A lot of folks have been telling us recently that @SpaceandTimeDB should deploy our ZK proof of SQL technology with the defense department. Although this is not our focus or something we plan to pursue, it brings up a larger discussion on the future of verifiable compute and how pervasive it will be to society.
TLDR: 1) ZK replaces BFT consensus for just about everything except sequencing 2) the ability to prove to party A that party B didn't tamper has broad applicability in tradfi and defense, but will be slow to adopt in both verticals for different reasons, explained below 3) AI agents will require ZK proofs to pass messages (output of computations) between eachother without manipulation. ZK is much more conducive for autonomous agents which swiftly transact programmatically than for clumsy humans. 4) In conclusion, any vertical where the negative financial/security/legal/personal/defense impact of data manipulation outweighs the non-zero computational overhead of ZK proving... will eventually leverage it
Let's get into it ๐งต๐
1) Bitcoin brought about the idea of BFT consensus for sequencing and inserting rows of transaction data into a globally replicated database, to ensure that database can never be tampered and data can never be lost. Ethereum expanded on this idea, with simple JavaScript-like scripts called Smart Contracts that execute over the "database inserts" of the chain like a stored procedure, to govern the writes to its globally replicated database of transactions. These scripts together can form decentralized applications that are *highly inefficient* yet tamper-resistant when executed redundantly across the various global chain validators.
But those scripts (Smart Contracts) don't need to be executed hundreds/thousands of times redundantly prior to BFT consensus โจifโจ zero knowledge VMs can instead execute them only once and prove their output. But here's the big question: where does their input data come from? Who sequences the transactions ("database inserts") in a fair, censorship-resistant way? BFT consensus is still needed to complement ZK in that aspect.
Now, you're probably thinking... what does this have to do with the defense sector? ๐คจ
Dec 13, 2023 โข 8 tweets โข 3 min read
Lot of discussion recently with crypto minds I respect about AI x Blockchain.
Here is my thesis as to why autonomous agents will be transacting primarily onchain in a couple yearsโฆ ๐งต
1) PKI INFRA. First off, this isnโt primarily about blockchain, but more about private keys. ๐๐ณ
The human instructing an autonomous agent has a credit card, but the AI does not. It has a private key.
Tbf, it could lever payment APIs like Stripe to transact digitally, but Stripe is regulated to Valhalla and is already starting to require human-approval-in-the-loop due to compliance oversight.
Even if Iโm wrong, consider that PKI infra is self-programable (rather than being at the mercy of a 3rd-party API) which lends enormous benefit to the code-generating bot at your bidding. More on this later.
