I. The Remarkable Story of Everyday O2O Commerce in China
Let’s say you were planning to visit China ten years ago. You would have your visa and flights booked, perhaps even a hotel. But the preparations wouldn’t end there. You’d have withdrawn plenty of RMB in cash, which you’d need for just about everything, except maybe your hotel. You’d think about bringing items like toothpaste, clothing, and stationery, as items for the Chinese market were often inferior, and many international brands were difficult to find (and, when you did find them, priced at a premium). You braced yourself for everyday activities, like finding a place to eat and getting around the city, as all of these posed considerable friction, requiring effort and negotiation, often culminating in frustration. In fact, it didn’t even matter if you were foreign or not, as most locals confronted similar challenges, outside the relatively small circle of the merchants and service providers they trusted.
Today, if you’re in China and have a smartphone (and some basic Chinese reading capability, a decent screen-translation app, or a helpful friend), most of these challenges are largely obsolete. Mobile payments are ubiquitous to the point that it’s routine to go weeks or even months without cash — restaurants, hotels, landlords, and even street vendors all readily accept mobile payments. You can now look up restaurants, have food delivered, rent a bicycle, buy a train ticket, or book a taxi within a matter of taps, with a huge and dynamic trust base of user reviews, recommendations, and background checks.
II. It’s Not Just ‘Silicon Valley In Shanghai’
Many of these innovations, of course, are hardly unique to China — the likes of Amazon, PayPal, Uber, Yelp, Expedia, and their competitors have become household names across the United States and much of the developed world. One could readily argue that much of China’s O2O commerce success is derivative (if not an outright copy) of these pioneering innovations.
And yet the scale and pervasiveness of adoption in China gives room for pause. China now leads the world in mobile payment (86 percent vs. 43 percent global average, according to a recent Nielsen report), and online commerce penetration (98 percent). This is not confined to China’s leading cities — food deliveries, taxi booking, and online shopping are widely available, and adopted down to Tier 3 and Tier 4 cities (i.e., cities that many people outside of China have never heard of). By some standards, even a Tier 3 city in China now offers many online mobile-access conveniences surpassing those in leading Asian cities like Singapore, Hong Kong, and Tokyo (never mind other parts of the world).
III. How? Innovation with Chinese Characteristics 
How did this come about? How did China go from being one of the world’s least connected societies to becoming one of its most, deploying and adopting business models and consumer solutions at unparalleled speed and scale?
The story is complex and nuanced, but here are a few key factors underlying this shift:

• Mobile-native population — China’s population largely skipped desktops and laptops – most of the population’s early Internet access was primarily through Internet cafes –  and went straight to personal mobile devices. For most mainland Chinese, their first connected device would have been a smartphone. This provided a large addressable user market for O2O commerce services on a rapidly-iterable development platform.  
• Urban density — Many O2O business models require a certain transaction volume, which typically means they need a minimum population density to be feasible. China’s high and dense urban population provides this.    
• Plentiful organisable labor — China’s large opportunity-hungry population, especially rural migrants, has fueled industrial success and is also a key enabler of the modern service sector. Further, a culture and history of large-scale organization has made it easier to build operational scale: a good example is courier fleets, a vital component of any O2O ecosystem. At the same time, the low cost of labor has allowed widespread consumer adoption without the need to charge a high and potentially prohibitive premium to cover manpower costs.  
• Ubiquitous adoption of mobile payments — In a way, China’s lack of extensive consumer banking infrastructure primed its population for next-generation solutions like mobile payments when they emerged,  accelerated by ingenious cultural marketing (e.g., Wechat’s hongbao or ‘red envelope’ promotion to encourage adoption of its payments feature). This in turn provided a platform for the next layer of services.
• Latent entrepreneurial capacity — Faced with many challenges to so much as survive over their not-so-distant lifetimes, much of China’s population has consequently become both hungry for entrepreneurial opportunity, and ingenious enough to, more often than not, make it work. As many O2O commerce models are two-sided platforms which need to attract not only consumers but also suppliers, an available supply of entrepreneurship is key to viability. For example,as Uber does not itself drive passengers to their destinations, it has to find and attract both drivers and passengers for its business model to work, unlike a traditional ‘one-sided’ taxi company which only has to attract passengers as it hires drivers in-house.
• Starting conditions of friction — The propensity to adopt a new solution depends on how cumbersome existing solutions are. The fact that many everyday acts of commerce in China encountered considerable friction to start with led to a user population open and willing to try out and adopt new solutions, however rudimentary.  

IV. Hypothesis: China as ‘Lead Leapfrogger’ 
As we survey these factors, one pattern begins to emerge — these barriers and enablers are, in essence, not technological but cultural factors, and further, they are all quite specific to China. This does not mean that all Chinese O2O innovation efforts have been successful: China’s O2O landscape, as with all ecosystems of innovation, is littered with failed ventures and broken dreams.  But here is the key: those that succeeded were the ones that managed to leverage these local factors, and effectively combine them with maturing and increasingly standardized global technologies like smartphones and mobile Internet. 
In other words, the story of China’s remarkable progress in O2O commerce adoption is not a story of pure technological innovation. It is a story of adapting the best-available global technologies to the distinct local business, cultural, and sociopolitical landscape, and of doing so better than almost any other economy or region. 
China, in short, is what we might think of as a lead leapfrogger — an ecosystem that manages to deploy and adapt a stack of available technologies to meaningfully bring about new patterns of living (within, for this essay, the domain of commerce). 
What does it mean to be a lead leapfrogger?

