Aggregators aren't open-ended

The landscape of tech is currently dominated by ad-supported consumer products and B2B SaaS companies. My sense is that this may be fundamentally thwarting the evolution of technology. Allow me to string together some theory into maybe 3/4 of a thesis.

Aggregators are network effect monopolies

An aggregator is special kind of internet business with a particular shape. Analyst Ben Thompson developed the concept of aggregators and aggregation theory to explain differences between industrial monopolies and contemporary software monopolies.

  • An industrial monopoly leverages a monopoly on supply to extract $ from demand.

  • An aggregator leverages a monopoly on demand to commodify supply.

A few examples:

  • Facebook leverages demand (users) to commodify supply (content creators).

  • Amazon leverages demand (customers) to commodify supply (3p merchants)

  • Uber leverages demand (riders) to commodify supply (drivers)

  • AirBnB leverages demand (travelers) to commodify supply (apartments)

Aggregators are able to do this by literally aggregating all of the demand in one place. They are hub nodes in the network, through which everything ends up traveling. How do they manage to become a hub? Through two mechanisms:

Together, these create a snowball of preferential attachment. Pretty soon you end up with a defensible position in the network that you can leverage.

Aggregators have a disturbing kind of evolutionary inevitability. Sick of ad-based social networks? Want to charge an honest buck instead? An aggregator will outcompete you on the basis of network effect growth. Aggregators are the sharks of networked software, apex predators perfectly streamlined by evolution to the contours of the internet.

Aggregators broke the evolutionary loop of technology

So, aggregators commoditize supply. Being commoditized is not something you choose for fun. You are commoditized when someone else leverages an advantage to commoditize you. That means there are a lot of motivated suppliers who want to route around an aggregator’s advantage.

We would expect suppliers to find ways to build a direct relationship with customers, and eventually accumulate their own network effects. This is exactly what used to happen on software platforms. Software found new ways to one-up platforms below in an effort to transcend commoditization. GUIs were built on top of command lines, browsers on top of GUIs, aggregators on top of browsers.

Each layer commoditized and built upon underlying platforms, generating an upward spiral of evolutionary complexity.

Layering is how technology evolves. Yesterday’s product becomes tomorrow’s component. That’s a commoditization process that produces the equivalent of heritable genes, which accelerate evolution via sexual reproduction, combining components to create new products.

But with aggregators, this spiral broke. Aggregators have managed to leverage network effect to make it impossible to build new platforms on top of themselves, refusing to reproduce in an attempt to become immortal. How?

The infinite variety of universal machines

Ashby’s law of requisite variety is a maxim from cybernetic control theory. It tells us who, in a system, has their hand on the wheel.

If a system is to be stable, the number of states of its control mechanism must be greater than or equal to the number of states in the system being controlled.

In other words, who has the most expressive variety drives the system. If you have more variety in a system than I do, then you determine where and how the system evolves.

Software is, by nature, a universal machine. It is infinitely expressive. When a system can reprogram itself, the variety of the system becomes effectively unlimited.

The Personal Computer allowed anyone to develop software for any reason. Software developers leveraged this expressive variety to invent never-before-imagined categories of things. Things like spreadsheets, browsers, the web, payment platforms, VoIP, wikis, streaming video, social networks. New euphemisms were coined, such as “over the top”, which is the kind of moral language you invent if you are a network hub, and people find a way to route around you. Infinite expressive variety and permissionless innovation produced open-ended evolution.

How about aggregators?

Aggregators break evolution by limiting variety

Ashby's Law of Requisite Variety means that aggregators must limit the creative potential of their product to retain control of the ecosystem. You can build a Facebook on top of the browser, but you can’t build a browser on top of Facebook.

This is why aggregators don't allow Turing-complete scripting. Turing's black box contains infinites states. Guaranteed loss of control. Instead, most aggregators are built around relatively expressive formats (text, images, video) that cannot be used to build a metaplatform. It’s an ecosystem in a petri dish. It cannot escape the lab.

Even app stores, which seem to want to be aggregators, restrict apps through baroque app store categories to define what is allowed to exist. They also ban end-user programming, as well as over-the-wire sandboxed code. These are all strategies to limit the variety of the ecosystem.

To be an Aggregator is to limit the open-endedness of your ecosystem for the purpose of maintaining requisite variety.

Containment can lead to contagion

For all its evolutionary inevitability, I wonder if the aggregation stalemate is stable? Innovation comes from open-ended ecosystems. Open-ended ecosystems are more expressive than aggregators, and they are fundamentally out-of-control. They thwart control.

One of the lessons from the theory of natural selection is that a degree of isolation can be a great help to a new strategy in gaining a foothold.
Cosma Shalizi, "What can emergence tell us about today's Eastern Europe?" SFI Bulletin, Winter 1999

Containment can lead to contagion.
David Stark

As long as some open-ended ecosystems exist, they will incubate disruptive new evolutionary strategies. We’ve seen two examples of this recently, as app stores tussle over open-ended technologies like WebAssembly, and cloud streaming gaming.

Separately, I wonder if the tools-for-thought space might not produce something beyond aggregators and SaaS services?

My sense is that, more than other networked software, tools for thought want to be owned. I don't want to store my second brain on someone else's computer. What if we instead had a small tool that was personal, multiplayer, distributed, evolvable? Maybe this is just a niche category, or maybe it could be the basis for a new open-ended ecosystem?