The Disruptability Index

8 Risk Factors Determining Internet Giants’ Vulnerability to Decentralization

Crypto is caught today between two opposite — and, in my view, equally wrong — narratives about the prospects of decentralization to transform the business landscape. On the one hand, the dreamers believe in the inevitable decentralization of everything. On the other, the cynics argue that consumers don’t care about decentralization and will stick with the services they already use. As so often is the case, the truth lies somewhere in the middle.

In my last piece, “The Future Of Network Effects: Tokenization and the End of Extraction,” I looked at 1) the outsized role network effects businesses play in our lives; 2) the “Extraction Imperative” which drives those businesses to ultimately exert monopoly power to increase the dollars or data they pull from the network; and 3) how decentralized, tokenized networks offer a compelling alternative.

In this piece, I identify and examine 8 risk factors that make a centralized network effect business susceptible to decentralized disruption. These risk factors include:

1. Economic factors (who pays the economic rents and how much they pay),
2. Competitive moat factors (how technological complex the service is, how big the switching costs are, how winner-take-all the service is, whether a nascent network can still add value), and
3. Third party factors (whether there are third parties who are incentivized to switch and can sway others, users’ expected loss from centralized censorship).

I then judge four of today’s top network businesses — Facebook, Twitter, Apple’s application ecosystem, and Amazon’s Marketplace business — against these risk factors, resulting in a determination of which are most vulnerable to the emergence of one of these tokenized alternatives.

Risk Factor #1: Business Model or “Who Pays?”

TL;DR: Network effects businesses in which the user pays for the service directly through their dollars are more disruptable than advertising-based businesses in which the users pays for the service indirectly through their data.

Broadly speaking, network effects businesses are organized in one of two ways.

In one scenario, end users of the network pay directly through commissions and fees on purchases. This is the default business model of platform networks and two-sided marketplace networks like Amazon’s Marketplace or the Apple app ecosystem. The platform’s profits come directly from the pockets of the users themselves.

As these networks become indispensable channels for sellers to access buyers, the network owners are able to exert monopoly power by increasing their commissions on sales. While at first these commissions squeeze the seller, inevitably they are passed on to the end consumer in the form of higher prices.

In the second scenario, end users of the network pay indirectly through exposure to advertising. In these networks, use of the product is often free to the user because the advertisers foot the bill. In return for this free access, users must give up their attention (by viewing ads) and their data (to be better targeted for ads).

Since advertising is still a somewhat competitive business, monopoly power tends to be exerted less on the advertisers and more on the users themselves. The more users are locked in to the network, the more license the service provider has to push the envelope on their collecting data practices.

In general, business models in which users pay directly (commissions and fees) are much more painful to network participants than business models in which they pay indirectly (attention and data), which are often largely invisible to the end user. Consequently, the potential for a decentralized business to disrupt a commissions-based centralized business is much higher, because lower prices can create high motivation for consumers to switch to an alternative.

Our rankings, then, explained:

• Facebook faces a low risk due to their advertising based business model.
• Twitter likewise faces a low risk for their advertising based business model.
• Apple faces more risk, as their model involves direct consumer payments for apps and in-app purchases that involve high commission rates.
• Amazon also faces more risk as the entire point of the platform is for consumers to make purchases that involve with high commission rates.

Risk Factor #2: Value Per Network User or “How Much Do They Pay?”

TL;DR: For businesses that charge users directly, the larger the economic rents per user per year, the more disruptable the platform becomes.

We saw in Risk #1 that the question of “Who Pays?” is critical to assessing disruptability because it is a strong determinant of how likely a network user is to look for alternatives. Risk #2 is an important extension of #1: how much do they pay? In short, the more that a consumer spends within the monopolistic network, the more economic rent is extracted from them by the network owner and the more incentive they have to leave and seek lower-cost alternatives.

Take, for example, the average amount a consumer spends annually through the app store versus the amount a Prime customer spends on Amazon. According to a SensorTower study, the average iOS user spent ~$66 on apps and in-app purchases in 2017. That is only 2/3rds of the price of Amazon Prime membership, about 20x less than the estimated $1300 Prime members spend annually, and about 10x less than the estimated $700 non-Prime members spend.

If both of these networks were to exert market power equally in a way that raised commissions and forced third parties to pass those margin cuts on to consumers in the form of higher prices, consumers on Amazon would obviously end up paying significantly more rent than app store users, and thus be much more motivated to switch to a lower cost alternative.

