will lighten the laborious complexity of encodings - and and yet still capture some worthwhile part of the subtlety of human itinerary.
I like that this became reality.