And how to kill, killer AI! By ( Chief Technology Officer at Nodle) Eliott Teissonniere & ( Co-Founder at Nodle, Science fiction author) Garrett Kinsman - As Civilization rapidly expands off earth, we must ask how we will govern ourselves. With a new space race set off between the US and China, people, cargo, and information will travel across the void. But what does this look like exactly? How exactly will the basic underpinnings of an interplanetary civilization be architected and governed? In our last article we explored how an would operate, and its implications. We also surmised that the economic driver for Martian exploration may not be resources or manufacturing, but simple tax “optimization”. interplanetary financial system The nearly $100 Billion USD that the top six Silicon Valley Giants underpaid in taxes (Amazon, Facebook, Google, Netflix, Apple and Microsoft) between 2011 and 2020, is close to the costs of the entire Apollo Program adjusted for inflation! ( $146.7 Billion 2020 USD, $25.4 billion in 1973 , ). source source As countries band together to make corporations pay their taxes, and the cost of spaceflight reduces exponentially, our push to the heavens begins to look more financially sound. It just might be cheaper to set foot on Mars and declare tax residency there than it could be for an interconnected group of organizations to pay their taxes. You can write off the expenses as well! Governing Mars as a Tax Haven As the Silicon Valley Monarchy uproots itself from the grounds of San Hill Road and impales itself upon the Martian dunes, questions of interplanetary governance come to rise. Under 1966 The Outer Space Treaty, Mars has been declared ungovernable. “Outer space is not subject to national appropriation by claim of sovereignty, by means of use or occupation, or by any other means.” Considering the current state of international law, we don't expect the Outer Space Treaty to last long. Historically speaking, almost every land that has been settled upon has been declared independent from the colonizing party. Look at America, Africa, or India. As soon as a Martian colony maintains some semblance of self sufficiency, it quickly declares independence. Even if a Martian Colony is not independent, there is nothing in the 1966 treaty stating it cannot declare a special economic zone. The Treaty does not even mention the word “tax” once. While the US-China cold war intensifies, we can assume that Martian colonies will be governed by two models, a highly centralized capitalist- communist approach, and a more decentralized capitalist - democratic approach. For this article we will assume that civilization will adopt a blockchain based, democratic decentralized autonomous organization model, also known as a DAO. A DAO brings people together to vote on “proposals” using decentralized technology. Such proposals can do quite a few things and are generally followed by an automated execution, a few non exhaustive examples are below: Funds management: a proposal could suggest spending funds, re-allocating the portfolio of a treasury fund, or even kickstart a colony in the form of martian tokenized bonds Constitutional, and legal management: a proposal could propose to amend a constitution, or a set of bylaws Dispute resolution: a DAO based jury system, like Kleros.io could be put in place in order to resolve disputes and limit escalations Elect Council members: In the case of a representative democracy. Command and Control: Direct and manage international policy and kinetic action. ** ** We suggest leveraging a “one-person-one-vote” governance model in order to guarantee that each citizen of the colony would have a voice of its own, without falling into the trap of establishing a plutocratic system. Each citizen could be added to a registry tracking their right to vote or not, this registry entry would be tied to their public address used on the local colony’s blockchain used to settle transactions and take decisions at scale. Thanks to the use of , it would even be possible to preserve the privacy of voters and thus limit biases when debating on sensitive proposals. zero knowledge proofs Governance Between Bases International Law is generally more of an art today than a solidified rule of law. The compelling fact about decentralized governance is that it can allow for two opposing forces to exchange without trust. An example would be trading energy for stablecoins. One could imagine governance interoperability standards in place for Decentralized Finance, energy and resource agreements, combat, or even Nuclear Command and Control (assuming civilization is still trapped within the current deterrence theory of mutually assured destruction). We would feel far more comfortable with autonomous, decentralized mutually assured destruction versus the current human controlled model. As civilization is not generally stable, and history is defined by conflict, let’s assume we end up in all out, interplanetary war. Martian bases will find themselves isolated, at war with each other, and under attack. DAO Threat Modeling Despite the tremendous command and control issues that decentralized