Path Dependency, Infrastructure, and Entrepreneurship

Every day, businesses take action because of infrastructure decisions made long ago. Venture capitalists rely on it. The theory of disruption presupposes that large companies will never look at markets with the eyes of a new organization. Instead, leaders of these organizations look at opportunity through a lens of resource optimization; trying to make the most money out of the least incremental investment.

But despite the handcuffs that tend to come with billions of dollars of outdated investment, attacking huge companies saddled with old assets isn’t always a simple decision. Sometimes the sheer amount of infrastructure in place across a single industry makes it nearly impossible to dislodge any one player. And for anyone setting out to tackle these monolithic industries, it’s important to understand why.

How infrastructure dictates strategy

Everything about who you are today is a result of where you come from. Some of the contributing factors to your position and perspective are more substantial than others - but the fact that your past plays a big role in your decision making process must be irrefutable. Businesses are no different. The decision of a telco to invest in a GSM standard isn’t one that can be reversed lightly - it informs a firm’s strategy for years.

In the mid-1980’s, an economist named Paul David introduced the world to a theory that would describe this phenomena. He called it path dependency. His ideas detailed how investment and standard setting in industries ultimately dictated the future of an industry. Once enough critical mass had been put behind a given technology platform, it was possible to predict the incremental investment companies would make - regardless of the quality of the platform.

The best example of path dependence is the Qwerty keyboard. If you’ve ever wondered why your keys are laid out in such an odd way, you wouldn’t be alone. In the early days of writing’s mechanization, typewriters frequently jammed. When keys were pressed too quickly, the typist would end up with a tangle of mechanical arms instead of a legible sentence. The Qwerty keyboard was developed to slow the process of typing by introducing the least optimal layout of keys. But even after computers became able to handle lighting fast inputs, the keyboard remained the same. Why? We’d invested so much in training people to type on an inefficient Qwerty keyboard that training people to type on any new format was always considered too onerous.

Today, the Qwerty pattern of path dependence holds true in many markets. Car lanes are a certain width because the carts that preceded cars needed to be the width of two horses. Healthcare maintains outmoded process and pricing methodologies because that’s how a wealth of doctors were trained decades ago. And many fortune 100 companies still have COBOL applications in use… a mainframe language that most current computer science majors haven’t ever thought about.

It’s common and understandable. Executives will invest in a production system and then spend years attempting to amortize its cost, regardless of new technologies that might emerge on the scene. Clay Christensen, Willy Shih, and Steve Kaufman called it the threat of marginal thinking. On the increment, the wrong investment seems like the optimal decision. Squeeze every dollar out of what you can - regardless of what advantages new platforms might yield.

But it’s not as simple as building a better business

Unfortunately, as an innovator, the dynamics of path dependence don’t uniformly run in your favor. It’s not enough only to build a disruptive business on the back of a scaleable technology advantage. Depending on what industry you’re sitting in, upending your competitors often requires a whole lot more than you’d imagine. And unless you are prepared to deliver all of what’s necessary, stumbling blocks await.

Consider alternatively powered cars; whether they’re electric, diesel, or biofuel. Companies have popped up and floundered for decades trying to unleash a more fuel efficient, environmentally friendly, vehicle. Until Tesla, none had succeeded. In part, Tesla’s strategy reveals why.

The automotive industry is highly interdependent. Despite the fact that Ford and General Motors don’t own gas stations, their cars appeal to us in part because of the massive amount of infrastructure investment made by gas station owners. A wealth of trained mechanics help us get our cars repaired when we need it - even if there is no conveniently located Ford outlet. And a vibrant reseller market helps us manage our long term costs of ownership.

Introducing a technical platform for a new car requires a lot more than introducing a new car. It requires delivering the same ecosystem of services that make the old platform appealing. This is why Tesla is investing in their Supercharger network, operating showrooms, and helping to manage used car sales. Even though it’s possible for entrepreneurs to predict that Ford and GM will be slow to adopt new platforms, it’s vital to anticipate the same inaction from their ecosystem. Without a strategy to overcome such barriers, it’s hard to topple the status quo.

The reason venture capital tends to flock to biotech and IT is because the industries are fairly modular. Single layers in the tech stack can be replaced easily. When it comes to places like the utilities or energy industry, it’s a whole different story (despite the inevitable march of modularity).

In these industries, the only way disruption can be pervasive is if the new model can grab hold of a small market and grow out. Containerized ships emerged on a single east coast freight route. Containerized ships relied on the rest of the ecosystem to transport their goods - even adapting their early architecture to ships operated during World War II. In order to ensure that people continued to get their goods around the globe, even the most disruptive change to emerge in the history of international freight had to plug into the existing model and slowly grow out. There was no possible justification for complete obviation of the billions of dollars of assets owned by the existing merchant marine. Transformation took decades.

