Having spent my entire career in the field of technology innovation, I have noticed a number of recurring themes that seem to fool all of us. For example, people talk about innovation as if the definition of that word is widely understood. It is not.
It seems a trivial point, but the definition is important. If most CEOs lament about lack of innovation, then they should at least know what they think they are lacking. The reason that so many corporations watch their “innovation programs” fail is that they never understood what they were trying to achieve in the first place, not in any meaningful sense.
Many definitions in the literature merely frame the subject so that the one defining it can peddle his or her pet theories. Innovation consultants and theorists are everywhere these days. Indeed, I am one of them, although I am also an actual innovator, of sorts.
However, given the creative capacity of speech, we are all innovators. Indeed, the brilliant David Crystal researched at length how informal street dialects of certain cultures with relatively low education are on many technical levels (linguistically and cognitively) more creative than so-called Standard English.
All words, not just innovation, are surprisingly difficult to define. Even supposedly concrete words, like “apple,” do not pick out concrete subjects in the world. This is the “referential” myth of early linguists, but also a widely spread dogma that almost everyone assumes about language. The other dogma is that language is a communication system. It isn’t.
If you think that the definition of an apple is obvious, then I recommend two things.
Firstly, read Paul Elbourne’s excellent book – A Slim Guide to Semantics. He spends the opening chapter trying to define the meaning of the word chair, and fails. Try it yourself. For example, you might immediately think of legs, but there are plenty of chairs without legs. And yet somehow we can use that word – and words like “Innovation” – with some degree of shared meaning.
Even Paul seemed to slip into referential modes of thought towards the end of his book when accounting for certain mysteries of how we can understand sentences without the referred objects in sight.
Secondly, consider the Chomskian thought experiment of a machine that produces an apple – i.e. without growing it from a tree. Is it a real apple? Or what about a pear that is genetically modified to look exactly like an apple and even taste like an apple. Is it an apple?
A simple elucidation offered by Chomsky is to consider the prince that turns into a frog. Even a child knows that the frog is not actually a frog, but a prince. It is a quite remarkable insight into the many puzzles of language.
Returning to innovation and its many meanings and mythologies.
Let’s ignore any definition for now and jump to the oft-posed question (in various forms) as to whether or not innovation is an art or a science. Clearly the questioner has certain assumptions in mind, which I’ll get to.
If you’re already a trained skeptic, then you know to avoid any binary question like the plague because almost certainly it will force the mind to ignore a 3rd (or nth) option due to a framing bias.
Oddly, this question is even more burdened by tricky definitions.
If you think defining the word “innovation” is a challenge, try art. And that’s speaking from the experience of someone who invented (“innovated”) a digital art product and has worked with myriad artists and “art world” folks.
But what about science? We might assume, given that science is supposedly the pursuit of precision or accuracy, that the use of this word is less ambiguous.
It depends, as do all words, on context. Take, for example, two common ideas that are perhaps dogmas:
- Science is mostly concerned with finding the truth – i.e. what really happens in the “real world”.
- Science and religion are opposite forces. Refer to Dawkin’s tweets for a pop-analysis.
I am not interested in the basis of these dogmas per se, but how easily they can be challenged by alternative perspectives or frames. Yuval Harari, the popular author of Sapiens: A Brief History of Humankind, gives a plausible account of how science (as an enterprise) is not really about truth, but mostly about the pursuit of power and how, in that mode, it is often co-opted by religious interests. He is looking at science from a historical perspective wherein he is focussed on myths that bind large social groups.
For a quick tour of his argument, read his article published on Salon.
Rather than answer the question of whether innovation is an art or a science, we might, as with all questions, explore motives for posing the question.
My motive is to pose the question as a kind of McGuffin to support the plot of this blog post. I confess!
But when managers and intellectuals ask the question, the motive is often to pre-excuse a possible lack of results. Of course, the phrase “lack of results” is a mystery in itself because that almost never happens. Innovation always get results, just not the ones we want.
This leads me nicely to the core of the question.
