I read an Entrepreneur piece by Julia Prakapovich titled “The Innovation Arbitrage Everyone Ignores, and Why It Creates Breakthrough Companies” and it gave me a useful phrase for something I see constantly in large organisations: innovation arbitrage.
The idea is simple. There is a gap between what is technically feasible and what an organisation treats as feasible. The gap survives because closing it creates friction in all the places people prefer to keep smooth: governance, risk posture, budgets, ownership, incentives. “We can’t” becomes a reflex. After a while, it even starts to sound like a law of nature.
In IT and transformation programmes, the pattern I see is painfully consistent. Everyone asks for speed, resilience, better data, more automation. Then the same operating mechanics stay in place: fragmented decision rights, project funding that ends at go-live, exceptions that multiply by geography, metrics that reward safe delivery more than durable outcomes. The system behaves exactly as designed. A lot of value sits inside that gap, waiting for someone to treat it as an execution problem rather than a cultural debate.
DARPA
DARPA is a US government agency created in 1958 to fund high-risk research with potential strategic impact. It is small, it places bets with uncertain outcomes, and it is built to learn fast. What matters for me is the operating logic. DARPA has a habit of tying ambitious goals to real constraints early, then forcing the work to meet reality quickly: prototypes that have to work outside a lab, interoperability issues that cannot be wished away, engineering trade-offs that must be made explicit. It pushes multiple disciplines to converge around the same outcome.
ARPANET is a classic example, because it started as a very practical challenge: get computers in different places to communicate reliably, even when parts of the network fail, and do it in a way that can scale. DARPA’s own history highlights how packet switching existed in different forms, and how progress came from making systems interoperate and moving toward protocols that could connect heterogeneous networks. It morphed into our current internet.
For an enterprise audience, the lesson is not “copy DARPA.” The lesson is that breakthroughs become much more likely when you couple ambition with constraints early, and when you organise the work so that learning compounds instead of getting lost in handovers.
The clean pipeline story slows people down
Corporate innovation often behaves like a relay race: research produces knowledge, applied teams produce POC’s and prototypes, delivery industrialises, and everyone hopes the baton handovers stay clean. Entrepreneur points to Pasteur’s Quadrant as a better lens, where work aims for deeper understanding while being anchored in real-world use.
This matters because the “relay race” model encourages a kind of polite separation. People learn in one corner, ship in another, and reality shows up late. Then the organisation discovers what it actually built: partial adoption, weak data quality, processes that remain broken around the tool, and unclear ownership once the project team disbands. Over time, the company concludes that breakthroughs are rare. Another interpretation is more useful: the operating model makes breakthroughs hard to land.
What I optimise shapes what I am able to see
Optimisation is never neutral. It changes perception. When I optimise for on-time delivery, I will get on-time delivery. I will also get local workarounds, exception mechanisms, and technical debt that quietly becomes tomorrow’s constraint. When I optimise for uptime, I will get uptime, and I will also get reluctance to change, fear of upgrades, and a bias toward solutions that feel safe. When I optimise for compliance, I will get compliance, and I will also see ambition reduced early so risk stays manageable.
Those objectives matter. They just do not solve the leadership problem when the strategic requirement is speed, resilience, and competitive differentiation. In that context, the real question becomes governance: who has the mandate to challenge the inherited assumptions that are slowing the company down, and who funds the capabilities that reduce the cost of change.
Where Dandeleon helps me in practice
This is where I use the Dandeleon model, in the sense Jeremiah Owyang described it: a multiple hub-and-spoke structure, designed for organisations with semi-autonomous units that still need a coherent core. I apply it as a way of structuring problem-solving and decision-making around constraints.
I start with multiple hubs around a strong central one, because global reality has multiple centres of gravity: business lines, countries, value streams, major platforms. Each hub owns an outcome in plain operational terms. Cycle time, cost to serve, forecast accuracy, customer effort, risk exposure, operational stability. If I cannot name the target condition and how we will measure it, the rest is theatre.
Then I run the spokes systematically, because constraints hide in predictable places. Data contracts and definitions. Security friction and access patterns. Platform fragmentation. Ownership gaps between run and change. Approval chains. Local exceptions. Vendor constraints. Skills and operating cadence. I force evidence. I separate constraints that are physical from constraints that are institutional. I also separate constraints that are real from constraints that have simply never been tested since the organisation last changed shape.
The output is a short list of constraints we can collapse, the capabilities required to collapse them, and the governance changes required so they do not reappear in six months under a different name.
Ghost constraints are expensive because they inflate the cost of change
In global organisations, ghost constraints travel well. They pass from one leadership team to the next because they sound prudent: “every country is different,” “security will slow us down,” “data governance takes years,” “legacy is too risky to touch.” Sometimes they are true. Often they are true only because we keep reinforcing them with the way we fund, measure, and organise.
That is the real arbitrage. The value is not in having more ideas. The value is in reducing the cost of change so the business can execute more strategies without multiplying complexity at every step.
This is why platforms matter. Identity and access that behaves consistently across geographies. Integration patterns that stop every project reinventing the wheel. Observability that gives operational truth across the estate. Data products with clear contracts and ownership. Resilience that is designed and rehearsed. These are the mechanics that turn ambition into delivery, and delivery into a repeatable capability.
Ghost constraints thrive on vagueness. The cure is brutal clarity: define the outcome, map the constraint, assign ownership, fund the capability, remove the exception. Repeat. After a few cycles, the organisation stops saying “we can’t” and starts saying “we chose not to, for now.” That is a healthier sentence.
