The Emerging Power of Decentralization
In the last few years since co-founding SingularityNET I’ve gotten fairly deep into the world of blockchain and decentralized networks. To say the least it’s been quite different from the cultures I was previously more professionally familiar with, such as the AI software industry or the AI / complex-systems research community. More exciting in many ways, as well as far more chaotic, confusing and volatile.
While I continue to believe — as I have since early childhood — that AI is the ultimately most radical technology humanity is going to invent and launch, I have also come to more and more firmly feel that decentralized blockchain-based networks have a profound capability for positive transformation of human society and economy … which may well come to play a critical role in bringing about a beneficial Singularity.
In this post I’m going to give some pointers for the curious reader who wants to better understand why I’ve come to feel that way. Many of the links and notions given here will be familiar to natives of the modern crypto community, but I would guess much less so to everyone else.
Amidst all the popular focus placed on the ups and downs of cryptocurrency exchange rates, many folks tend to forget that the core motive for decentralizing software networks was not originally about money — but rather about increasing the odds that humanity’s technological infrastructure evolves and gets used and gets built in a creative, beautiful and broadly beneficial way. The basic intuition is that central control structures, when they intersect human psychology and sociology, tend to lead to corrupt and screwed- up networks — whereas decentralization provides real alternatives.
Decentralized control is of course not a panacea — decentralization is just the absence of centralization, and it can be better or worse than centralized control depending on how it’s implemented. This is true for software networks just as it is for nations; centralized governments tend toward corruption and inefficiency and have many other faults, but a decentralized anarchic alternative like the one we see in say modern Somalia is not really an alternative. If decentralized governance means rival bands of warlords roaming the countryside forcing youth to join their armies, then centralized parliaments and armies and such start seeming not so terrible.
Bitcoin — the first blockchain network — was founded on the vision of decentralized money, which remains an important vision, though by now it’s only one among many interesting applications of blockchain tech and decentralized philosophy.
Modern decentralized finance goes far beyond decentralized money to decentralize all manner of other financial instruments —making them possible without any centralized custodian. My SingularityDAO project aims to apply DeFi to help transform token offerings into a viable path for software startup fundraising.
Practical realization of the concept of smart contracts provided the first practical technology for implementing a secure global computational process network with decentralized ownership and control. Unfortunately the smart contracts implemented on Ethereum, today’s leading smart contract network, are currently ridiculously expensive and slow. However Cardano’s Plutus smart contract platform, slated to launch publicly later this year, seems to have technical promise to fix these issues; and there is also the Ethereum 2 initiative aimed at providing the original smart contract platform with a radical upgrade.
Bitcoin and Ethereum 1’s “proof of work” consensus mechanism achieves decentralized governance at the cost of massive wastage of computational power; Cardano bypasses this by using “proof of stake” instead, and Ethereum 2 will also. Proof of stake means that those who hold enough tokens created by a decentralized network, have voting power in approving transactions on the network. “Proof of reputation” mechanisms are earlier-stage but go beyond this, allowing those who have a high reputation in the network for whatever reason to have significant power in making decisions leading to consensus.
Proof of reputation requires sophisticated and relatively hacking and gaming proof ways to measure reputation in decentralized networks; we have prototyped and experimented with various ideas in this regard in the context of SingularityNET’s weighted liquid rank reputation system (see analysis in the context of artificial societies and multi-agent marketplaces)
Even the most advanced proof of stake and proof of reputation systems out there now still rely on a single replicated or sharded ledger recording all transactions in the system — a design pattern that will pose increasing scalability difficulties as decentralized networks grow in size and importance. The TODA protocol solves this problem via liberating decentralized networks form the need for having a massive common ledger. The common ledger is replaced by a massive “TODA tree” data structure which however can be stored in a distributed and decentralized way across the machines of numerous network participants.
Connecting multiple decentralized networks together into a “decentralized network of decentralized networks” (which is arguably e.g. what the human body is and what human society is) can be done in software on various levels. The DAIA (Decentralized AI Alliance) aims to pull together multiple blockchain projects to make their tokens and networks inter-operable. My SingularityNET Foundation is launching a variety of “layer 2” (SL2) projects that will have their own tokens that interoperate with SingularityNET’s AGI token.
Samuel Smith’s beautiful analysis of meta-platforms ties in closely here — he argues that a meta-platform, which allows others to build platforms on it, provides superior economics to mere platforms, in a similar way to how platforms provide superior economics to companies making individual products/services and selling them to individual consumers.