Knowing where you fall on the spectrum

I’ve been spending a lot of time recently grappling with how to look at different markets in relation to this problem. The framework that I’ve come up with is rather simple in explanation and difficult in implementation.

Namely, two concepts that dictate the way disruption will play out in these types of situations:

  1. The fixed cost of deployed infrastructure in the market

    The fixed cost of deployed infrastructure is an indicator of the amount of path dependence in a given market. Things like property, plants, and equipment must be considered here. But other intangible fixed investments must be considered as well (i.e., typing training, etc.).

    In markets where there is a great deal of deployed infrastructure, the decisions of existing players is likely to be more predictable. It’s easier to play traditional disrupter when incumbents are tied to obsolete legacy assets.

  2. The level of interdependence in the market

    The challenge of looking at your competitors’ deployed infrastructure alone is that it often ignores the level of dependence ecosystem participants. When you look at your competitors’ partners, suppliers, and customers - it’s vital to gauge how much they’ve invested in the existing ecosystem. If they’re tied to the fate of the people that you’re trying to displace, it means a harder go of it all.

    Sometimes the burden here can be overwhelming, simply requiring too much investment for any single entrepreneur to overcome. Even if, for example, a company came up with a great alternative to the US highway system. Unless the new solution required orders of magnitude less capital to replace the existing infrastructure or could be deployed in microcosms of the market, it’s unlikely that it would ever dent the market.

When you consider these two influences on the course and predictability of disruption, you get a simplified set of situations. Namely, you find markets where disruption will tend to run rampant where infrastructure costs are high and interdependence is low (i.e., Hard drives). You find markets that require full stack innovators to attack an entire ecosystem, where interdependence is high but infrastructure costs are relatively low (i.e., automobiles). You find markets where both infrastructure and interdependence is high, often resulting in collective inaction (i.e., Utilities). And finally, you find very adaptable markets where there’s low infrastructure and interdependence (i.e., professional services).

Knowing where you sit tells you whether you need to integrate up and down the stack, delivering everything from product to post-sale services. It tells you whether you can expect collusion from incumbents. And it tells you whether venture financing or public investment provide more realistic paths towards rollout. It’s not as simple as just having a superior technology. If you’re in a market where you’re fighting the entrenched interests of the entire ecosystem, having a 10x cost advantage relative to your direct competitors might not even be enough.

Even if they don’t articulate it, investors and corporate financiers are thinking about this the minute you walk in the door. You should be too.
12 responses
How do off-book disbenefits or bads (call them ignored, unrecognized, unmeasured or misunderstood negative externalities) of a project or an infrastructure ecosystem affect the continued existence of the (typically protected or shielded) system that otherwise should be doomed and replaced, but that instead enjoys an existence or fundamental role much longer or greater that its true fully accounted value or cost-benefit would allow? I.e., Path Dependency as a concept works as a pseudo-mathematical didactic or graphic, but not as a fully or even reasonably informed market mechanism artifact. Path Dependency in its simplest form is like a Majority Report / Winner-takes-All pronouncement that is obvious, prima facia, with no experimental, methodological or procedural finesse or stewardship regarding other findings (i.e. no Minority Report as a simple metaphor)... If Path Dependency could be paraphrased as "You are dependent on how you got there," it also should be characterized as well as "You also are dependent on what you did not notice, did not know, did not realize or did not do." I.e.: What did NOT happen? And, regarding "What COULD have or SHOULD have happened?": Full-Cost and Full-Benefit accountings and evaluations, along with various other suitable metrics, and extensive incorporation of externalities and also of cross-subsidies or joint interactions in production or consumption, would both challenge and enrich the Path Dependency approach.
Thanks for this insightful reflection and the structuring matrix. I am working a lot on cluster promotion and the promotion of local and regional innovation systems. During the last months I reflected a lot about opportunities to promote radical innovations in rather rural areas in Germany, where a rapid structural change and in many areas also an economic decline is going on. You article gave me great ideas to reflect about the depending eco-system of the rural customers, their behaviour patterns as well as the opportunity of a more differentiated identification of potential businesses or sectors that might be able to innovate their products or their business models in a less costly way and in the beginning in a small market area. Thanks for this thought push!
Good read. I connect it what IBM is trying in the cloud, battling Amzn and in the processor space, battling x86, while at the same time milking the mainframe installed base
9 visitors upvoted this post.