Since taking a break from my daily notes in this space about a year ago, I’ve published a handful of longer pieces on Medium; and for some of you who used to follow me here, I thought you might be interested in these. The most recent one is copied below and the link for all the others is this. Maybe one day I will return to the daily exercise, and until then I hope you will stay well and check in on occasion.
Networks are having a moment, or rather, they’ve been having it for quite some time, long before the big contagion and the flattening of curves, but the moment is expanding. Stephen Wolfram is working on a new physics model that would explain the universe in terms of network evolution and lead to the long-sought Unified Theory in the field. Niall Ferguson published a book that presents modern history as a series of network events, governed by network behavior and attributes. It is generally recognized in financial markets (themselves dense network structures) that the modern economy was borne of the InterNET (emphasis added). Most currently, cryptocurrencies — a revolutionary but natural extension from all that — are inherently network structures.
And yet there has been very little in the areas of finance and economics — in the mainstream at least, really nothing — to seek to bridge the analysis of value and strategy with what would seem an important counterpart in network science. If science is in this case too strong a word, network theory would be a good enough place to start. While the footprint of commercial networks has been growing to the point where regulators are even taking note, the resulting arguments don’t seem rooted in any network concepts or their context, which renders the debate — at a minimum — incomplete.
Perhaps the financial conventions that we still resort to, predicated on old-economy metrics and trends, are adequate enough. It is nevertheless non-trivial and now also opportune to try to understand the network asset on its terms. What follows are high-level notes, almost a preface, in support of an investment thesis that may be backtested with promising results. This is only a summary, purposefully light on supporting detail and examples that could one day turn it into someone’s massive book. But the interested reader should be able to consider cases, circumstances and criteria from one’s own observations that will fit the mold.
1. In an economy dominated by technology advancement and the digitization of virtually all things, there is a diminishing distinction between companies in the so-called “technology” segment and all the others. On the supposition that a time will soon arrive when the distinction will have altogether vanished, the more important difference is between companies that offer product or service solutions with and those without network effects.
2. The latter are in a state of constant need to upgrade, update, reinvent, ride cycles up and down, and reduce pricing on their offering, or else become disrupted or commoditized in a competitive environment that is moving ever faster. The former — businesses with network value — are better able to withstand attack and well positioned to drive down the costs of growth, because the network mesh is both a base and self-perpetuating driver.
3. But not all networks are the same, in fact they’re mostly very different from each other. At the highest level, these include marketplaces, exchanges, communication systems, platforms, artificial intelligence and other connected data tools — to name some of the principal categories — which may also combine several of the listed fields. And by the same token, network effects — defined as the improvement of network experience or value with the addition of new sources of engagement — are also different among the varied types. These can be broadly classified as user or data network effects — the first a matter of popularity and the second a matter of depth — which may also for some networks work in combination.
4. At levels below the very high, there are more nuanced distinctions that define the nature of the asset, often relative along a continuum, rather than binary or absolute. Examples of such qualities include the centralized, decentralized, and distributed topologies; the single-, bi-, and multi-directional data flows; and the single and manyfold layers of the network, which often co-exist and may in fact be linked, internally or externally to the individual business unit. There are also differences of strength between the ties that link the nodes together, and there are differences in size and numbers of the clusters that ensue.
5. These things and others shape the network profile, which in turn shapes its value and potential. Distributed networks, for instance, all other things being equal, may be more valuable than the centralized variety at the opposite extreme, because the single center-point of the former is a vulnerability to its whole. But all other things are not equal, and thus the network whole is best to understand without such oversimplification. Perhaps the one true constant in the general assessment of all sorts is the value of engagement, an attribute that’s always worthy, regardless of the other qualities described.
6. The so-called FAANG contingent — a less than ideal grouping as each of the five constituents is a different type of network from the others — is a highly visible sampling of the digital network asset class. Because they’re big and public and have evolved in more or less transparent ways, they make good subjects for more general analysis that can be carried over to the less developed cases. The fastest growing (if not already dominant) forces in many if not all the major industry segments — transport, finance, commerce, education, health, security, most recently biotech and manufacture — are additional examples.
7. In all the branches and the realms, it is difficult to the extreme to build a network from ground up. Unlike a product or a service that is designed, produced, and introduced into the market — successfully or not — a network is a complicated being, almost biological in nature, and subject to improbable conditions to take root. And when it does — a miracle in ways, which follows an inflection point that’s hard to manufacture — the network must be nurtured like an organism. The influences and the outcomes (which is to say, the behavior of nodes and clusters) aren’t always easy to anticipate, and there further comes a time at which the growth will asymptotically stall. When this occurs, new use cases or network offerings may provide a lift, but it isn’t always known during such times if the desired effect will materialize. Conversely, a robust and growing network can create great optionality, which may among other things enable an expansion into contiguous network areas.
8. Despite the frequent overlaps of categories — for instance, in finance and commerce, in information and entertainment, messaging and transactions — it has been observed that where user network effects are a core network driver, the resultant entities tend to be unique. Think, Instagram, LinkedIn, Twitter, YouTube, in the social realm, despite the basic similarity ingrained in messaging. It’s also been observed that as new categories form, the landscape tends to winner-take-most or even winner-take-all (power law) competitive effects. The same is true of the behavior within the network, as big clusters tend to become bigger and even dominant at times. (The influencer economy is a result of such network patterns.) What all of this describes is a tendency to centralization, even among the decentralized, that the subject networks may seek to mitigate.
9. Financially, the network’s most attractive features are profitability (after the inflection point) and predictability (as network effects take form). In combination with the winner-take-most advantage previously described and the high-gross-margin nature of the software business in the digital economy, the cash flow of the operation can be superior to almost any other business type. In cases where this is not so, it is advisable to wonder why. Perhaps the subject isn’t quite the network that we might assume it is, while, on the other hand, it’s also true that there are companies that unsuspectedly reveal themselves as such.
10. As certain network platforms have grown and reinvested cash over the years, the concern has recently arisen that these have taken on monopolistic forms which have to be controlled. Should there be a breakup of some kind (which wouldn’t be a first, e.g., Ma Bell), it may be worth remembering the nature of the living organism. Plants can continue to grow after a branch or two have been sliced off, and even branches may evolve into a tree or two once they’re replanted. After a while, we may be back to where we started, or someplace very different and strange, because the consequence of change in complex systems is often unintended.
One way perhaps to summarize all the above and draw a financial conclusion in terms that match traditions and accepted standards, is this:
In the digitized economy, networks and network effects are capital, and seeking to develop these is a capital investment; technology advancements and disruptions are important, but on their own are an expense.
In the long term, however, network effects may also turn negative and spiral down, while the technology expense can in the short term create value.
Reinterpreting the networks (2020)
Interpreting the networks (2017)