Preface

A Problem | Envisioned Solution | A Eureka


A Problem

Last days of Spring 2018.
Blockchains look both very promising and arbitrary.
I am a complete outsider in Toledo, south of Madrid.
Pretty place.
Scientists have taste.
No one is talking about blockchains.
They are busy upvoting each other's papers with like-minded scientists.
Special speakers didn't play this game.
And I didn't have the cards to take part.
I could only hope to connect with the best.

One of the invited guests was professor Petr Skobelev.
He is a pioneer in a type of applied Artificial Intelligence:
adaptive resource management.
He's coming from an engineering background, that is, he makes things work.
He had just developed a virtual market in space.
It solved conflicts through consensus in a fleet of autonomous satellites.
They were running his peer-to-peer algorithm to coordinate independently.
Impressive that the protocol operated under stringent cosmic conditions.
Cooperation on earth was not diplomatic enough.
Petr studied at Samara State Aerospace University during the Cold War.
Earlier, but after the Cold War, at the lunch buffet, he had been kind.
He had given me the reference of an important book for him.
It contained a mathematical proof for markets driven by protocols.
When I passed by him again, outside, by chance, I could not resist.
I stopped him, even though he was on his way to a conference room.

I quickly inquired in a weird way to say hello:

How could I build an adaptable protocol?

Skobelev first reaction was:

You'd need a system to manage the protocol itself.

I wasn't very happy with what the answer was giving me:
the idea of a transcendental system next to the unaware protocol.
When Petr and his followers promptly left, I noticed a side patio.
I went there to think under the sun.
I was hopeful for the intricacies of vegetation around the deadpan stone.
Leafy benches and columns reminded me gardens and statues of Sanssouci:
the down-to-earth stronghold of the Enlightenment in Europe.
But I got stuck.

My conclusion was that adaptability wouldn't be an inherent property.
You'd need a system to manage the manager and the manager of the manager…
getting further away from change signals.
Miners are validating Bitcoin transactions.
There was then a need to manage miners and mining pools appeared.
There was then a need to manage mining pools and control appeared.
The protocol lost some intrinsic value for the sake of management.
The outside cause of an action for change felt wrong.
Externalities are not understood from the inside.
Organisms autoregulate to get rid of managerial systems sucking resources.

The regulation mechanism must be internal.
Be it in minds or protocols, a supernatural motive creates a disconnect:
a greater friction than that of a need for change.
Ethereal injunctions stiffen systems to save the beauty of idealized laws.
Bounded rationality assures a greater complexity than what we can think.
As a result, we endure great pains misaligned with a simplistic Almighty.
Yet, it makes us rigid to resist mostly small temptations.
They only break rules when growing in times of crisis or revolution.
I wanted to align frictions with a simple Almighty.
They could shape adaptability within a protocol's world order.

Envisioned Solution

Organic shift alleviates pain caused by immobile institutions.
Microscopic discord could feed a cost function to adapt protocols.
We want the decisive agility of a supervisor in a single chain of command;
but also the regulation of complementary functions in an organism.
Let the supervisor be a function rather than a separate system.

In traditional software development:

Input + Rules = Output

But in supervised machine learning:

Input + Output = Rules

Likewise, a DLT is learning consensus on the fly from past transactions:

Transaction + Validity = Consensus

Consensus is then applied to new transactions to determine validity:

Transaction + Consensus = Validity

Likewise, in the judiciary:

Case + Laws = Judgment

The question is, how do we learn:

Case + Judgment = Laws ?

Newly formed laws would integrate back the protocol like the consensus.
An input and a protocol for a desired outcome compose usage functions.
The system does not learn in this setting.
The agent triggering a process learns but has no control over the system.
Agents will look to gain control over the system, or switch to another.
They could become supervisors, who need controllers in a vicious circle.
This control paradigm is a directed acyclic graph lacking backpropagation.
The controlling agent is not part of the system for a reason.
We want to avoid situations where the controller designs a biased system;
which could lean towards personal ownership rather than usage.
We want a systemic judgment but keep it in check with usage.
Let the agent be a feedback function of system usage.

