When the military say ‘move fast and break things’ it usually means something different to that in the world of technology. Yet that saying has been adopted from the tech world by defence ministers and officials, reflecting a conviction that government is too slow, too ponderous, and often too bereft of innovative ideas. If only, the thinking goes, government—read Defence—were faster and more ruthless, then many problems would be resolved.
Reflecting on language
Simply picking up the language of the tech world does not of itself confer change. The problems of friction and the fog of war apply no less to technology development.
That said, language does matter. The refrain ‘pick the winners, shoot the losers’ is likely to frighten an already risk-shy public service, increase the chances that more imaginative, higher risk ventures are overlooked, and likely favour existing behemoths, which will argue that while slow and expensive, they are more certain.
But innovation is neither certain nor straight forward. ‘Shooting losers’, as one might in a linear process, foregoes learning from failure and partial progress, critical in innovation. It may well do away with future winners. Repeat disappointment need not be a sign of abject failure; in some cases it may be simply that right solution has not yet been identified, as Edison’s trial of thousands of possible lightbulb filaments famously demonstrated.
‘Picking winners’ implies knowledge and certainty that is at best questionable except in hindsight—maybe. Working out what matters, and what we need to respond to, is tough.
The current environment is characterised by near glacial inevitability, through demographics, for example, coupled with continual but unexpected landslides—financial contagion, pandemics, invasions, and technological breakthroughs.
After all, who in Australian policy circles would have foreseen the emergence of ChatGPT late last year, its rate of adoption, let alone its predilection to hallucinate? Not the Defence Strategic Review leads, for example, who saw little need to discuss AI.
‘Moving fast’ in policy circles also implies a degree of insight and control over the means of innovation that is also, at best, ephemeral.
Those means comprise economic systems, including market mechanisms; relative industrial capacity including manufacturing capability and supply chains; and the creative ecosystem including entrepreneurs, educational institutions, and, importantly, nonconformists, notoriously unresponsive to command authority.
Government officials and agencies may be in a poor position to even know how to move fast in the industrial and technical realms in which they have little experience.
Without deliberative action, it’s more likely that internal drivers of behaviour—the slow-winding constraints of government budgets, financial frameworks, procurement policy, military culture, and, of course, election cycles, along with long-held assumptions—will dictate behaviours.
It’s the inexorable nature of the pacing mechanisms—the metronome beat of budgets, the dislocation of military postings, the grind of procurement, the regular shifting of election goal posts and the inertia of business processes in a big, messy institution—that will likely ‘break’ efforts to innovate.
It’s worth thinking about those pacing mechanisms. They have evolved over time to meet the needs of a stable environment. They ‘fix’ behaviours and impose linearity and certainty, usually from the perspective of insiders. They deliver control to the hierarchy and reward consistency and repetition. And they tend to be impervious to language and dismissive of new ideas and innovation—except for reinforcement.
That suggests if we want defence innovation to succeed, at the very least we need to understand better the tempo of innovation, rather than simply assuming innovation will simply work in current structures.
Roughly, we can identify three time horizons of defence innovation: ‘fast’, ‘intense’ and ‘considered’.
‘Fast’ innovation is short-term. It concerns available technologies of the here-and-now—what users can lay their hands on, typically from currently available stocks and commercial off-the-shelf and ‘MacGyver’ them to meet operational problems.
In short, we don’t have to wait for magical new technologies to arrive to make a difference; thinking about the old and familiar in different configurations matters.
Further, as David Edgerton writes in The Shock of the Old, war has been industrialised and civilianised since the late nineteenth century. Hands-on innovation using readily available materials and components allows testing of ideas, industrial capacity, and supply chains in the short term, providing input into policy, planning and decision-making. It enables a better sense of readiness, rather than awaiting some yet-to-be realised capability, such as nuclear-powered submarines.
The pacing mechanism for fast innovation is immediate user need, at a tactical or operational level. Naturally, the ‘best’ environment for such fast innovation and adaptation is the battlespace itself. Fast experimentation is evident in Ukraine’s use of drones, for example—as are the lessons available to the main supplier of the most popular small, cheap drones, China’s DJI.
Australia could avail itself of a rare chance to test its systems of innovation and production against wartime conditions should it supply Australian made drones, Hawkeis and counter-drone precision weapon systems to Ukraine.
‘Intense’ innovation occurs in the short-to-medium term. It concerns problem sets that can be articulated reasonably clearly, typically at a strategic level, even as the solution has yet to be determined. But they require deep technical expertise, either concentrated in a highly specific field or developed across several domains, as in the Manhattan Project or the invention of stealth aircraft.
Further, these problem spaces are deeply consequential, highly ambitious and more than proofs of near-term capabilities (those are more likely to fall into the first category, of ‘fast’). They also have long, wide tails of consequences—implications for a range of other domains in science and technologies, spurring innovation and economic developments in other fields.
Because most technologies are intrinsically civilian, such ‘n+1’ or ‘n+2’ ambitions will also likely be mostly civilian even as their application and constraints may be military. A system that seizes control of an adversary’s AI and digital control systems might be an example.
The pacing mechanism for intense innovation is defined by the problem but managed by the organisation. That allows for diversity of organisation types, from a Manhattan Project, which leveraged the US Army Corps of Engineers, to private sector’s Lockheed Skunk Works, to British intelligence’s Bletchley Park and Australia’s Counter IED Taskforce during the Afghanistan conflict.
Australia should invest in a number of intense innovation efforts, each targeting an ambitious problem set. For example, Ted Goranson’s situation reasoning would be an ideal candidate for a Bletchley Park. Another could lie in quantum computation once a suitable problem was defined.
Last, there is the ‘considered’ innovation—the longer-term research that’s undertaken by universities and less and less by institutions such as CSIRO. This is the sort of research that may have defence implications five- to ten years, if not 20 years, hence, and the foundational research, such as the physics research in the first half of the 20th century that eventually led to the development of lasers in the 1950s and 1960s. Examples include the work done at for well over a decade now at the University of Queensland on hypersonics, and that has led to Australian quantum computation capability.
This should not be as contentious as it has become. In 1970, Vannevar Bush reflected that after the Second World War, it had been easy to convince the politicians of the value of work of scientists that for two generations had been regarded ‘by many as interesting but hardly of real impact on a practical existence, [but] had been basic to the production of a bomb that had ended a war.’
The wavefront of potential applications is extensive. Rather than trying to corral all such research into shorter defence innovation needs, there remains real value in protecting that long-term research wavefront, while devoting other resources to short term applied innovation. The pacing mechanism for long-term research must be the pace of progress of the science itself.
Australia has not the resources or expertise to cover all possibilities that may emerge in a fast changing, febrile technologically competitive world. There is too much risk in placing all our bets on too few hoped-for magic bullets, especially in the short-term. Choices must be made as to where best to prioritise and over what time horizon.
Sticking with the systematic (slow) pacing mechanism present in bureaucracy and wanting to be associated with clear winners ignores the nature and differing paces of innovation. More will need to be done, but building a model that can be shaped around urgency (within deployments, for example), depth (across the forward estimates, as a minimum), and patience (using something like the Future Fund, as a lever) is a first step.
The idea, then, is not ‘move fast’, ‘break things’, or even ‘pick winners, ditch the losers’. Rather, it is ‘apply, pace and learn’. Less sexy and action-oriented, perhaps, but possibly more assured of outcomes in both the short and long term.