Decentralizing AI via implementing intelligent processes on decentralized networks could lead in a variety of directions, but it does seem especially amenable to creating systems instantiating Weaver’s notion of Open-Ended Intelligence
Decentralizing not only AI but the whole Web seems imperative if we are to avoid centralized government/corporate capture of the Internet network. The Elastos protocol and framework gives one fantastic, rapidly maturing way to do this; Dfinity may possibly comprise an alternative.
The acute need for decentralization in order to create a fairer and more broadly beneficial economy, as technology wreaks havoc on human employment and moved us toward Singularity, was analyzed by me in a blog post titled How to Save the World .
Just as SingularityNET aims to decentralize AI processing and Elastos aims to decentralize the Web, Ocean Protocol aims to decentralize Big Data storage and Nexus Earth aims to decentralize communication even to the put of putting up their own satellites to connect multiple ground-based devices communicating with each other in a peer to peer network mesh.
Decentralization of control of inter-human communication is also a critical aspect, with e.g. minds.com serving as a decentralized alternative to Facebook, Hive (successor to Steemit) as a decentralized alternative to Reddit, and my own soon-to-launch Xccelerando Media project as a decentralized-community alternative to the centrally-guided communities associated with traditional tech media websites. These applications are important because traditional and social media are increasingly what shapes the mass mind — so if these mind-shaping processes are centralized, they will among other problems end up programming the mass of humanity to agreeably follow centralized dynamics, i.e. to grow up and learn to love the centralized institution modulating their behavior implicitly via dynamically adapting the media they see and how they interact with it.
The connection of decentralized networks with the world’s legal structures is also a source of complexity and subtlety, and not only as regards finance. Is a decentralized network comprising a combination of humans and software agents effectively thought of as a for-profit company, or a nonprofit organization, or a social club, or what? The matter is confused even more by the fact that many of the “utility tokens” used to connote value or manage payment in today’s decentralized networks live in a legal gray area relative to the regulations in many jurisdictions, with unclarity as to whether they are currencies or securities or new forms of intangible assets or what.
The notion of a DAO or “Decentralized Autonomous Organization” is rightly gaining momentum — and there are frameworks like Aragon that make it simple to create organizations (of humans, software agents or a mix) that operate without need of any legal jurisdiction or any mechanism beyond a set of smart contracts. The legal and formal governance issues associated with DAOs can get somewhat subtle. And it’s also the case that practice DAOs generally need to interoperate with individuals or organizations that are pinned to specific physical jurisdictions — which means they will sometimes be treated as if they had some specific legal jurisdiction, e.g. the one where the servers sit or where the DAO’s programmers have citizenship or residence (a theoretical counterexample would be a DAO that ran on servers on boats in international water, having bought the servers and boats and other materials and commissioned the software development via a company that bought them from various parties but then self-destructed). Various jurisdictions have explored giving DAOs unique legal status and the US state of Wyoming has already taken practical steps in this direction.
There is also an important notion of “progressive decentralization” — in which an organization starts out with a more centralized control structure, and then gradually transforms itself into a DAO or something similar. This is often a more realistic way to do things given the early-stage and poorly-understood nature of the DAO structure — folks may understandably be reluctant to throw their resources (time, money or IP) fully and immediately into a new form of entity with unpredictable dynamics and risks (including risks of susceptibility to sophisticated bad actors). SingularityNET and a number of other projects in its ecosystem are structured according to a progressive decentralization framework — where there is some traditional centralized corporate governance initially, the degree of decentralized control increases year on year, and the idea is that eventually the governance will be entirely democratically community-driven.
Many of the regulatory complexities involving current DAOs and other decentralized projects have to do with their intersection with the finance world, via various forms of utility or security tokens. However, it’s worth remembering that there are also ways for decentralized networks to exchange value that don’t rely on money-like one-dimensionally-valued tokens. Various forms of multidimensional money have been explored theoretically, and at SingularityNET we have prototyped versions of “offer networks”, which allow exchange of value via decentralized AI-mediated barter networks, without the need ever arising to quantify the value of an entity being exchanged via something like money or shares.
The replacement of today’s centralized software infrastructures with decentralized alternatives is still in an early stage — but given exponential advance, the shift from preliminary to mature can sometimes happen with unprecedented rapidity. We had better hope the shift happens quickly, because while we work on these cutting-edge decentralized technologies and improving their scope and speed and cost, governments and huge corporations are ingesting more and more of society’s wealth and information into their centralized networks. Of course many of the things these institutions are doing are positive (while sone are much less so), but where humans are concerned as time passes “power corrupts and absolute power corrupts absolutely” — so if we want a global computational infrastructure that is reasonably likely to remain beneficial as we move toward Technological Singularity, decentralized technology appears the only route.