My first quest in this vein will be to re-read Blood Meridian (a book where many events happen and all of them are bad) with an eye toward what protocols are absent or what malicious protocols the Judge installs.
“Protocols can be understood as infrastructural technologies for producing non-events at scale, often without a central coordinating mechanism. For that reason, we like to call protocols engineered arguments. They are not universal agreements, but rather mostly predefined sets of rules that allow individual actors to make trade-offs without becoming embattled.”
Why the emphasis on a lack of central coordinating mechanism? Quite a few of your examples including that of hospitals has a central administrative organ, that creates protocols and tracks them right?
Also could protocols also been seen as the currency of a bureaucratic system, that’s optimised for recurring work at scale?
A protocol issued by an admin center does not necessarily possess a central coordinating mechanism. Once that protocol, like Watch Team Back Up, is out in the wild, it enables individuals to work with each other directly rather than rely on a third party acting as central broker to the whole network. It's an important characteristic because it allows for coordination to scale without being rate limited by, say, an organization-wide approval process.
do you have a definition of what you consider a "protocol"? i ask because the list of examples is so expansive, that i'm not sure putting them all under one umbrella is useful.
“A protocol, in a broad sense, is an agreed-upon way of doing something, shared among people or systems, to coordinate actions toward a desired outcome. We usually hear the term in computing or diplomacy, but protocols are everywhere: computer networking schemes, climate treaties, hand-washing practices, interactions with royals, treatment of football injuries, medical research design, fire-fighting practices, diplomatic etiquette, and blockchains have all been described as “protocols”. The concept spans technology, society, and culture.
Protocol thinking means consciously viewing problems through this lens of shared rules and processes. Instead of only looking for one-time fixes or relying on individual effort, protocol thinking asks: What structured approach or agreed process could address this issue repeatedly and reliably? This mindset treats challenges as coordination problems that can be solved by designing better interactions and agreements among participants. The advantage of protocol thinking is that it offers a way to “level up” problem-solving in almost any context by harnessing collective behavior in a structured way. From a neighborhood organizing a safety plan to a global agreement on climate action, thinking in terms of protocols helps break big challenges into manageable, repeatable steps that people can execute together.”
Hey Timber — the "dynamic non-events" framing is the part of this I'm going to be thinking about all week. The most valuable thing a protocol does is make something NOT happen — the breach that didn't occur, the error that never fired, the patient who didn't wait too long. That's an impossible thing to sell because you're selling absence.
I build custom AI agents for businesses and I run into this exact problem. The best measure of a well-built agent isn't what it did — it's the 2am client inquiry that didn't go unanswered, the booking that didn't fall through, the owner who didn't burn out because the system caught what they would have missed. Trying to show ROI on a non-event is the hardest pitch in any field.
The Bastiat connection is smart — the broken window fallacy is basically "we over-index on visible activity and under-index on invisible prevention." That's the whole AI agent market right now. Everyone wants the flashy autonomous workflow. Nobody wants to talk about the guardrails and protocols that keep it from doing something stupid. Bookmarking the canon list. Following.
What does ‘protocol’ do that ‘infrastructure’ does not?
Great essay.
This is sick - can't wait to see how it develops
A few other books worth considering:
When Formality Works (Stinchcombe)
Engineering Rules (Yates and Murphy)
Underground Empire (Farrell and Newman)
See also, Caroline Levine, Forms: Whole, Rhythm, Hierarchy, Network
This is epic for the landscape of (future) models for continuing education, too. Excellent.
My first quest in this vein will be to re-read Blood Meridian (a book where many events happen and all of them are bad) with an eye toward what protocols are absent or what malicious protocols the Judge installs.
“Protocols can be understood as infrastructural technologies for producing non-events at scale, often without a central coordinating mechanism. For that reason, we like to call protocols engineered arguments. They are not universal agreements, but rather mostly predefined sets of rules that allow individual actors to make trade-offs without becoming embattled.”
Why the emphasis on a lack of central coordinating mechanism? Quite a few of your examples including that of hospitals has a central administrative organ, that creates protocols and tracks them right?
Also could protocols also been seen as the currency of a bureaucratic system, that’s optimised for recurring work at scale?
A protocol issued by an admin center does not necessarily possess a central coordinating mechanism. Once that protocol, like Watch Team Back Up, is out in the wild, it enables individuals to work with each other directly rather than rely on a third party acting as central broker to the whole network. It's an important characteristic because it allows for coordination to scale without being rate limited by, say, an organization-wide approval process.
Got it. Would languages be a good analogy for protocols? Or a very limited version with a defined grammar/syntax?
That enables specific communication/coordination between people, but also objects.
I am working on a coordination protocol. I have published an extensive list of standards for this effort: https://github.com/coordination-structural-integrity-suite/suite and AI assistance to go through the very developed, nearly legal standards: https://coordination-structural-integrity-suite.github.io/ai/
An addition to the canon - Keller Easterling’s Extrastatecraft.
That is a good one!
do you have a definition of what you consider a "protocol"? i ask because the list of examples is so expansive, that i'm not sure putting them all under one umbrella is useful.
Hey Rui: I would recommend looking here: https://docs.google.com/document/d/111EDpTAN-HJCwqMwQlGHojEA5b2A3Awg333HtJybjJg/edit?usp=drivesdk
“A protocol, in a broad sense, is an agreed-upon way of doing something, shared among people or systems, to coordinate actions toward a desired outcome. We usually hear the term in computing or diplomacy, but protocols are everywhere: computer networking schemes, climate treaties, hand-washing practices, interactions with royals, treatment of football injuries, medical research design, fire-fighting practices, diplomatic etiquette, and blockchains have all been described as “protocols”. The concept spans technology, society, and culture.
Protocol thinking means consciously viewing problems through this lens of shared rules and processes. Instead of only looking for one-time fixes or relying on individual effort, protocol thinking asks: What structured approach or agreed process could address this issue repeatedly and reliably? This mindset treats challenges as coordination problems that can be solved by designing better interactions and agreements among participants. The advantage of protocol thinking is that it offers a way to “level up” problem-solving in almost any context by harnessing collective behavior in a structured way. From a neighborhood organizing a safety plan to a global agreement on climate action, thinking in terms of protocols helps break big challenges into manageable, repeatable steps that people can execute together.”
Hey Timber — the "dynamic non-events" framing is the part of this I'm going to be thinking about all week. The most valuable thing a protocol does is make something NOT happen — the breach that didn't occur, the error that never fired, the patient who didn't wait too long. That's an impossible thing to sell because you're selling absence.
I build custom AI agents for businesses and I run into this exact problem. The best measure of a well-built agent isn't what it did — it's the 2am client inquiry that didn't go unanswered, the booking that didn't fall through, the owner who didn't burn out because the system caught what they would have missed. Trying to show ROI on a non-event is the hardest pitch in any field.
The Bastiat connection is smart — the broken window fallacy is basically "we over-index on visible activity and under-index on invisible prevention." That's the whole AI agent market right now. Everyone wants the flashy autonomous workflow. Nobody wants to talk about the guardrails and protocols that keep it from doing something stupid. Bookmarking the canon list. Following.