A Rant About Linear Models of Innovation

Why, oh why, are they still the default for policymakers?

Happy Monday! If you’re affected by the horrible heatwave, I hope you are keeping cool. Stay hydrated, and don’t forget to check on elderly relatives and neighbours!

Today’s post is a rant about linear conceptions of innovation, and how they dominate policymaking even after decades of evidence to the contrary. I don’t necessarily have any policy prescriptions or suggestions in this post, but I want to vent about how bloody stupid it is.

Photo by Denny Müller on Unsplash

Following the Second World War, the dominant model of how innovation happened was one of a “linear continuum where stages of implied causality and progression follow a straight path.“ As G. Bruce Doern, David Castle, and Peter W. B. Phillips wrote in their book Canadian Science, Technology, and Innovation Policy, this idea was

anchored in basic or curiosity-driven research that led to R&D where more applied research and prototype developmental production systems were tested, and ultimately this led to commercialization, sales, and market development. A prevailing, if often tacit assumption has long been that the volume of commercializable outputs from this continuum is proportionate to the inputs, hence the longstanding debate about the levels of investment required to grow and sustain an innovation economy.

However, this model has been known to be insufficient since at least the 1990s.

Geoff Mulgan

recently pointed this out, saying how “decades of study have shown the limitations of this linear model, and the equally flawed assumption that commercialisation and spin-offs are the key means through which R&D drives growth.”

Instead, innovation is inherently non-linear. I love this definition from Richard Owen, Jack Stilgoe, Phil Macnaghten, Mike Gorman, Erik Fisher, and Dave Guston:

Innovation is not a simple, linear model with clear lines of sight from invention to impact, and where accountability for such impacts can be traced. It is an undulating path, sometimes with dead ends, involving many, often loosely-connected actors. It is a complex, collective, and dynamic phenomenon.

Doern, Castle, and Phillips also have a similar, if slightly more academic, framing:

At the centre of this conceptual interpretation is innovation as non-linear and networked interactions, often involving collaborative, networked science, where research and technological change actually occur and are achieved in diverse and even unpredictable ways. The concepts of national innovation systems, local-regional innovation systems, and clusters emerged internationally and nationally to capture views about the non-linear nature of research and innovation.

There is an absolute treasure trove of research and academic literature that fleshes all of that out.

Yet despite that, linear models of innovation still dominate our policymaking. This can be seen in ISED’s 2021 Building a Nation of Innovators. That document includes 33 references to continuums, explicitly says “innovation exists along a continuum,” and gives this figure as an illustration of innovation as a very linear process:

ISED’s Innovation Continuum. Source: https://ised-isde.canada.ca/site/innovation-better-canada/en/building-nation-innovators

Source: https://www.icegif.com/wp-content/uploads/2024/06/facepalm-icegif-3.gif

While the most recent ISED departmental plan fortunately does not include any references to continuums, that linear model is still inherent in their structure and programs. Indeed, their core responsibility in this space is linear in framing “Science, Technology, Research and Commercialization.”

Canada isn’t unique in this regard, even if we are particularly bad, repeat offenders. In Innovation for the Masses, Neil Lee observes how “the idea of an automatic linear relationship, whereby R&D spending automatically generates innovation and then growth, still dominates policymaking” around the world.

The costs of remaining wedded to these flawed ideas are substantial. Not only do they continue to emphasize policy prescriptions that don’t work (more inputs!), but they also exacerbate the siloing of innovation. Too often, innovation policy focuses on narrow technological innovation. But as John H. Howard has argued in the Australian context:

Defining innovation policy narrowly as technological innovation risks neglecting broader forms of innovation—such as organisational, social, cultural, environmental, urban, and service innovation. Such a definition also limits policy attention to measurable technical outcomes, marginalises non-technological sectors, and may stifle cross-disciplinary creativity, diversity, and adaptability, which are increasingly critical in complex, knowledge-driven economies.

Our poor innovation and societal outcomes reflect, at least in part, our inability to design policies that understand how innovation works in practice. Why policymakers still default to these linear models is beyond me. However, this problem does fit into what I’ve argued before: that our policy paradoxes, such as this, are in many ways more pressing than the “innovation paradox” itself.

Before any different policy suggestions can be deployed, we first need to solve the problem of why policymakers continue to follow deeply flawed models after decades of contrary evidence.