Byzantine Askemos Language Layer

issue: On Trust In Code

date author attachments
2013-07-24 13:17:42 +0200 jfw 1

I've been working for a while on an arguable tamper-proof system. It's designed to following closely the ethical and judicial ideas which form the foundation of at least western democracies. (To be sure it's never becomes illegal here.) Program code is therein handled as if it was a contract. Facts in the sense of "seen and understood to be true by the legal system" are established by execution of the same program at various hosts. Those act akin to a witness. The common understanding available to most peers is taken then to be true. There is an obvious recursion: the code itself has to be established as a common known fact before it can be executed (to arrive by a common checksum over an transaction). This recursion terminates, once it finds the "first contract" - the technical equivalent of a constitution.

That's how it works. We can create secret realms at top as well as public registries (a.k.a version control systems with full audit trail down to the hashsum) easily.

For a while we tried: we had a config parameter, which would allow to execute secret or time-variant code. Forget it! Impossible to secure without an omni-privileged administration. And even then: even this power can not proof to itself that the system has not been tampered with. We removed the switch.

[↗986] How the FISA Court Undermines Trust