governance poses to a real world settlement, there are many non-kinetic attacks that can be played out. We see later in this piece how these attack vectors could be critical to the survival of the human race. : Whichever voting system is supported by the Martian colony, it will need to take in account the risk of vote buying, where extraordinarily wealthy individuals could provide people a monetary compensation for voting in favor of certain proposals. This is another case where the use of privacy enhancing technologies, such as Zero Knowledge cryptography, could improve the status quo by making it incredibly hard to organize vote buying attacks at scale. For instance, it wouldn’t be possible to prove that somebody actually voted for a certain option, thus it would be incredibly hard to verify that somebody actually did the correct vote and later compensate them, which would make the economics of vote buying attacks collapse. 1) Vote Buying, Sybil attacks : But what if your cryptography has underlying problems? While this opinion is highly controversial, during an interplanetary conflict we should assume that our cryptographic primitives may come under attack through backdoors or advances in computing theory. Cryptography is based on one way functions, essentially math that makes information easy to “encrypt” in one direction, but very hard to “decrypt” the output unless you know the right variables. The only issue, as described it best, is that “none of the functions currently in use have ever been definitively proved to be one-way functions — we don’t even know for sure that true one-way functions exist.” This means that cryptography is not provably secure, it is only “Hard” to attack. With the right computer, some presume that a quantum computer, or even a galactic computer, could create cryptographic vulnerabilities. In addition, some people believe that modern cryptographic standards, which are designed by the same agency created to break codes (The NSA), may have included cryptographic backdoors in the mathematics they selected for common cryptographic standards. Because of this, decentralized Protocols like Polkadot are using non NSA cryptographic primitives (Polkadot even goes so far to describes their reasoning ). Assuming the Internet is still online during an interplanetary conflict, a simple runtime update should be able to solve non hardware based backdoors or cryptographic vulnerabilities (Assuming you can detect such a cryptographic attack is even taking place, and fix it swiftly). 2) Quantum Key attacks Quanta Magazine here Malicious code caused to crash the network could be inserted into a governance or blockchain update proposal. Unless properly audited, could be used to take down or harm the network if voted into effect. 3) Malicious Governance proposals: Another way to attack a DAO is to attack the hardware on which it runs. This can be through hacking, or even backdoors inserted into the hardware itself. Significant work these days is going into securing hardware based voting machines. DARPA is doing some really incredible work here with the , and we are going to assume this technology, combined with open source HSMs, will be integrated into most consumer hardware by the time we are voting on Mars. With decentralized roots of trust, and potentially billions of nodes, we can make hardware attacks very expensive to perform. 4) Hardware attacks: SSITH Program The Internet as we know it has a fairly centralized key scheme and could be fairly easily shut down. We should assume that some of these issues would be solved when deploying an interplanetary internet (mostly to improve scalability and efficiency). Nonetheless, an angry ICANN could certainly revoke the certificates or change the routing tables of core interplanetary governance components. We hope that ICANN adopts more decentralized governance models (like ENS support) in the future, and they have adopted a during the recent war in Ukraine. 5) Internet attack: neutral stance While NTP has been before, Network time protocol attacks can pose threats to interplanetary governance. The Primary time contributor of NTP is the US Navy. We should assume Mars would have its own clocks, not relying on the US Navy. Interplanetary NTP poses some interesting issues and . 6) Attacking time. weaponized research As DAOs evolve and move off earth, we will begin to see new types of DAOs emerge, and this is where things truly get interesting. Muskian Governance Lord Elon Musk proposed a simple tweak to the classical direct democracy approach, which we will call Muskian Governance. We can modify our democratic governance with a core constitution. This approach then includes proposals that when voted into block are time bound. Amendments to this constitution are proposed in governance proposals (EMIPs) which are then voted upon periodically. These proposals once voted onto place have an expiration date based on the block. After a set number of time (As determined by number of blocks, or decentralized time protocol) the proposal will expire. ** ** Dark DAOs As we mentioned before, Zero knowledge (Zk) voting will be essential