ORDER/SUBSCRIBE           SPONSORS           CONTACT           WHAT'S NEW           INDEX/SEARCH

Why Greatness Cannot Be Planned: The Myth of the Objective

by Kenneth O. Stanley and Joel Lehman
Springer, Heidelberg, Germany, 2015
141 pp. illus. 11 col. and b&w. Paper, &29.99
ISBN: 3319155237; ISBN: 978-3319155234.

Reviewed by Harold P. de Vladar
Parmenides Foundation
Art & Science
University of Applied Arts Vienna


If you ever wondered what creativity is, this book will ease your curiosity about it. Why Greatness Cannot Be Planned argues that in order to fulfill ambitious goals, we must give up aiming for specific objectives. Instead, we should open ourselves up to an unexpected, more fascinating and progressive undirected exploration of new possibilities. Stanley and Lehman, researchers in artificial intelligence, argue that because innovation involves complex stepping-stones that cannot always be anticipated, we should focus on exploring new possibilities-regardless of how these perform according to preconceived standards (objectives). This ensures new creations, some of which will be highly valuable. The view advocated in the book is so obvious and inconspicuous that it passes unnoticed, but realizing it is enlightening. In the words of the authors: "All of us can transform the present into the future. None can transform the future into the present."

It is a provoking read, focused on a rationale that at first sounds simplistic and even counter-intuitive, but one of which is well-known to artists. That is, how the free exploration of themes eventually hits the creative jackpot.

The authors first focus on a couple of computational models of their own which explore the 'novelty search strategy'. One of these models, picbreeder [1], is a game where one or several images are picked and somehow combined to give an offspring population of new images. The process can be iterated ad infinitum. The authors explain why in this game it is impossible to breed a specific preconceived image. In other words, you cannot decide beforehand whether you want to breed an image of, say, an elephant for example. Instead, images are bred without aiming for any particular design and consequently, truly enthralling pics will be discovered, such as alien faces, skulls cars, etc. I tried picbreeder, aiming to evolve a flower (a rather simple albeit pre-set objective). I avoided novelty, i.e. getting distracted by any other images that would catch my attention. Yet after more than two hundred clicks, I quit because having been unable to reach my objective, I got bored. This toy model certainly illustrates the central point of the book.

This novelty search strategy is explained by invoking a metaphor: a treasure hunter who does not seek for something in particular (he cannot know what treasures he will find). His virtue lies in skillfully collecting various valuables (stepping-stones) by searching in different places. They give the example that making a TV could not have been a realistic objective for cavemen since the stepping-stones to TVs are not TV-like objects. Instead, first electricity had to be discovered, cathode tubes invented, and so on. Similarly, I could not possibly have 'picbred' a flower because the stepping-stones from the initial pics were not flower-like.

After dealing with the toy models, several chapters discuss why focusing on objectives can be deceiving (like me 'picbreeding' a flower), or even disastrous. There are strong and direct critiques to controversial issues such as education, the functionality of the scientific community and others. These critiques are valid and necessary. They explain how focusing on indicators (strictly speaking, objectives) can lead to misguided results. This happens when instead of trying to improve the systems themselves, policies target improvement of indicators. The outcome is often manifested in strategies that increase these indicators but allow the problems to become worse. An extreme example that they propose is to breed 'intelligent bacteria' by applying an IQ test to the microorganisms. However ludicrous it sounds, the problem bears a deep truth as the stepping-stones to intelligence are not intelligence-like and cannot be assessed through objectives such as increasing IQ. Intelligence could arise only through non-directed exploration, implemented by biological evolution.

Along these lines, a complete chapter is dedicated to discussing why education fails and how schools and colleges could be much better off by abandoning scoring through indices that rate their performance. Another chapter they discuss in detail plays on the 'rituals' of scientific functioning. (If you are a taxpayer, you should peep into this and demystify how science works!). These social apparatuses [2] have their own ways to measure objectives, thus, in the authors' own words: "[...] the focus is always on the ultimate destination rather than on interestingness or novelty [...]. So it can't be a treasure hunter-" (their emphasis). Similar cases are made for technological innovation (e.g. the TV example).

I personally found these ideas to be largely one-sided and exaggerated from a very simple computational toy-model to big issues. Therefore, the book proposes 'solutions' to many issues based on the notions of novelty search but I regard these to be a bit simplistic. However, to be fair, proposing solutions to social problems is beyond the scope of the book.

The last two chapters rethink two fascinating subjects. The first one is a reinterpretation of the theory of evolution in terms of the novelty search process. This is very welcoming as the theory of evolution largely assumes objective optimization. The authors take the chance to clarify some popular misconceptions of the evolutionary theory.

The second study case addresses the field of artificial intelligence. A.I. basically designs algorithms that are 'smart' according to a handful of test problems that are a consensus of the research community. Paraphrasing the authors, research in A.I. should focus not so much on algorithms that perform better, but rather on algorithms that lead to consideration for new algorithms.

It is unclear whether the myth of the objective extends as a metaphor, of which the big problems in life, science and society represent a true example. Since one cannot prove a negative, it is hard to dispute failed attempts for innovation. Furthermore, due to the fact that there are always stepping-stones, be it biological, technological, artistic, scientific, social, etc., it is hard to argue against the myth of the objective. Irrespective of that, I believe that they are largely right. Moreover, the views advocated in the book can be very useful for understanding a big deal of what happens around us.

The writing style of the book is slightly dry and formal, so at first read it is not particularly catchy and entertaining. However, due to the short length of the book, it does not become tedious. Overall, the piece is clearly written and the message is delivered very well, clearly demonstrating its value. An interesting fact is that this book is a product of such serendipity, which is a clear demonstration that the non-objective search itself leads to interesting outcomes!

I personally liked the book as it made me think and question several aspects of my own research subject in evolutionary biology, of my artistic practice and of my life and career. Although it certainly did not distract me from pursuing my personal ambitions (this dispelling is anyhow not the point), it did provide me with a new perspective from which to look through.


[1] See http://picbreeder.org

[2] G. Agamben, What Is an Apparatus? And Other Essays (Stanford University Press, 2009).

Last Updated 1 August 2015

Contact LDR: ldr@leonardo.info

Contact Leonardo:isast@leonardo.info

copyright © 2015 ISAST