• The notion of leapfrogging in the context of technology, innovation, and development refers to the upgrading of an entity’s infrastructure or practices in a manner that ‘leaps’ over the historical or industrial sequence of (typically linear, sequential) progression. For example, going from 2G to 4G telecommunications without bothering with 3G, or, as with some island nations, skipping fixed-line infrastructure and going straight to nationwide wireless Internet.  
• The notion of lead here draws from the concepts of lead users, a term often used by startups to refer to the customers who are the first to experiment with, make use of, and influence the development of new products because they have specific, advanced, or articulated needs which stand to benefit most from the new solutions.
• In most cases of leapfrogging, the solution leapt to is one that is already well-established, often off-the-shelf (such as buying an iPhone as one’s first computing device) — which is to say, that typical leapfroggers tend to be late and not early adopters.
• So to be a lead leapfrogger is something different. It is to combine the non-sequential progress of leapfrogging with the specific, advanced needs of a lead user.

V. Technological Innovation Converges, Cultural Innovation Diverges
What difference does this make? Doesn’t all innovation and progress entail some form of leading and leapfrogging? 
Here’s the key: this expands the conventional narrative of innovation and progress. In this narrative, progress is linear — one moves from yesterday’s to today’s to tomorrow’s technologies. Leading innovators work on discovering and defining the cutting edge of the future, while everyone else catches up while usually settling for a precedent set by the leaders.
Lead leapfrogging matters because it falls outside this narrative: a lead leapfrogger works on a new edge, breaking precedent, and defining a new domain of possibility. It suggests a set of innovation trajectories that are divergent, not convergent. What does that mean?

• Many technological innovations tend to be convergent, in that they usually narrow down to the optimum best practices and the state of the art. For example, figuring out the best way today to maximize the number of transistors per square inch on a chip. This works well with problem spaces like physics or engineering.
• But culture is nonlinear: the problem spaces and resource landscapes in a given society at a given time are highly contextual, resulting in many possible optimum outcomes. This opens up the possibility and opportunity for divergent innovation, centering on contextual and highly-adapted solutions, that is likely to result in varying patterns of solutions and outcomes. 

So consider this implication of China’s O2O story — in domains like commerce which are highly culturally-contextual, successful innovation is likely to be divergent. China isn’t necessarily the alternative model for other ecosystems to follow; rather, its story suggests a broader landscape of many possible successful models, a landscape which subsequent leapfroggers are likely to discover and inhabit. In other words, Indonesia’s O2O innovation trajectory is more likely to look different from both China and the US than it is to resemble either of them (and similarly for Vietnam, India, Kenya, etc.).
In a landscape of convergent innovation, markets tend to be winner-takes-all, favoring near-monopolies like Microsoft on operating systems, Google on search, and Amazon on retail. But in a divergent innovation landscape, there is more room for fragmentation, niches, and long tails – which suggests that even in a world where we all use the same smartphones and cloud computing platforms (convergent innovations), how we actually use and adopt them in our everyday lives is likely to diverge, flowing to address the underlying cultural patterns and needs.
VI. Big Data & AI Aren’t Just Tomorrow’s Technologies — They’re Really Tomorrow’s (Divergent) Culture
Let’s consider a further implication. How might this impact the evolution and development of what is likely to be our most world-shaping invention to date — AI?
Modern AI, as epitomized in models like unsupervised learning (with deep learning approaches notably being the most successful to date), is in essence, a process of algorithmizing data, i.e., having data generate its own algorithms. Instead of prescribing what an AI should do, we give it examples of ‘done’, and (largely) leave the AI to work out its own formula of doing. The resulting algorithms are often inscrutable to human analysis, but have proven to be capable of high levels of effectiveness, given sufficient quantities of sufficiently meaningful data.
What this means, however, is that as effective as these AIs may be, they are epistemologically – and therefore behaviorally – bound by their data. If AI is algorithmized data, then the data defines the algorithm. They are necessarily blind (barring explicit intervention) to everything that is not embodied and modelled in the data. And what they are blind to, they cannot meaningfully decide or act on. In a world run on AI, that which the data does not model, effectively does not exist.
The attributes, connections, and behaviors we capture in our data — and how we model them — shape what our AIs value, and how they behave, i.e., their culture. Our culture, to the extent we model it (or don’t) shapes the cultures of our AIs, which in turn increasingly and pervasively shape the world in which we think, interact, and live. This means that our data and their AIs are fundamentally not so much technological as they are cultural products. They are not technologically neutral, but inherently and ineluctably cultural. Our conversations, our policies, and our designs of our AIs – and the data and models they feed on – needs to recognize this.
Finally, to the extent that our data and AIs are cultural products, they are – as we reasoned earlier – prone more to diverge than converge. If China, as lead leapfrogger, has set a precedent for open-ended, divergent cultural innovation, then we can expect to see a diversity of cultural practices emerge. And in turn, as these practices are respectively modeled and captured, we begin to accumulate diverse and divergent pools of data – that then undergo algorithmization and develop into diverse and divergent AIs. What begins to emerge in this extrapolated imagination is a veritable ethnodiversity, not a mono-deity, of AI values and behavior – a world that is more Babel bazaar than SkyNet singularity.
This was written exclusively for Digital Asia Hub. For permission to republish or for interviews with the author please contact Dev Lewis.