Machine Trust Language (MTL): Human-Machine Collaboration

Andreas Antonopoulos’s articulation of network-enforced trust primitives (Oct 2015, Feb 2014) could be extended more broadly into the concept of Machine Trust Language (MTL). While blockchains are being popularly conceived as trust machines, and as a new mode of creating societal shared trust, Andreas addresses how at the compositional level, this trust is being generated. The key idea is thinking in terms of a language of trust, of its primitives, its quanta, its elemental pieces, its phonemes, words, and grammar that can be assembled into a computational trust system.

Blockchains are a network-centric trust system that can make and enforce promises. A network is not just a decentralized architecture; a network can have functional properties built into it. Network-centric or network-enforced functionality can thus enable a more complex level of activity. As XML standardized, facilitated, and undergirded Internet I: the Internet of information transfer, MTL could similarly for the Internet II: the Internet of value transfer.

Trust Primitives: Technical Details
The atomistic building blocks of trust, trust primitives, arise from blockchain scripting languages; they are the programming functions or opcodes used to specify different situations. Some examples are OP_CHECKSIG (a script opcode used to verify that a signature is valid) and OP_CHECKLOCKTIMEVERIFY (a script opcode used for a transaction output to be made unspendable until some point in the future).

As human language components are aggregated into different levels (phonemes, morphemes, lexemes, syntax, and context), so too can blockchain trust primitives. These indivisible blockchain trust particles, trust quanta, can be assembled into larger trust structures like payments. One example could be a micropayment channel with bidirectional settlement for vendor payment, for example entered in 1000 blocktime confirmations for 10 millibits. There could be libraries of standard trust primitives that are always included, for example, to verify the signature or multi-signature status of any transaction. The possibility of fine-grained trust primitives is limitless – a very small instruction set can be used as a toolkit for innovation that is composed into infinitely complex macro expressions. Some other examples Andreas mentions in addition to payment channels are stealth addresses, payment codes, and multisig escrows.

More sophisticated examples of in-built blockchain trust are already starting to become conceptual standards. One is Lighthouse, a cryptowallet that has crowdfunding (the ability to pledge funds to an address) as an incorporated feature; essentially a decentralized network Kickstarter program. The Kickstarter functionality is in the program (there is no custodian); just as Bitcoin allows digital currency transfers without a central bank, so too the Lighthouse wallet coordinates crowdfunding for projects without a central intermediary like Kickstarter. A whole series of similar network primitives with embedded trust functionality can be envisioned. These could include crowdfunding, reputation-checking, backfeeding (emergent collaboration), insurance, multisig, payment channels, peer-to-peer tipping (ProTip), compensation, remuneration, micropayments, IP tracking, backup (specified blockchain transaction record-keeping and archival), and advocacy (via third-party oracle like Smart Contract and Early Temple).

Trust as a Feature: Human-Machine Social Contracting
When trust becomes a ‘mere’ assumed included feature as opposed to a marveled at and explicitly designed functionality, we will have really arrived somewhere as a species. In some sense, the entire apparatus and infrastructure known as society has been produced to instill and manage trust. Deception had an evolutionary benefit, but is perhaps a quality that can be reconfigured, first in machine-mediated human interaction, and later in human biology. The longer-term endgame of blockchains-as-algorithmic-trust is human-machine collaboration, particularly in the application of shifting from the labor economy to the actualization economy. Given the increasing potential prevalence of machines in human existence, a looming topic is the kinds of social contracts that may be appropriate to establish between machines and humans. For example, consider what trust primitives might be needed to write a smart contract with your personalized home robot. To open a payment channel with your home robot, first could be identifying the relevant exchange streams for services and data. These might include personal data, life-logging, backup, diagnostics, advice, empathy, sound-boarding, home maintenance services, payments, and record-keeping; a list of operations that make sense to conduct in a ‘payment channel’ structure (e.g.; two-way open transfer over time of value between parties per triggering events).

A New Kind of Language
Here the concept would be considering the possibility space of all language and noticing that there could likely be a bigger range of language than has come into existence so far. There are human languages, computational languages, math, logic, and other systems of semantics and signifying. As seen with examples like math (Husserl), computing algorithms (Wolfram), intelligence (Yudkowsky), and self-assembled locomotion (Lipson) and life forms, what has been seen through the human example may be but a few nodes in a larger possibility space. The bigger query would be what new kinds of language can be made with blockchain trust primitives. Not just solving human problems (e.g.; creating automated trust structures) but creating new languages from these new functionalities. One next step could be applying linguistic theory (Chomsky, etc.), concept theory (Lakoff, Kant, etc.), and mathematics, logic, computation, complexity math, machine-learning, and deep-learning theory to creating platforms for the emergence of new kinds of language. The first task might be to optimize for obvious new types of trust language that might be possible and that might solve low-hanging fruit problems like offloading the cognitive and behavioral energy effort of deception to move to Brin’s Transparent Society. Blockchain trust could be for society what the quantified self fourth-person perspective was for the individual (a trustable independent objective arbitrator of information about reality).

Philosophy: A New Kind of Qualitative Language
A language of trust is undeniably qualitative. Trust is exactly the qualitative easing necessary for society to function, including in more intensive human-machine collaborations, and in larger scale universally-global and extraterrestrial singularity-class endeavors. Is it possible to reach a place with computational language to say what cannot be said with human language? Perhaps not in traditional 1s/0s computational language, but with a new kind of language of qualitative trust primitives, maybe yes. Wittgenstein famously said (the type of) all there is that can be said in the Tractatus, and in this crystallization pointed to what cannot be said, in three domains, ethics, aesthetics, and religion. Now thinking in terms of trust primitives and other qualitative primitives changes the question of what kinds of sentences and language can be written; the grammar and Wittgensteinian language games that can be enacted with blockchains; in an AI DAC and other applications. There could be many diverse blockchain cliometrics implementations in MTL; e.g.; the measurement of social qualitative factors like the amount of liberty in a political system. The notion is qualitative primitives and qualitative machine language; having a pourable bag of trust elements as components. There are trust primitives, and possibly many other kinds of qualitative primitives, for example freedom, autonomy, and choice primitives; idea primitives and innovation primitives; all of these could be on tap in a multi-faceted qualitative machine language to configure a life of crypto enlightenment.