Why decentralization? – Part I

Why is decentralization so important? It’s what cryptocurrency seems to be all about.¹ Yet I couldn’t really put my finger on what it means exactly, or why it is important. Neither could anyone I asked. Most agree it intuitively feels better, but to say why proved to be much harder.

The scope

In three posts, I’m trying to get a grip on why decentralization feels intuitively correct, while at the same time being hard to characterize. It starts with defining decentralization and highlighting its importance. Then I also want see if it is possible to think of a model that balances centralization and decentralization, instead of having to strictly choose between them. To finish we’ll have a cursory look at where decentralization could beneficial, or even essential.

• The first post asks Why decentralization?
• The second post suggests How to decentralize.
• And the third post looks into Decentralization as an answer to Life, the Planet and Everything.

The main objective

Decentralization is currently a hot item.² But before we choose to go all-in on this potentially disruptive innovation, I think we first have to realize exactly what it is we’re choosing for. These 3 posts are not intended as a comprehensive answer. Instead they intend to make us think about what adding decentralization means for us and our lives. And if we are willing to take the bad with the good.

A secondary objective

Besides its meaning for humans, an additional emphasis is put on the importance of decentralization for autonomous machines. Since this is a blog that looks into the implications of a machine economy.

What is decentralization exactly?

Although “decentralization” is considered the gold standard in cryptocurrency, the exact definition or even its distinction from distributed networks is not always made clear.³ The most basic and neutral definition of decentralization is “the dispersion or distribution of functions and powers.” But what does this mean? And why is it important? And is there a point where further decentralization has reduced benefits or even severe disadvantages?

Before looking deeper into the why, how and what of decentralization, let’s have a look at centralization first. What is so bad about centralization that we want to change it? Or even get rid of it?

Disadvantages of centralized services

Centralized services tend to be very comfortable to use. Somebody more competent than you takes care of all the thinking, engineering, testing, safety features, updates, plus the upfront costs. And you can even get all this for free! What’s not to love?

More and more people are waking up to the problems the tech giants’ core business model present. If you’re not aware yet, read Surveillance Giants by Amnesty International.

“Google and Facebook dominate our modern lives – amassing unparalleled power over the digital world by harvesting and monetizing the personal data of billions of people. Their insidious control of our digital lives undermines the very essence of privacy and is one of the defining human rights challenges of our era.”

Kumi Naidoo, Secretary General of Amnesty International

Is Free of Charge truly free?

Centralized services like Facebook, WhatsApp or Google are often free of charge. Others are freemium⁴ where the free part advertises for paid additional features.

However, this does not mean the free part doesn’t cost you anything. In practice you are paying for it in another way. ‘If you’re not paying, you’re the product being sold’ is usually true here. Your preferences, habits, routines, locations, personal connections, and reading history are mapped. The most innocent application for this is marketing. You pay for all the free stuff through the products you buy.

Paying with priceless freedom

On top of paying a premium on your purchases there is an even greater price you could be paying: freedom. Your online behavior is being weaponized against you and the people you love.

With the abundance of data collected about you, it is trivial to extract your political leanings, social beliefs, and moral values. But also behavior like gambling habits, infidelity, sexual preferences, and drug use⁵. In turn, this allows people with that data to covertly manipulate you into voting differently, undermining the foundations of democracy. Or you could just be extorted. And flat out refused for health and car insurance, because of a real or perceived issue in your past.

A counterbalance to curb corporate control

Let’s keep dystopian sci-fi fictional. Monopolies and oligopolies are the authoritarianism of economics. We need to move away from that, before they start governing every aspect of our lives, not just the digital part.

Decentralization by itself does not solve the problems above, but it does provide a basis for you to regain control. For that we go back to the basic definition of decentralization: “dispersion or distribution of functions and powers”. Why is this valuable?

Decentralizing against unknown disasters

We need to let go of the idea of full control. Or that one entity having all functions and powers is able prevent us from harm. Whether that’s a god, government, or the free market system. Highly optimized centralization works perfectly if nothing is changing. But change is one of the few constants in the universe. So having a society based on the assumption of stability is an exercise bound to fail.

We’ve extrapolated our ability to manipulate our surroundings to the point that we think that full control over everything is possible. We only have to map and eliminate any potential problem. The problem is that this way of thinking runs into serious issues with black swan events, because of their unpredictable nature.⁶

Graceful degradation

In order to deal with known unknowns and unknown unknowns we need a different approach. Not aim for unbreakable perfection, but instead design for graceful degradation. Microsoft’s ElectionGuard is a nice example of this. Instead of trying to create an unhackable voting machine, it indicates when a vote was likely compromised. This helps guarantee the integrity of the process, without depending on perfect execution.

