Emma-Jane Mackinnon-Lee
Lead Author · Local Confidential Computing

> Emma-Jane Mackinnon-Lee divides her time between Edinburgh and London. In Edinburgh she keeps a small workshop with a bench of development boards, a logic analyser, a pair of bench supplies and a shelf of discrete TPM modules collected over a decade; in London she spends her days in reading rooms and at standards meetings. Her subject is local confidential computing: the practice of running sensitive code inside hardware-isolated environments on the device that collects the data, and of proving that fact to a neighbouring machine without asking a cloud service to vouch for it.
> She came to the history through the engineering. Building offline attestation rigs for remote sites, where the network might be a satellite link that drops for a day, forced her to ask why the mainstream designs assumed an always-reachable verifier. The answer lay in where each design was born. TPM attestation grew up inside enterprise networks with a directory server down the hall; Intel SGX remote attestation was first built around a vendor attestation service; the confidential-VM generation of SEV-SNP and TDX was built for hyperscale clouds. Each design carried the shape of the place it was first deployed. Her chapters trace those shapes, and her lab work tries to rebuild the same guarantees for a farm gateway, a tide gauge or a community energy meter.
> Her method is to reconstruct each generation of trusted hardware from primary documents: architecture manuals, specification revisions, conference papers and the patch notes that followed every published attack. She reads the TCG TPM 2.0 Library specification the way a compiler engineer reads a changelog, comparing revisions to see which commands were added after which incident. For the attestation chapters she worked through the IETF RATS working group drafts that became RFC 9334 in January 2023, mapping the Attester, Verifier and Relying Party roles onto every earlier system so that readers can compare them on one diagram.
> Alongside the writing she maintains a reference collection of attestation evidence formats: TPM quotes, SGX and TDX quotes, SEV-SNP attestation reports, Arm CCA tokens, DICE certificate chains and Entity Attestation Tokens, each with a small decoder and a note on which fields matter for an offline verifier. The collection is used in her bench experiments where two boards with roots of trust verify each other over a serial cable and exchange signed receipts for a measured piece of work, the smallest possible unit of a machine-to-machine economy.
> She argues throughout the history that the most important shift of the last decade is not any single feature but the move from privacy as a promise to privacy as a property: a design in which an operator cannot read the data, rather than one in which an operator has agreed not to. She is careful to show where that claim holds and where it does not, giving side channels, microcode updates and supply-chain trust as much space as the guarantees themselves.
> Writes and works in English; reads specification drafts and their errata in the original, and keeps a habit of quoting register names exactly as the vendor manuals spell them.
- ·TCG TPM 2.0 Library specification revisions, read side by side
- ·RFC 9334, Remote ATtestation procedureS (RATS) Architecture
- ·Arm Confidential Compute Architecture and Realm Management Monitor specifications
- ·AMD SEV-SNP and Intel TDX architecture whitepapers
- ·Conference papers on transient-execution and controlled-channel attacks
- ·Standards meetings in Europe and North America
- ·A university cleanroom in Grenoble for the open-PDK chapters
- ·Remote coastal field sites where attestation must work offline
- 01Secure Coprocessors and the Trusted Platform Module
- 02Arm TrustZone and the Secure World
- 03Intel SGX and the Enclave
- 04AMD SEV to SEV-SNP: Confidential Virtual Machines
- 05Intel TDX and Arm CCA: The Confidential VM Becomes Universal
- 06Attestation Standards: RATS, DICE, OpenTitan and Caliptra
- 13Epilogue: Local Confidential Computing and the Machine Economy