Explaining our rankings:

• Facebook is low risk because risk factor #2 is restricted to business models in which the end user pays commissions and fees.
• Twitter is low risk for the same reason as Facebook.
• Apple has medium risk because although it’s business model is based on commissions from user transactions, users spent relatively little in the ecosystem.
• Amazon has more risk because users spent a significant amount of money with the platform, and therefore, have a high incentive to look for lower priced alternatives.

Risk Factor #3: Early Network Value

TL;DR: The more that an alternative network can provide value before it reaches full scale, the more disruptable the incumbent is.

When a user switches to a new, less mature network, there are inevitably sacrifices in terms of the value that network can deliver. The Bootstrap Problem refers to the idea that networks, in general, aren’t valuable until they reach a critical threshold of participants.

In some cases, however, there are also unique, temporary properties of the relatively smaller scale of networks that can provide incentives for switchers. This value tends to be in the form of less competition for attention.

Social networks are largely built around influencers and content creators who use those channels to build and grow an audience for distribution. Larger networks are more valuable in the sense that they have larger total addressable audiences. However, those mature networks are also more crowded with people competing for that attention. Nascent social networks are good places for influencers to build a following that they can scale as the platform grows.

Likewise, there is an argument that in certain marketplace contexts, third party sellers can similarly benefit from a reduction in competition. Much has been written about how Amazon’s algorithm has almost entirely replaced the brand, with algorithm hacking consequently replacing brand building. Third party sellers working through a different marketplace might face less competition for attention and more ability to stand out and differentiate relative to their peers.

In the chart below, the higher the potential for temporary early network value there is, the higher the risk factor.

• Facebook is low risk because there is very little value to a network based on the offline social graph when a very limited percentage of the total social graph is using the network.
• Apple is low risk because a mass reduction in available buyers would render most app development economically illogical, making early network value non-existent.
• Amazon Marketplace is medium risk because while an insufficient number of either buyers or sellers would render a competitor irrelevant, it doesn’t take many sellers for a consumer to find some utility, and the lack of competition for the attention of the consumers who are there is great benefit to those small handful of sellers.
• Twitter has more risk because numerous times, new social networks (see: Vine) have popped up and demonstrated their value in helping a new class of influencers emerge who have, in those early days, less competition for attention than they face on Twitter.

Risk Factor #4: Simplicity To Copy

TL:DR: The less technologically complex a network platform is and the cheaper it is to rebuild its core functionality, the more at risk of disruption the platform will be.

Technological complexity creates defensibility. Companies whose technology is harder or more expensive to recreate make it harder for competitors to reach the point where they can offer a comparable user experience. Even when early adopters are highly incentivized, most decentralized alternatives will need to reach a minimum threshold of UX parity to engender consumer switching.

Let’s take the four network businesses we’re analyzing as examples:

• Facebook is low risk because the software is complex and expensive to recreate. The company has spent billions of dollars over the course of more than a decade building out hundreds of valuable features (events, groups, photos, video, timeline, etc) along with an advanced and highly-trained newsfeed algorithm that largely determines a user’s core experience.
• Twitter has more risk due to a base product that is comparatively simple. Of course, they’ve also spent years and millions of dollars to build on the core, such as providing smarter recommendations about who to follow and which content a user is going to be most interested in. Still, spinning up a clone that replicates the core micro-blogging feature is not technologically difficult. This helps explain why it’s already happening with decentralized companies like Mastodon and Peepeth.
• Apple’s application ecosystem is low risk, as it is one part of a massively complex mobile operating system. Trying to re-architect this sort of system would be an extremely demanding task.
• Amazon is medium risk. Attempting to re-architect the marketplace business to a high standard of user experience in the core functions of discovering, purchasing, returning, and reviewing items is a low-to-mid complexity technological task. It would take longer to recreate and train the complexity of Amazon’s recommendation algorithms. However, given that the majority of shopping on Amazon is intent-based, that feature is somewhat more ancillary than it is core.

Risk Factor #5: Possibility of Multiple Winners

TL;DR: Networks where users already participate in multiple versions of that network (i.e. where multiple winners are possible) are more disruptable than networks where, on average, people only participate in one (i.e. winner-take-all).

The more “Winner Take All” a network is, the fewer “versions” of the network the average person participates in.