to preserving the privacy of individuals participating in governance. The participants, functions and potentially even the existence itself of a Dark DAO can be nearly impossible to detect. Previously we coined the term DAO of Devices, where the DAO itself incentivizes hardware like base stations, sensors or servers to be deployed. Hardware that runs Dark DAOs can be hidden in the form of disposable computers, subsea servers, satellites on Lagrange points, and even base stations on distant planets. Distributing blockchain consensus across space and time creates many challenges in performance, of which we dove into in our previous post on interplanetary finance. Distributing blockchain validator nodes across the solar system would make such a Dark DAO extremely difficult to attack. Autonomous DAOs As Artificial intelligence advances exponentially, we will quickly see humans electing AI as core members of a governance committee. These AIs will submit proposals, and eventually be elected to vote on proposals. There is some incredible based around this concept, and it makes one ponder as to what kinds of proposals and votes that an AI would make in regards to current geopolitical events. anime As a first step, AI would leverage big data to create proposals that humans could then vote on. For example leveraging climate models to propose changes to energy and economic policy. AI and humans could further compete on proposals, with competing AI ranking these proposals and modeling their potential implications. This poses a great possibility of enhancing and improving democratic functions. As systems become increasingly relevant, autonomous command and control become key topics to debate. Distributed AI (multiparty neural networks) is a growing field where many different AI powered “nodes” work together to solve problems. Distributed AI helps preserve privacy by processing data locally, and sharing what is learned throughout the network. Startups are working to bring distributed AI on-chain, allowing AI to participate in trustless, decentralized governance models. The most interesting aspect of distributed AI running on-chain is that it is nearly impossible to kill. autonomous swarm weapons While we are still some years away from general, on-chain AI, we imagine a great amount of conflict brewing between AI governance mechanisms and human controlled elements. How would humans stop an AI DAO, distributed across a solar system or galaxy? Intergalactic DAOs Since the invention of the airplane in 1903, the speed of controlled flight has increased exponentially. As civilization expands beyond our host star, and AI helps us solve , we imagine a space opera of intergalactic DAOs spreading out across the heavens. AIs governing billions of inhabitants will make peace, and go to battle with each other, and they may not be human controlled. key civilizational issues Due to limitations on speed of light, simply communicating between the closest DAOs could take decades. Without an Intergalactic transdimensional DAO ( ) how will governance proposals be synchronized across the galaxy? Based on the speed of light we might even be able to calculate the size of a DAO could grow before communication breakdowns lead to new ones being formed. History would tell us that even a few days of communication delay can cause empires to fall apart. Could an AI DAO replicate itself and govern societies across a wide area? see Tachyonic antitelephone We like to think that AI and humans will live constructively towards a common goals of abundance. In order to map our threat model we must have to imagine a malicious AI taking control over a DAO. You cannot simply “pull the plug” on a distributed network with billions, or potentially trillions of nodes. This has given us an interesting topic to speculate upon. How do you go to war with an intergalactic DAO? Our best guess so far is to hack the blockchain itself. We can assume that key nodes will be relatively localized due to speed of light performance limitations. These would be wise old nodes with centuries of high ranking proposals. If such a system can be compromised, and a malicious proposal can be crafted in such a way to take down the network (Likely with the help of other AIs), then AI based DAOs could be attacked from within. Trustless operating environments and autonomous error recovery would make such an attack truly of epic proportions. We envision a science fiction epic where future elite hackers fly to distant and ancient rainforest temples, on the hunt to locate and disassemble subterranean governance infrastructure. Perhaps the hackers communicate with other compromised nodes within the solar system using a portable, free space optics, x-ray datalink. Such hackers would have to time the attack with the motions of the planets, maintaining line of sight beneath the forest canopy to execute the proposed governance update at just the right time. We liked the idea so much, we contributed to on the idea. a science fiction trilogy Whatever happens, the future of DAOs will be incredibly exciting, as we push what it means to be democratic, spread across space and time.