When asking the question is innovation an art or a science, ignoring that most of us don’t know what either of those words really mean beyond surface assumptions, in my experience the motive behind the question is to assert the following:
- Innovation is about unknown unknowns (yes – the Rumsfeld kind) and so subject to a kind of magic, often called serendipity or creativity. This magic is essentially a non-deterministic process that we can’t predict and so, like creativity, is a kind of art. (This betrays the fact that most art is actually highly predictable within certain boundaries and often process driven – i.e. not magic.)
- Science is about precision and measurement and, best of all, deploys the dependable scientific method! This is a process. Processes are not magic because you just follow them and, in this case, they produce results. In other words, innovation is not science so you can’t expect results (as if science actually produces the ones you want anyway).
Of course, in this context, the questioner is probably trying to say that when innovating, we can’t apply the scientific method (metaphor for “process” and “results” and “precision”) and therefore the only option is to hope for magic.
Maybe it’s the kind of magic that Steve Jobs and Jonny Ive used to talk about. Ironically, that kind of magic is probably more like the scientific method than most of us like to think. It certainly depends upon massive amounts of quantifiable research, the kind of research that is foundational to their creative engine.
The problem with the “art-or-science” frame is that there really is possibly no such thing as a scientific method to begin with. We only think there is, mostly because we are taught a dramatically simplified history of science wherein scientists posed questions, did experiments and got results all by simply “following a method.”
This is clearly plain nuts.
What method did Newton follow exactly that enabled him to be puzzled by the natural states of objects that had been “successfully” explained for two millennia before him? What method enabled him to somehow imagine the existence of an invisible force, which he himself considered absurd and quite literally described as an occult force – i.e. magic!
Whatever it was, it was not “the scientific method.”
Furthermore, how many scientists actually start with a hypothesis – the holy grail of the scientific method – versus conjure one retrospectively to fit the data once they have developed some kind of insight – often by creative leap (i.e. magic) – that fits the data?
The answer is: surprisingly few.
Even a brief moment of introspection should tell us that this ought to be the case. Our thoughts, even about trivial subjects, never mind complex hypotheses, don’t fall out of our heads in nice linear lines of enquiry. They jump around: “Creative leaps” are just normal parts of human behavior and this is how science takes place. The so-called method is mostly a narrative.
Returning to the art-science question, a key assumption is that art is subject to all kinds of “unknown unknowns” whereas science is about “getting results”. In case I have to spell it out, the question is often used by managers to suggest that there are no guarantees of results because innovation is art and art is not about “results.”
In other words, they are trying to excuse failure from the outset. However, as we shall see, they have misunderstood how both art and science work.
As Fermi said: “There are two possible outcomes: if the result confirms the hypothesis, then you’ve made a measurement. If the result is contrary to the hypothesis [“failure”] then you’ve made a discovery.”
Whilst this notion is often used to validate the importance of failure in science, something which goes unnoticed by innovators who prefer to think about innovation as an art in order to excuse failure, note that Fermi did not say anything about the following:
- How did the experimenter come up with the hypothesis?
- In what way is “failure” actually a discovery?
Maybe you’ve already guessed that the answer to these two questions might reveal that the “scientific method” is more like an art than we first thought.
So regarding the question of whether innovation is an art or science, the answer is almost too obvious to state.
It is both, mostly because these two bodies of activity are similar to begin with.
We should – and often do – use as many techniques as possible from art and science where appropriate.
Most of us already do this perhaps unknowingly or informally, but I think it would be far more useful to innovation teams if they were to be more explicit about which tasks are more amenable to artistic techniques and which are more amenable to scientific techniques.
This will help set expectations and, most importantly, help defend against myriad cognitive biases (e.g. poor framing) that can easily kill innovation projects. Certain parts of innovation are highly amenable to analytical approaches, like science often is, and parts of it are highly amenable to and only available to creative techniques that characterize art.
It is when we fool ourselves into using “creative techniques” (which often default to crude heuristics) in place of analytical techniques and directed experimentation that we can send our innovation projects into oblivion (long before they actually fail).
Let me give some concrete examples of when innovation is art and when it is science, although hopefully by showing that the edges are always blurred between these two approaches.
Leaving aside definitions of what art is, we can agree uncontroversially that it is largely about aesthetics. Much of what gets discussed as “design” in the software industry (like “UX design”) is involved with aesthetics. Arguably, even the more technical aspects like usability are about aesthetics in the broader sense. Interaction is a form of aesthetic experience in addition to its utility.