Take social networks. Social networks are relatively winner take all. People don’t participate, in general, in two versions of the same network. There is no “other” Facebook; there are only differently architected, differently-focused versions of a social network, such as LinkedIn’s professional version or Instagram’s mobile-first, image-centric version. People may be willing to maintain presences in these similar, but distinct, networks, but far fewer will be willing to participate in two with the same community and feature set. This is expressed, for example, in Instagram slowly burying Snapchat into irrelevance.

Two-sided marketplace networks tend to be less winner-take-all. People are used to shopping in many different online stores that suit different purposes and needs. Even Amazon’s greatest power users of Amazon are likely to still buy things from other eCommerce outlets

Winner-take-most networks are easier to disrupt than winner-take-all networks. When consumers are used to engaging with multiple similar-functioning networks simultaneously, it lowers the perceived barriers to trying a decentralized alternative. In other words, trying something new doesn’t signify giving up the old. So the easier it is to participate in multiple similar networks at once, the greater the risk factor.

A few more notes on our rankings:

• Facebook is low risk because, after nearly 15 years, it’s clear that consumers want a single digital representation of their social graph.
• Twitter is medium risk because people have shown themselves willing to try similar types of experiences. It is unsurprising that, in Peepeth and Mastodon, Twitter already has two decentralized competitors.
• Apple is low risk because in almost every case, consumers chose a single mobile platform and stick with it. In the context of any individual, mobile operating system is a singular decision.
• Amazon is more risk because even with its dominance of eCommerce, Amazon customers still shop from other online marketplaces and stores, and Amazon sellers still sell in other ecosystems. eCommerce is by far the least winner-take-all of the network effects businesses.

Risk Factor #6: Sunk Costs & Switching Costs

TL;DR The lower the sunk and switching costs to the user, the more disruptable the network is.

The potential for a network to be disrupted doesn’t just lie in how strong the motivating factors are, or even the emergence of a plausible alternative. Another essential factor is how much investment has gone into the network (i.e. sunk costs) and how much it costs to switch to the new network.

In many network effects businesses, these are “soft costs,” such as the time it takes to re-upload data about oneself on a new social network, or the social cost of missing out on activity happening on the main network. In other categories, however, these costs have a real dollar price tag.

Perhaps the best example is operating system platform networks. “Buying in” to the Apple or Google mobile ecosystem means investing hundreds of dollars on buying a new mobile device. Switching to a different ecosystem forfeits that expenditure.

The fewer the sunk costs and the softer/lower the switching costs, the higher the risk factor. A few more notes on our rankings:

• Facebook is medium risk. Users on Facebook haven’t spent money to be there, but do have some sunk costs in the form of the digital record of photos and events that they’ve invested in building.
• Twitter is medium risk because although setting up a profile takes little time, it has significant sunk costs in the time it takes to cultivate followers. Huge amounts of effort go in to building a following.
• Apple faces low risk of disruption because the sunk costs and the switching costs are high. It costs a significant amount of money to buy into a phone ecosystem, a sunk cost which must be repeated again to switch.
• Amazon is medium risk because while switching does not represent a significant cost outside of the inconvenience of re-entering billing and shipping info, Amazon has a form consumer lock-in through the sunk cost of Prime Membership and the buying history from which to make better buying recommendations.

Risk Factor #7: Third Party Rebels With Incentives To Encourage Consumer Shift

TL;DR: The more a network relies on a category of third party businesses who have an incentive to switch to a different platform, the more disruptable the network.

Many networks don’t just have the owners and the end users, but a third category of participants that interact with the end users in some way. The way these third parties respond to the Extraction Imperative can have a big impact on whether they have an incentive to encourage consumers to switch to alternative networks.

In the context of networks with an advertising business model, the third parties are the advertisers and their incentives are broadly alignedwith the network owners’ imperative to command attention and extract data from their users. For advertisers, more attention and better data means cheaper and more effective ads. Because of that, they want the user platforms they work with to capture as much data and attention as possible. Advertisers have no incentive to encourage existing network participants to leave because then they lose access to the advertising channel..

Marketplaces, on the other hand, are often built on the backs of third party sellers who use the network platform to reach customers. Sellers have strong incentives to minimize the commissions and fees they pay and the restrictiveness of the policies to which they are subject. Thus, as reasonable decentralized alternatives arise that are better for their bottom line, they are likely to use their resources (email lists, advertising budgets, word of mouth) to help convince consumers to switch.

