The Meaning of Open

Alex Komoroske
Jan 17, 2018 · 7 min read

There are a lot of misconceptions about what open means, when it is the right strategy to apply, and the fundamental tradeoffs that go along with it. It’s very easy to cargo-cult the notion of open — using it in an imprecise or half-baked way that can obscure the real dynamics of an ecosystem, or even lead you in the wrong strategic direction. Here are a few important things to know about openness in the context of ecosystems.

Openness is a spectrum. A thing is not “open” or “closed” — it instead exists upon a spectrum of how open the system is.

Openness is about power differentials. Fundamentally openness is about the existence of power differentials between entities: the ability for any one entity to exercise capricious power. This is an abstract notion that can be hard to reason about or apply in practice. The simplest real-world test is this: what is the worst case scenario if any specific entity in the ecosystem went greedy, incompetent, evil, or corrupt? How long would it take for the ecosystem to recover? In some cases, the answer is that the ecosystem couldn’t recover, which is clear evidence that the ecosystem has a powerful closed portion. In other cases, the ecosystem keeps right on trucking with nary a blip — a sure sign of an open ecosystem. This practical test helps identify the points of power, and also where along the spectrum of openness this ecosystem falls as a whole. The difficulty to recover should account for all of the various real-life forces (some of which are real but hard to discern) like network effects, the power of defaults, the cost of distribution, the friction of app installation, the power of established brands, ingrained user habits, etc.

You are biased about your own position in an ecosystem. We are biased to believe the best about ourselves and the goodness of our motives. But ecosystems are the sum of inter-dependent incentives, which means a proper analysis requires understanding everyone’s true incentives — including your own. To do this properly, you must see yourself as others do. An easy trick to mitigate this bias is, every place that your own entity’s name appears, replace it with the name of your fiercest competitor and see how you feel. That’s likely similar to how others feel about you (or would if you ever became as powerful as your competitor). If you feel bad about it, that’s a sure sign your position in the ecosystem has arbitrary power.

A single closed portion can have severe impacts on the ecosystem. In some cases, an ecosystem can have a closed portion, but due to the way the ecosystem is set up that one entity does not have undue power and the ecosystem is still mostly open. This is the case where there’s a powerful, well-behaved entity, but that entity is fundamentally easy to replace if they were to become badly behaved — for example, because the switch cost for users is very small. In other cases, an ecosystem is made up of 90% “open” parts, but a single, well situated, powerful entity has a vise-like grip over the ecosystem. Ultimately a careful analysis using the test outlined above is the best way to uncover which situation it is. If you’re trying to design a system that is open, but have missed the extreme power a specific piece of it wields, you may have designed a system that might as well be closed.

Open systems create gravity wells. Systems that are truly open tend to attract others to join them at an ever-accelerating pace. In ecosystems that are ruled by a despot no matter how successful other participants in the ecosystem are, they fundamentally just empower the despot to have more leverage over them, because they have more to lose and their success feeds the despot’s success. In open systems, on the contrary, participants see that they don’t have to fear their own success fueling their own increasing subservience to a despot. Each individual entity who can’t plausibly build their own similarly-sized proprietary ecosystem to compete — the overwhelming majority of entities — is incentivized to pitch in on the open ecosystem. Investment in an open ecosystem by any one entity helps the entire ecosystem as a whole. This fact, combined with the fact that ecosystems generally get exponentially more valuable the more participants there are, means that in many cases over sufficient time scales truly open ecosystems create gravity wells, sucking more and more into them until they are nearly universal.

Monetization can break what makes the ecosystem open. It’s easy to design an ecosystem where you can monetize due to the power you have. It’s much harder to design an ecosystem where you can monetize because you provide a differentiated, valuable service that people would want to pay you for even if you weren’t in a powerful position. One common mistake is to assume that you will be successful in your strategy and fast forward many years until you have power in a part of the ecosystem, which can then be monetized. But that power differential is — in many cases — precisely the same thing that makes the overall system not be open, which undermines the analysis in the first place. The gravity-well effect only works if the system is truly open, and that effect only kicks in when no entity has undue power. Often we don’t see this because of our bias about our own motives and how others perceive us. This is not to say that it’s impossible to monetize an open ecosystem — quite the contrary —but rather that you must have a good understanding of the value you bring and why that it is a sustainable complement to the ecosystem. Trying to gain as much power as possible to maximize future potential monetization is a surefire way to create an ecosystem that can’t activate the self-sustaining flame and thus never amounts to much.

Minimizing your own power can help the ecosystem be more open overall. In some cases you discover that you are in a more powerful position in the ecosystem than you desire, which threatens the openness of the entire ecosystem. Others will assume that you will exercise all of the power you have available to you, and telling others just to trust that you won’t isn’t sufficient (how would you feel if your fiercest competitor told you to just trust them?). In these situations it is possible to cede power that you have that you don’t want, by making it hard or impossible for yourself to exercise that power in a capricious way. You can do this by committing publicly to principles and sticking to them even when it is costly, actively empowering other members of the ecosystem, open sourcing key components that you control, standardizing interfaces that you implement so that others may implement them too, enacting governance structures that invite others to the table with real power, and many other ways.

Openness is the granddaddy of “commoditize your complement”. A core business strategy is to commoditize your complement. When a particular piece of an ecosystem is commoditized, the switch cost between providers is effectively zero, allowing full-on competition to take place, driving prices as low as they can go. This is great for the consumers of those products, but it’s a terrible place to be in if you’re one of the producers of the commodities — one of the reasons businesses historically resist commodification with gusto. An open system is one that has no point of undue power, meaning that it’s effectively fully commoditized. If you provide a service that is complementary to the rest of the open system, it is in your long-term incentive to make sure the ecosystem is as open as possible — including by limiting your own power in some cases. Some of the most successful business models in the world have been providing useful services that are complements to open ecosystems.

Openness has enormous benefits — but also large downsides. Open systems tend to be extremely resilient because they can recover from any one entity going bad. They also tend to grow ever larger in a self-sustaining feedback loop and become nearly universal — many orders of magnitude larger than similar closed systems. But they are not without downsides. Open systems will always be messy, chaotic, and hard to control and will never be clean, polished, or simple. In fact, the relationship between chaos and openness is fundamental: an ecosystem that gets incrementally more clean is incrementally less open. This is because an ecosystem naturally tends toward entropy over time as entities’ individual incentives cause every niche of the ecosystem to be explored. An ecosystem that stays “clean” likely has an entity in a powerful position within. If you expect any one ecosystem to be simultaneously resilient, universal, polished, and easy to control, you’re going to get extremely frustrated. That’s not to say it’s impossible to influence an ecosystem — only that you should know what you’re signing yourself up for. Successfully influencing an open ecosystem will be like playing a four-dimensional game of chess and even the best strategies will take years to bear fruit. If those downsides aren’t palatable to you, you should should consider alternate strategies.

Open is not intrinsically a moral good. This is something I believed with every fiber of my being for many years. Openness does tend to correlate with morally good outcomes. But openness is just a means to a particular end, and like any other tool, it should be judged primarily on whether it achieves good ends. Openness tends to erode the power of arbitrary power points in a system (i.e. gatekeepers), and in many cases that leads to a better outcome for society writ large. But too much openness can erode all points of power, meaning that the entire system can become a runaway engine seeking to satisfy our basic urges with ever-increasing efficiency. Gatekeepers have historically used their power capriciously of course — but they’ve also used it to help align incentives to give users the things they want to want, effectively aligning users’ short and long-term incentives. As with all things, the key is finding the right balance to make sure the ecosystem is in healthy equilibrium and achieving good ends.