Within art, much is known about certain aesthetic universals, like the principles of symmetry, color, harmony, gestalt and form. Indeed, as a metaphor, we could say that all design is concerned with finding “the golden mean” of a product.
So, when thinking about design (of products and services) it would certainly pay to have an understanding of aesthetics. In this sense we are borrowing from art even though many of the principles of aesthetics can eventually be stated as rules (i.e. more like following a process than a creative leap).
The creative leap comes into play when trying to cross the considerable chasm between aesthetic principles and the particular configuration of aesthetic ingredients that will achieve the golden mean – i.e. how to use aesthetics to make a compelling product.
It is clear that this process is highly variable and fraught with all kinds of mistakes. Most products fail, often more to do with aesthetics (in its broadest sense) than utility.
The first mistake is confusing artistic merit with artistic efficacy. They are very different.
Artistic merit is something that mostly only the artist will understand, often embedded as highly intellectual content that other intellectuals will appreciate. Indeed, much of the “art world” is about intellectual exchange.
Artistic efficacy is about the success of the art given some narrow purpose, which in the case of a product is to attract buyers. When attracting buyers, some of the product’s “content” has to appeal to what the buyer likes from his or her perspective.
One way to explain this is to consider decorating with art, a subject that I know much about from my activities as Director of Labs as Art.com.
People who decorate do not know much about art and are not interested in its intellectual content. People who know about art are very few. This does not mean that a folk-decorator will not appreciate good art. They do, and usually for two key reasons:
- Universal aesthetic principles (like color, harmony, form, symmetry etc.) that the artist has used and that our brains are “hard wired” to appreciate (mostly for evolutionary reasons).
- Essentialism principles wherein the product has a hidden “essence” that speaks to the user, such as it’s by “a famous artist” or is “a well liked piece” etc. These are really elements of “brand” or “myth” that our brain is “programmed” (rather than “wired”) to appreciate (via culture). For example, the myth that expensive bottled water is some kind of self affirmation (of health or wealth etc.)
But the key reason that a folk-decorator (“product user”) favors a particular piece of art is: “I know it when I see it.” This applies not only to art, but to products and services generally.
I am puzzled by the number of folks who think that Lean product development revolves around putting a low-fi prototype in front of customers and asking their reaction as if this is the core of “know it when I see it” without any prior regard to universal principles of aesthetics, product content or context.
This approach relies solely on a kind of information exchange (almost at the level of utility) and ignores how far essentialism or artistic merit can play a role in artistic efficacy – i.e. the many ways they could have presented the “same” low-fi prototype and got a different reaction.
Even more interesting is how Lean is often presented as a kind of pseudo-scientific technique. It even borrows language from the so-called scientific method, like “hypothesis” and “testing” and “data” etc.
Many Lean advocates fail to mention anything about research and figuring out some kind of model for how a user ought to react to certain ideas before presenting product proposals: a kind of “theory of product.”
A model is very different to a hypothesis. A model is more like an argument with evidence from other fields, and even available data about the field itself, all pointing to why a certain idea (which might eventually surface as a hypothesis) should work or not because we are able to predict the effects of its constituent parts (components of the model).
However, this is not easy and requires knowledge from a broad range of topics and a keen ability to synthesize according to the underlying principles of those topics. Underlying principles are almost never taught in schools or colleges, so no surprise that we are ill-equipped to identify them or even think to identify them. People who can build these mental models are called interpreters by Roberto Verganti in his excellent book Design-Driven Innovation.
Unpacking the phenomenon of “I know it when I see it” is where we can cross back over from the field of aesthetics to the domain of science because there are methods available to predict the outcomes of the “I know it when I see it” experience. Sometimes we call this “user-centric” design, but that is yet another misleading definition because many great products are as much “designer-centric” or “culture-centric” as they are “user-centric” – i.e. they serve more than one master or user.
The interpreter interprets and then articulates, or proposes, the user’s tastes on their behalf – i.e. helping the users to see why they need a product feature/benefit/myth, or guiding them towards a particular “know it when they see it” experience.