Hedging against black swans

What concepts can you implement when you’re up against the unknown? There is no way to prevent such disasters from happening. I don’t believe even an AGI⁷ – if it ever comes into existence – will be able to perfectly predict everything.

What we can do is dampen the severity of unpredictable disaster. But this requires a system that is designed to breakdown in a somewhat controlled and predictable way. In short:

• limit initial exposure
• reduce potential impact
• utilize backups
• and if everything else fails – buy time.

Compartmentalization

Compartmentalization is separating two or more parts of a system. Distributing functions over multiple parts diminishes the attack surface, and reduces its potential damage.

Containment

Containment aims to prevent a malfunction from spreading through the system, which can result in the system cascading into collapse.

Redundancy

Redundancy adds extra elements not strictly required to operate, and work in case other elements fail. They then take over (parts of) their function.

Mitigation

Mitigation is about actions that dilute the effects of a disruption. This is not as much about minimizing the total damage as it is about spreading the load on the system.

These concepts above intend to give decentralization hands and feet. It’s not about one concept of purity, but instead a broad view of what it can entail. Decentralization is not one singular concept that you can choose or reject. On or off. It is a vague, multilayered approach, that is not good or bad. It is a tool that has pros and cons; benefits and costs.

Playing the devil’s advocate

Now comes the confusing part. Above we see – hopefully – a strong case for decentralization.

But some form of centralization is not bad. Pooling resources creates resilience too. Centralization colloquially means efficient. Having a common culture helps relate to others. The concept of centralization aids in sharing, cooperation, and interoperability. And I like that installing a new iPhone is almost a non-event, because Apple controls every aspect of it. It allows me to focus more time on my family, or correcting stuff on the internet.⁸ It’s not all bad. Decentralization is not the be all and end all. At some level a form of centralization will exist, and might even be desired.

The cost of decentralization

Besides change by itself being costly, there are also other costs to consider with decentralization. In short it means the buck stops with you. You are responsible for dealing with errors and losses. Even if they are not your fault. Or even not due to your actions. IOTA’s Trinity Attack Incident is a good example where the error was largely outside of the control of the users. At most you could blame the users for having trusted the wallet; a single point of failure.

This brings us to an interesting paradox. The absolute freedom absolute decentralization brings might be less desirable than it seems at first sight. Is there something in between that could be more appealing than either absolute responsibility and absolute dependency?

Balancing efficiency and resilience

Automation aims to optimize. Its goal is efficiency. This has brought us many benefits as a society. We are richer, healthier and safer than ever, because of automation. But that’s our current state. And efficiency is only valuable from the perspective of that state.

The reason we need to decentralize is resilience. We’ve overoptimized in many areas right now. Any disruption is becoming a problem, because the efficiencies only work in a limited frame of reference. We need more redundancy built in to deal with black swans, because black swans are an unavoidable emergent property of highly organized systems.⁹

But looking at it from a complex adaptive system’s point of view¹⁰, we’ve been optimizing at the cost of diversity. Which decreases resilience. The result is that we’re increasingly less capable of dealing with unexpected change, and more in need of ever more control. Control that might only be an illusion. Regardless of how you look at it – economically, socially, ecologically – this is not sustainable.

Decentralization as a movement, not a target in itself

The goal of change should be movement towards a better balance. Not the polar opposite of the current system. We shouldn’t try to achieve perfect distribution. It’s an undesirable extreme that is as equally bad as excessive centralization.

Decentralization to me is a process. One in which we develop a diverse fabric upon which we build a future proof society. A fabric that gives individuals autonomy; the capacity to choose to cooperate and collaborate. Or not. A steady balance between autonomy and delegating responsibilities. In a system that can easily bend without breaking.

We need the redundancy decentralization offers. Whether it’s because of the next pandemic, effects of climate change, or “just” trade wars.

If there’s anything the Corona pandemic shows, it’s that we can change rapidly when we’re forced to. It’s uncomfortable and somewhat unsustainable, but we can if we must. But will we if the perceived urgency is not high enough?

Next: A model for harmonic decentralization

In the next post, we’re looking into a model that combines a distributed network with elements of centralization in an intuitive way. It could be a carrot to move people towards a more resilient system, before they’re forced with the stick of frantic panic.