These rankings are relatively self explanatory:

• Facebook is low risk because it doesn’t have a large cadre of third parties who shape the consumer experience and want an alternative. As explained above, advertisers actually have an incentive to want more, not less information.
• Twitter is low risk for the same reason as Facebook.
• Apple is medium risk because although there are many third party app developers who wish they didn’t have to pay 30% of their sales to Apple, as consumers only have one phone there aren’t any realistic alternatives.
• Amazon has more risk because not only are sellers frustrated and fed up, many are actively engaged in other ecosystems and looking for alternatives.

Risk Factor #8: Expected Loss From Censorship

TL;DR: The more likely a network is to censor its users, and the more damaging and costly that censorship is to those users, the more disruptable the centralized platform becomes.

One of the tools available to centralized network owners is to censor certain types of activity, including banning users who have violated their terms of service, or who they simply don’t agree with on an ideological level. The conversations swirling around the banning of shock political entertainer Alex Jones from major social media networks have dramatized this reality recently.

Censorship can be extremely costly, particularly in the cases where the users being restricted or banned have invested a lot into building their presence on the platform, whether that be followers on a social network or buyers on a marketplace. The larger the expected loss from censorship, the greater the risk to the centralized platform that users will seek decentralized alternatives.

“Expected Loss From Censorship” can be summed up by the equation below:

Expected Loss From Censorship = Damages from Censorship x Risk of Censorship

The “Damages from Censorship” is a measure of how disruptive censorship is to a censored party’s business or livelihood. This cost is related to 1) how integral an audience channel the network is for the particular business in question; 2) whether the business conducts a sale through the platform doing the censoring or whether it is a lead channel to a different platform.

The “Risk of Censorship” is a measure of how likely to ban or restrict participant activity a network is, based on previous actions.

Both advertising-based and consumer transaction-based network businesses carry risk of censorship, and in many cases, the damages can be large. While social media bannings tend to capture more news attention, bannings that happen on marketplace platforms like Amazon can directly cut off people’s livelihoods, creating higher expected loss.

Here’s how I’ve ranked the four platforms: .

• Facebook has medium risk as it represents a key audience channels for many types of business. Facebook has a higher relative risk of censorship than Twitter in the post-Cambridge Analytica context, as it works hard to prove that it will take appropriate action against suspect users.
• Twitter has medium risk because while the company has been much more resistant to censor members than Facebook, and the cost when it does censor users can be quite high for people that use Twitter as a primary audience channel.
• Apple is low risk for the simple reason that most apps that violate Apple’s terms of service never get approved in the first place. This mitigates the risk that an app developer will put in time and resources to develop an experience only to have it shut down later due to censorship.
• Amazon is high risk because both the risk and cost of censorship are comparatively high. In the case of Amazon marketplace sellers, suspension can come at any time, for reasons that aren’t clear, and last indefinitely. This is only getting worse as the company makes a push to curb fraud that seemingly has many innocent parties caught up in the sweep.

SUMMARY

Now that we’ve covered all 8 factors, let’s compare them all in a summary table:

When taken together, the 8 risk factors I’ve identified suggest that Amazon’s Marketplace business is more at risk of disruption than the other major network effects businesses examined. In no way does this suggest that Amazon isn’t a strong business, or that it doesn’t have extremely powerful competitive moats. It does mean that there are more vulnerabilities that a well-designed tokenized network could potentially exploit.

This Disruptability Index is designed to start a conversation and inspire thought and debate. It is not a comprehensive review of every network effect business or even every risk factor. Like all such things, the ratings I give each company are subjective.

I hope that by identifying and exploring these risk factors in depth, we might as a community uplevel the decentralization conversation beyond simplistic arguments for binary futures (“Everything will be decentralized!” “Nothing will!”) and into the nuanced gray areas of probabilistic outcomes. I believe that decentralized projects, organized around blockchains and incentivized by tokens, will offer alternatives to today’s internet network monopolies with varying degrees of speed and likelihoods of success. By exploring what specific factors make a centralized network vulnerable, we can improve the threat posed by blockchain-based alternatives.

Please consider this the beginning of the conversation. I’d love to hear from you — what factors do you believe matter most? How do you see these criteria applying to other network effects businesses? What other risk factors do you think we should be considering? Feel free to share below, or find me on Twitter @kjer.

And if you liked this content, follow me on Medium for more!