In other words, it’s like knowing which ingredients you could show a customer that would satisfy their “know it when I see it” affirmation (“artistic efficacy”), but only showing them those ingredients that you want to show them because these are the ones that you know how best to deliver (“artistic merit”) because you are able to interpret a “theory of product” from the available data and aesthetic principles.
Research plays an essential role in interpretation and is very much part of the scientific method, such as extensive background reading in order to form strong mental models of what you think is going on in a particular field based on what others have done before you. It is no coincidence that Jonny Ive designed the original iMac after he researched methods of soft-forming, like how jelly beans were made. (By the way, artists also conduct research.)
One important method of research is making use of available data via various predictive methods to find patterns in the data – i.e. attempts to reveal the unknown unknowns. A competent data scientist can often reveal far more insights reasonably efficiently than someone spending months conducting user-intervention methods like surveys and focus groups.
The analytical techniques of a data scientist are highly scientific, or technical. However, it is worth noting that the job title “Data Scientist” is hugely misleading (and fungible). As you no doubt have guessed, the word “scientist” is problematic.
The only thing scientific about analytics and machine learning are the algorithms and processes themselves. However, they are still driven by the creative act of forming hypotheses, many of which “reveal” themselves by merely “playing around” with the data. Indeed, the first step of data science (after cleaning the data, which is a key part of science) is to “play with the data” in order to form a view of what it “says” and what it might “say” when subjected to certain modes of analysis.
And when playing with the data, who is to say what kind of creative leaps might occur just like when Newton found himself puzzled by the natural state of objects. In my view, innovation projects should assign “play time” wherein the innovators play around with whatever data they can muster about the problem space. There are various techniques for “playing” that we can borrow from data science.
One goal of playing around and then performing hard analysis should be to determine the “principle components” that most heavily affect user perceptions. In other words, which parameters of a product (or problem) space are users most sensitive towards and which of them are principle rather than tertiary (i.e. possibly causal versus correlative). This helps to figure out which of the universal aesthetic principles apply to the problem space, or in what combinations.
By combining these more scientific techniques of analysis with universal aesthetic principles (in the broader sense of aesthetics) we are better equipped to perform what really happens at the core of all creative endeavors, be they artistic or scientific or anything else: namely interpretation.
When you assemble all of the available artistic elements, like aesthetics (and there are many more) with scientifically curated data (like principle components) you still have to interpret what it all means and how best to combine the knowledge into a product that has both artistic merit (enough of it) and artistic efficacy.
This process of interpretation is what artists and scientists do alike. It is their core skill. And it is often done best by individuals who are so well grounded in various intellectual theories that they can literally see what others can’t. They have a higher order creative synthesis capability that is perhaps impossible to articulate. However, this does not mean that it is not informed by accessible artistic and scientific principles. It is not voodoo!
To illustrate the interpreter’s skill, I am tempted to use the cliche metaphor of the jazz musician, which I happen to think is particularly apt in a number of ways.
Firstly, unknown to many outsiders, the best jazz musicians are well drilled in many non-jazz musical motifs. For example, many of Oscar Peterson’s piano studies and exercises (piano lesson here) are classical motifs. He was heavily influenced by the classical composer Rachmaninoff. He also practiced an insane amount of hours, many of them playing classical piano, which I would regard here as “background research.”
In other words, Peterson was widely versed in the broader aesthetics of music theory and was able to interpret how to bring these components together in moments of contemporaneous innovation: jazz improvisation.
Secondly, jazz (and music creation generally) is an unbounded generative capacity of the human mind, similar to language, which is where I began this post. My own theory (and informs partly the work I do in trying to formulate computational models for creativity) is that the interpretive capability is highly related to a language capability, although limited mostly to the sub-conscious role of language (its primary role) which is an instrument of thought.
Why this is interesting is because just like any musician can increase his or her skills through deliberate practice, I believe so too can any innovator. However, we are yet to see much work in this area regarding what drills, exercises and motifs could one learn and practice to become a better innovator (or thinker for that matter).
To summarize, the frame of “art or science” applied to innovation is almost meaningless. This is not because innovation, as I hope I have shown, involves elements of both art (like aesthetics) and science (like data analysis), but more instructively because science and art are, at their core, similar processes of creative leaps. The so-called scientific method is merely a narrative that explains post-hoc what a scientist did to arrive at some kind of conclusion. It is not really a method at all.