Many thanks to Quintus Kilbourn, Dan Boneh, and Tomasz Stańczak for feedback and review. Special thanks to Alex Obadia, Nicola Greco, and team for organizing ARIA’s Trust Everything Everywhere workshop where many discussions with attendees, especially Chris Lu and Christian Schroeder De Witt, inspired this post.
Can we trust everything without trusting anyone?
As the world around us becomes more automated via advances in AI and robotics, the question of who and what to trust becomes central to anyone's life. The proliferation of intelligent products, with advanced sensing capabilities, perfect memory and precise recall, and agency in the real world, should make us wary of who "holds their keys". Technology products will have the power to dynamically shift and mold the reality we're living in, this is already starting to happen in the digital world of bits and will happen in the real world of atoms soon.
In the face of this brave new reality, it becomes imperative to develop defensive technologies that can not only protect us against potential damages at the individual level, but also shift the equilibrium to a safe and balanced growth path at the societal level, which can ultimately unlock the steepest path to economic growth and humanity reaching the stars. If we don't do this, the concentration of political power and economic wealth in the hands of a few centralized controllers risks canceling out all the social benefits of technological progress or even send us backwards to a state that some have described as techno-feudalism.
We're already working on technologies that can counter-balance the force of the extreme progress in intelligence and automation: cryptography, privacy, open source and decentralized systems. By investing and accelerating development of these, we can unlock the safe growth path that can lead us to a post-scarcity utopia as the one described in Ian Bank's Culture series, a world in which we have technical infrastructure that is secured and aligned, we don't worry about big risks because we've designed them away, and we share the benefits of progress with everyone. To realize this it's important to design and govern defensive technologies with care to minimize collateral risk: a new cryptographic protocol could enhance defensive abilities but could also enhance offensive abilities by allowing a group of attackers to coordinate undetected.
In this post, we'll go over the most promising avenues for secure technology in both cyber and physical space, and we will also discuss path of adoption that can yield the most benefits to society in the years to come.
Many projects in the Ethereum ecosystem are focusing on these areas, by diving deeper into new specific technologies and new advances we want to highlight a few strategic areas that will be key to realize the vision of Ethereum as the global settlement layer for the machine economy.
At the same time, for this vision to materialize we need to collaborate closely with researchers and developers outside of the Ethereum community (e.g., AI security, safety, cryptography, mechanism design), this post should also serve as a signal that we care about these areas in Ethereum and we already have the building blocks of a platform that can be very useful at solving the problems they care about.
This area is vast. In one perspective the entire blockchain industry is building digital coordination machines that leverage immutable code and cryptography. The intersection of blockchains and coordination of intelligent digital agents and robots is particularly interesting going forward. There's three key areas:
All these coordination primitives together, plus the scalability of mature blockchains, provide the necessary neutral substrate for machine coordination. Ethereum, the largest and most secure programmable blockchain, can be the global verifier of a machine economy secured by cryptographic proofs. Starting from digital agents, to then expand to robots and the physical world.
Society runs on software. Software runs on hardware. There has been tremendous improvement in open-source software over the past decades, which now supports key operations of the internet. From cloud computing infrastructure with Linux, to money, finance, and other social technologies with Bitcoin and Ethereum. Web infrastructure built on Apache and OpenSSL secures and delivers nearly every website we use. However, if there is a hardware vulnerability all the assurances of open-source software usually go out of the window. You can always pay overhead to split computation across different machines, but it's hard to make up for hardware security issues at higher levels of abstraction. That's why it's important to make progress in open-source hardware (OSH).
Hardware is hard because the production process involves many more steps and it's much more capital intensive than software. Moreover, there is a history of closed source hardware and secrets maintained in the physical world by large organizations. However, there's been a few recent initiatives such as GF180 and SKY130 that have open sourced PDKs for down to 130nm process nodes. This is roughly where closed source hardware was 20 years ago, and these open source chips are mainly used for educational purposes today. But if this process continues there will soon be production-grade chips in increasingly more applications.
Beyond fully open chips (where both RTL and GDS are public), a faster-moving category has emerged: open-RTL but closed-GDS designs. These don't depend on the fabs opening their processes, yet they still provide meaningful transparency and security benefits. Projects such as OpenTitan, Caliptra, and XiangShan exemplify this trend. While open RTL stops short of full openness, it allows the community to inspect, catch vulnerabilities, and iterate on logic design, and can even make it feasible to open firmware layers where many hardware trojans and subtle bugs hide.