In the learning mechanism:

Transaction + Validity = Consensus

Validity is the sum of feedback functions from agents:

Transaction + Feedback * Agents = Consensus

And usage remains:

Transaction + Consensus = Validity

As a result, learning dynamics of THE answer for an adaptable protocol are:

Case + Feedback * Agents = Laws

A supervisory authority is not required to evolve protocols.
Commands are a poor substitute for broken cycles of self-regulation.
Regulated usage and learned adaptability can be made complementary.
The two system functions keep each other in check.
Compliant usage capitalizes on learnings from free usage in a closed loop.
A belief system governed by such feedback, maximizes users' utility.
In a belief system of rules, laws keep evolving like a consensus.
A fluctuating level of disagreements drives change.

Note that a reliable rule must simplify circumstantial experiences.
Out of diversity, simplicity isn't simplistic.
Feedback must be far greater than complexity to synthesize it.
Only then statistical order emerges from chaos in a comprehending rule.
That is why ordinary rules are believed to apply when not dictated.
In a pure social setting, people behave according to learned habits.
Internalized laws either avoid conflict or maximize life chances.
Thus, verified rules guide behaviors for growth and survival.
Interactive rules are shared because one life learns only a tiny amount.
Legislative cooperation is a lifesaver.

From a tribal perspective, we were often on the very verge of extinction.
Up till 10000 BCE, the world population was often well under one million.
The earliest known testimony of written laws dates from 2100–2050 BCE.
Murder, robbery, and rape were capital offenses in the Code of Ur-Nammu.
Later, the Ten Commandments still appeared in a self-preservation context.
But a single set of ten rules for inner peace cannot cope with us anymore.
Between divides of an overpopulated world, it is now often believed that:
the middle finger of God inscribed commandments on two tablets of stone.
We reached another extreme where vital coordination challenges hard rules.

In Ani's Papyrus, 42 sins ensured cohesion to 42 administrative regions.
Everyday laws unified society and the kingdom of ancient Egypt together.
The deceased confessed cooperation in the Book of the Dead's judgment.
Division was the ultimate death sentence for the kingdom.
Accordingly, the Devourer of the Dead, Ammit, punished divisive behavior.
The chimeric creature was made of the three largest human-eating beasts.
But life regulation across different beliefs cannot rely on common laws.
Means of relative regulations have yet to be devised to face diversity.
Presidential whims do not have grounding in democratic foreign policies.

Many hope to avoid the deadly struggles leading to a global empire.
The end state of centralization is even more to be feared.
No empire could last for very long.
The homogenization of the rule of law is the end of resilience.
Monotheism forced by heretic pharaoh Akhenaten had been hastily canceled.
A kingdom's fate is to be diverse or to be divided.
The disparition of leveled powers introduces a single systemic risk.
History taught that internal checks and balances do not resist authority.
A global civil war could implode humanity.
Hopefully, governance does not have to be designed around a single will.

A Eureka

Fast forward in Summer 2021.
I heard about Hayek's book "Denationalization of Money".
I had it in mind when I read "The Use of Knowledge in Society".
It is a paper from the same author.
The book was not as easy to download.
I was almost tipsy, very late on a Friday night.
I had gone home because Covid restrictions were closing places early.
I could not share my good mood outside like usual.
I finished the night listening to Fauré's Requiem while reading Hayek:
skipping paragraphs talking about methodology because I couldn't follow.
And eating almond paste cookies because I had nothing better to do.
Why I was listening to a requiem was not obvious to me on that night.
Friends opined that nocturnes for piano would have been more appropriate.
Nietzsche's spirit was singing the death of God.

Two weeks before Easter 2022, I woke up from the disillusioned.
I recalled what Skobelev told me four years earlier in Toledo.

His last remark was encouraging, he said:

You're looking in the right direction.

I had been looking bottom-up.
I went down the rabbit hole again and found something above me.
The system to manage the protocol was not transcendental.
It is the immanent price system of Friedrich August von Hayek.


02-04-2022

Charon Charpentier