Exponential curves gradually, inexorably grow until they reach a limit. The function increases over time. That’s why a force like gravity, which grows exponentially as objects with mass get closer to one another, eventually leads to a black hole. And at the middle of this black hole is a point of infinite mass, a singularity, within which the rules no longer apply.
Financiers also like exponents. “Compound interest is the most powerful force in the universe” is a quote often attributed to Einstein; whoever said it was right. If you pump the proceeds of interest back into a bank account, it’ll increase steadily.
Computer scientists like to throw the term “singularity” around, too. To them, it’s the moment when machines become intelligent enough to make a better machine. It’s the Geek Rapture, the capital-S-Singularity. It’s the day when machines don’t need us any more, and to them, we look like little more than ants. Ray Kurzweil thinks it’s right around the corner — circa 2045 — and after that time, to us, these artificial intelligences will be incomprehensible.
Businesses need to understand singularities, because they have one of their own to contend with.
Business has been about scale
For centuries — since at least the start of the industrial era — business has been about scale. As a business student, I was constantly told that bigger companies have the upper hand. Economies of scale are the only long-term sustainable advantage, because with scale you can control markets, set prices, own channels, influence regulators, and so on.
The embodiment of this obsession with scale is the corporation. You may have issues with today’s companies-are-people-too mindset, but remember that they were initially conceived to allow huge projects like trans-continental railroads to happen while shielding the investors from the equally huge risks. Before corporations, it took a monarch to build something truly epic.
The corporation wouldn’t be possible without an organization that could itself scale. Daniel McCallum first realized that organizational charts and spans of control let the railroads scale, and we haven’t looked back. Just as standardization made the mass production of everything from cars to armaments possible, so the organizational chart made global companies possible.
Scale is so entrenched in our society that it’s built into our fundamental economic indicators. Gross Domestic Product (GDP) rewards national productivity rather than, say, individual productivity or citizen happiness. Can’t make your GDP grow by improving things? Grow your population.
Thanks to software and big data, however, scale’s importance is waning.
Why software changes businesses
Marc Andreessen once observed that software is eating the world. Once a process becomes digital at one end, and digital at the other, it quickly turns digital in the middle. As the inputs and outputs of industry become increasingly digital, the middle — the organization — becomes software.
Software has two important attributes that fundamentally change how businesses are run:
First, software can be analyzed. Digital systems leave a digital exhaust, an analytical breadcrumb trail that happens automatically. An employee doesn’t record how long it takes them to do something; software has no choice but to do so.
Second, software can be optimized. Managing humans is messy. It is fraught with emotion and governed by employment law. But nobody cares about pitting two algorithms against one another in a battle to the death. HR is rough and toothless; software optimization is tough and ruthless. Humans retire; code gets a faster processor.
Analysis and optimization lead to a closed loop of continuous improvement. They give us the exponential function.
Heartless? Maybe. If Hollywood has taught us anything, it’s that singularities aren’t good for those left behind, as “The Terminator” and “The Matrix” suggest. Closer to home, one look at the runaway risk of algorithmic trading or the creepy dystopia of wiretapping proves that we haven’t yet figured out how to harness our connected world for the greater good.
But remember that the Terminator was a cyborg — literally, a cybernetic organism. Cybernetics is the study of feedback loops. According to Wikipedia:
Cybernetics is applicable when a system being analyzed is involved in a closed signaling loop; that is, where action by the system generates some change in its environment and that change is reflected in that system in some manner (feedback) that triggers a system change, originally referred to as a “circular causal” relationship.
The business singularity is about creating a business that analyzes changes in its environments and turns them into system updates. The smartest companies know this. They instrument every facet of their business, and figure out how to tweak it. I joked the other day that Google’s business plan is really to get to the singularity first, because after that it won’t matter. Maybe that’s more right than it seems. Maybe organizations that get to the business singularity first won’t care about their competitors.
It’s the cycle, stupid
Companies that learn to harness the power of data iteratively stop worrying about scale, and start worrying about cycle time. To them, everything is an experiment, a chance to optimize. They analyze everything, and feed this back into themselves, continuously engineering their improved successor. Scale might happen — in fact, it probably does, because software is easy to replicate — but it’s a natural consequence of circular, causal loops.
It’s this cycle of learning and optimization, accelerated by software and the data exhaust of a connected society, that pushes businesses toward a limit, a point at which they stop behaving like organizations and start behaving like organisms. Importantly, companies on that side of the business singularity will seem opaque to us, shifting and transient, unthinkably agile. To them, we’ll seem sluggish, predictable, and unwise. Like ants.
This seems a bit fanciful, and smarter folks than I have called it mere hyperbole. So let me offer an example by way of virality.
To an analytics wonk, the number of people who adopt a product because an existing user told them to is measured with the viral coefficient. If every user invites at least one other user, you have a business that grows by itself. Hotmail rode virality to a $300 million exit because every email it sent carried an invitation, a natural vector for infection.
But there’s a second viral metric that’s much less talked about, and sometimes more important: viral cycle time. This is the delay between when you sign up, and when someone else does because you told them to.
Back in the early days of YouTube, there were several video sites competing in the rapidly-growing online video sector. YouTube wasn’t the best. It didn’t even have the best viral coefficient; companies like Tabblo were doing better. But what YouTube did have was really, really good cycle time. People tended to share a video with others more quickly on YouTube than on competing sites. As a result, YouTube quickly left the others in the dust.
Companies like Google and Amazon care as much about the cycle time at which they learn as they do about their ability to generate products and services. Scale is a consequence of iteration or a side-effect of replacing things with software. Everything these companies do is an experiment. Scale is OK because it gives them more test subjects and increases the confidence level of their results. But scale isn’t the point: quick learning is. As my Lean Analytics co-author Ben Yoskovitz says of startups, the goal is to learn. They get better, more efficient, and the next cycle is infinitesimally tighter. The curve bends, inexorably and imperceptibly; they approach the limit.
There’s another definition of singularity: a peculiarity or odd trait. Today, companies that are passing through to the other side of the business singularity look weird to us. They invest in solar cells and goggles. They offer their own infrastructure to competitors, or open source it. They behave strangely (to us, anyway), trading profit for iteration. They get uncomfortably close to customers and critics.
I don’t think accountants have a metric for “how fast the organism learns,” but they’d better get one soon. For modern businesses — built with little capex thanks to clouds, marketed with little investment thanks to social media — learning is a company’s greatest asset. Learning faster is enough to unseat titans of industry. Those on the other side of the business singularity live by cycle time; those on this side seldom think about it.
I’ve definitely abused the notion of a singularity here. Maybe this isn’t as tectonic a shift as the rise of sentient machines or the middle of a black hole. But it’s more than just the evolution of businesses, because it’s the migration from a physical world to a digital one. We’re moving from a business ecosystem where those who have scale win, to one where those who have better cycles of adaptation and learning win.
The cycles themselves are driven by data and software. It’s something I’m hoping to explore in detail during the Data Driven Business Day at Strata Santa Clara in late February.