IN THE early 1990s, 14 computer scientists at the University of Maryland were sharing an 80-megabyte hard drive. The drive was often overloaded, with expendable files taking up space in neglected sub-directories. Finding anything was like blindly reaching along all the branches of an overgrown tree.
There had to be a better way, thought departmental professor Ben Shneiderman. So he wrote a six-line algorithm that visualised the drive as a rectangle. Vertical divisions split the rectangle into smaller ones, representing directories, which then subdivided horizontally to show subdirectories. Each of the smallest rectangles corresponded to a megabyte of storage space, so free space was visible at a glance.
He called his invention a "treemap", and it was adopted by computer labs around the world. It soon found other uses, such as in an interactive chart of stocks and shares, still popular today.
These hierarchical treemaps "epitomize the recent growth of information visualization", writes Manuel Lima in The Book of Trees: Visualizing branches of knowledge. And as big data engulfs labs and lives, the need for such powerful visualisations will only increase.
Lima, a digital designer and information guru, thinks visual literacy, including the ability to express ourselves graphically, is as important as reading and writing. True to this visual orientation, he has provided us with a fine field guide to tree forms past and present in a sumptuous book that places pictures firmly in the foreground. Full-colour reproductions of charts from many of the world's great library collections graphically connect 2014 with the 1000-year evolution of using trees as a mode of visual communication.
Trees were symbolically important for most ancient cultures, often worshipped and frequently present in art. Their association with immortality and their branching structure made them natural scaffolds for genealogies, showing, for example, the lineage of Christ and of royalty. They visually established pedigree and, equally crucial in medieval societies, helped to control inbreeding by showing how closely people were related to a potential spouse.
Yet, as Lima's book shows, the greatest impact of trees was in the realm of taxonomy, as visual representations of abstract religious and scientific concepts. Religion illuminated the way, with 13th-century scribes drawing trees to show relationships between scriptural texts, to aid memory and encourage exegesis – the practice of critical interpretation of texts common in monasteries.
According to Lima, these tree illustrations supported "combinatorial invention and creativity". His idea of exegesis is overly modern (monasteries were not tech start-ups) but it's easy to see how visualisation nurtured more systematic thinking. And, in turn, more systematic thinking nurtured more elaborate visualisation.
Lima convincingly singles out 13th-century Spanish philosopher Ramon Llull as a key figure, whose encyclopedic Arbor Scientiae (Tree of Science) presented a unified vision of knowledge. His 16 domains of science, from the moral to the celestial, are each represented by a branch, and all are supported by a single trunk fed by 18 roots. The roots are also labelled, with nine bearing divine attributes such as wisdom, and nine signifying logical principles, including contrariety.
Over the next five centuries, the roots were pruned, but the tree of knowledge flourished as a metaphor – think "branches" of science – and evolved as a visualisation model.
In fact, the French Encyclopédie, the Enlightenment's foremost encyclopedia, was prefaced by a tree diagram. This schematised its contents with as many as eight levels of branching. Interestingly, in the 1751 edition, the tree was abstract, rendered entirely in type with nested brackets as branches.
The powerful combination of taxonomic complexity and visual simplicity in this tree foreshadows many of the contemporary ones in Lima's book. Here, the typographic tree is one branch of a tree diagram, reaching from medieval drawings to Shneiderman's treemaps.
Lima has skilfully organised his book to reveal these developments. In addition to his section on rectangular treemaps, his chapter on radial trees is especially absorbing, all the more so because there's nothing overtly arboreal about them.
For instance, a "species-level supertree of mammals" shows rodents, monotremes and the rest of us as nodes on a circle. Connective brackets sequentially link up species, genera, orders and families to a common ancestor at the centre. The radius is a 166-million-year timeline, and a smaller concentric circle is drawn at the 65-million-year-mark.
Originally published in Nature (vol 446, p 507), this supertree revealed how little mammalian diversification was affected by the dinosaur mass extinction. In The Book of Trees, it exemplifies the potential of visualisation "to explain and educate; to facilitate cognition and gain insight; and, ultimately, to make the invisible visible", as Lima writes in his preface.
This is all fascinating. But Lima's book lacks balance. Reading his intellectually sparse introductory text and captions, you would never guess that tree diagrams have been criticised by various experts for half a century.
This omission is all the stranger given that Visual Complexity, his previous book, drew on several critiques, including some by French philosophers Gilles Deleuze and Felix Guattari. There, Lima summarised their views approvingly: trees are "authoritarian, unidirectional, and stagnant", contrasting them unfavourably with network maps.
At that time, Lima's contrast between trees and net maps was too extreme and his distinction between them too stark. Now he seems to have shifted to the opposite position.
His examples in The Book of Trees show that tree diagrams, especially interactive ones, can be remarkably dynamic.
However, for any diagram to be properly interpreted, its limitations must be fully grasped so that we know which qualities it can't represent, or, at least, can't represent effectively. For example, the Mercator map of the world is good for sea navigation, but bad for judging the relative size of continents: one of its key limitations is the representation of landmass. Likewise, suppose you wanted to use a tree to show the World Wide Web's structure. This would represent it very poorly, since its structure is a network, not a hierarchy.
The problem of limitations is especially true for tree diagrams that don't resemble trees – and fewer and fewer do. Lima's guidance is sadly absent where it's needed most.
Despite these reservations, there's much to be gained by exploring Lima's trees. What he lacks as commentator, he makes up for as a curator, and his subject couldn't be more apposite.
As data visualisation becomes ubiquitous, we typically look at diagrams as simple infographics. Being reminded of the complex, old-growth forests of medieval scribes or Enlightenment savants cultivates our appreciation of contemporary trees – and may inspire us to combine old forms with new in creative ways.
Keats, Jonathon. 2014. “Why do we love to organise knowledge into trees?”. New Scientist. Posted: April 7, 2014. Available online: http://www.newscientist.com/article/mg22229630.800-why-do-we-love-to-organise-knowledge-into-trees.html