Why verifiable location
The framing behind my research -- why credible location-contingent commitments matter, and what it takes to make them work.
My research focuses on verifiable geospatial technologies and their relevance to the spatial governance of intelligent machines.
For several years I have been building and prototyping at the intersection of spatial data science, secure hardware, cryptographic protocols, smart contracts, and decentralized data protocols -- DIDs and verifiable credentials, IPFS, and the like. That work has spanned maritime security, arms control, climate monitoring, and transport. A few core insights have held across all of it:
- Verifying location is only part of the challenge. The real crux is the credible location-contingent commitment -- a technical mechanism that enforces a location-based policy, rather than merely asserting where something is.
- Making that commitment credible means defining the spatial extent and the policies that apply within it, then evaluating a location against those policies to produce a predicate -- a clear result that a system can act on.
- Technically robust verification matters most exactly where someone has an incentive to lie or mislead. In low-stakes settings a self-reported coordinate is fine; in high-stakes ones it is worthless without a way to raise the cost of faking it.
- Spatial data has quirks that resist naive treatment. Coordinates are continuous and imprecise, regions are extents rather than points, and the same measurement can mean different things to different verifiers.
- Location is often sensitive, so verification has to work without forcing disclosure. Privacy-preserving techniques -- zero-knowledge proofs among them -- let a system establish that a location predicate holds without revealing the underlying coordinates.
These patterns hold across use cases even when the mechanisms differ. A drone proving a delivery and a data center proving where its chips run face the same underlying problem -- convincing a skeptical, remote verifier of a claim about where something happened -- and reach for different tools to solve it.
Right now my focus is on advancing the frontier of technical capability, specifically verifying the location of sensitive, static AI deployments: the advanced chips whose whereabouts are becoming a governance lever.
A note on the web3 framing
Much of my earlier work carries a "decentralized" or "web3" framing. A lot of it was designed to integrate with smart-contract applications, so that vocabulary came along with the tools. I unpack that framing in Web3 is a set of design principles.
I have come to think of blockchains as useful in a narrow but real set of cases: where mutually untrusting agents need a shared digital system to coordinate around, as markets and nation states do. Smart contracts also share an interesting property with AI agents. Data submitted to them cannot simply be trusted -- a coordinate is just a number, and nothing about receiving it tells you it is true. The usual client-server assumption, where a server trusts its own inputs, does not hold. That property, rather than any enthusiasm for blockchains as such, is the through-line connecting this earlier work to the questions I care about now.