You should design trust infrastructure

And if you can’t, learn to see it. It shapes your life!

About a 20-minute read, I reckon ⏰

1. Hello!

Hello! I am Baz. I am a cofounder of Tonk. 

On this rainy Saturday morning in May, I’m sat at my desk, at home, here in East London. I’ve spent the day catching up on Firefox tabs, collating data from the past week, and organising my thoughts. These are precious days where I can zoom out from the day-to-day startup noise and see where we’ve been, where we are and what the future looks like.

This is written for you. It’s my attempt to share a way to think about the world, so that you can solve a particularly important set of problems that humanity is facing. This set of problems is unique to this current historical moment. My intention is to equip you with a lens, so that you can spot patterns in the world that can be solved with (what we’re calling) “trust infrastructure”. 

This is a bit of a long read so here’s a map. I’ll cover:

1. How trust shapes our shared social flourishing;

2. The invisible infrastructure we all take for granted when cultivating trust, from markets to contracts and culture;

3. How the internet broke those ancient tools by forcing us into feudal networks and isolated monocultures;

4. Proposals for a new trust infrastructure, and why I think they can usher in a web that’s more productive, more liberated and way more fun.

Warning - yes, I am going to talk, in part, about blockchain and cryptography. Strap in.

“Trust infrastructure is everywhere. It is all around us, even now in this very room. You can see it when you look out your window or when you turn on your television. You can feel it when you go to work, when you go to church - when you pay your taxes. It is the world that has been pulled over your eyes to blind you from the truth.”

2. Arnaud said some important things about trust

Arnaud Schenk is a smart + cool guy who recently said something really important, about trust infrastructure. So what did he say, and what is “trust infrastructure”?

I’ll give my subjective summary of what Arnaud said here:

1. Trust is really - like REALLY - important. For everything.

2. Trust doesn’t really work on the internet, sadly. Oof.

3. This new thing called “trustlessness” is a really important part of the answer!

4. But it’s not the whole answer. Building trust > trustlessness.

5. Funnily enough, the tools we created to make trustlessness possible (new cryptography) can be repurposed to build trust instead. Call these tools “trust infrastructure” and now go build trust on the internet pls.

You should go and read the original, if you’re not following and fully onboard. It’s really good! However I have noticed some terminological confusion when discussing the blogpost and workshopping its implications with some really smart people over the last few weeks. Here I’ll do some ✨conceptual clarification ✨to help get us all on the same page before adding some missing conceptual rungs that will help us to see the shape of an optimistic future.

2.1. Conceptual clarification on key terms

Start with the idea of a claim such as “I ate some pizza last night”. Claims are purported statements of fact made by agents. Agents are often individual human beings, but could also be institutions, or other intelligences.

Then, when Alice claims to Bob “I ate some pizza last night”, Bob can’t empirically verify if Alice’s claim is true. In general, Bob can only empirically verify what he can sense around him, which is limited to his local spatio-temporal existence in a massive universe of unverified facts. In order to make plans for his life, Bob has to figure out whether to trust Alice’s claim. 

Bob faces this challenge for Alice’s claims about (a) her past observations (b) her promises about her future actions and (c) stuff going on right now that only Alice can see, or the auxiliary details of which she wants to keep private.

Bob doesn’t listen to Alice’s claims in a vacuum; they both exist in some social context, such as school playground, a Whatsapp group or a dinner date. 

A claim is untrusted iff Bob doesn’t believe it. Then we can say that for some given context, a claim is trusted iff Bob believes it, where that belief rests on his prior belief that Alice is (in general) truthful to Bob about this sort of claim, in that sort of context. We can also say that a claim is trustless iff Bob believes it, where that belief rests on something other than Alice’s truthfulness to Bob (about that sort of claim, in that sort of context). In other words, Alice could be a compulsive liar, but when they make a trustless claim to Bob, Bob knows it to be true, even though Bob can’t empirically observe the claim, nor trust Alice in general. 

We talked about trusted claims, but we can also talk about high-trust relationships, or trustful relationships. A trustful relationship is a relationship between Alice and Bob such that the claims they make to each other are trusted. Similarly we can also talk of trusted spaces, composed of many trusted relationships in a knitted mesh. Analogously, we can talk of trustless relationships and also trustless spaces.

We have “tools” (in the broadest possible sense of “tool”) to cultivate these relationships and spaces. We’ll get more into it later, but as Arnaud gestures, we can build trustful relationships by cultivating (a) friendliness and (b) common knowledge, and trustless relationships by applying cryptography.

Call these tools trust infrastructure. Trust infrastructure, as I hope you can see, is everywhere. Just like physical infrastructure, our trust infrastructure warps and evolves over time. We should be surprised if our trust infrastructure were to retain the same form forever. 

It sounds weird, but it should also sound obvious, that our trust infrastructure can and should be overhauled - just as any infrastructure should, from time to time. I want to help you see where our current trust infrastructure is letting us down, by sketching its history - what we have, and what the state of the art is. If we can see the history, then we can see the glaring gaps today.

3. A brief history of trust infrastructure

To understand the specific ways in which our current trust infrastructure is kinda sucky, we have to first understand how trustful and trustless relationships are possible in the first place.

3.1. Why trust anyone?

Testimonial epistemology (the study of the principles upon which we believe what other people say) is a pretty-much-solved domain of philosophy. The basic idea is that we believe people because they’ve proven themselves to be reliable truth-tellers, much like we believe a clock that keeps time. 

I’m not going to give a clean analysis of what makes someone a “reliable truth-teller” but I will point to two overwhelmingly powerful factors that determine their reliability and therefore our ability to trust them: friendliness and shared (sub)culture.

Friendliness. All other things being equal, it’s better for people when they believe what’s true and not what’s false, and if someone cares about you, then they’ll do their best to not lie, or misrepresent the world. So friendliness matters. I know this sounds obvious but it will matter for later, when we consider how we come to trust strangers and enemies.

Shared culture. Most knowledge is specific; if we describe any part of the world in any level of detail, we need specialist knowledge. Bob may be overwhelmingly friendly to Alice, but if they don’t speak the same language or share the same reference-points, then Alice can’t understand (let alone trust) Bob. This applies not just at the macro level (French vs. English) but also at the micro level (two English speakers, where only one can speak fluently about crypto). Specialist knowledge is partly composed of specialised concepts describing some domain of the world (think of the distinction between people who understand “cryptocurrency” vs. those that understand the fine-grained conceptual distinctions between “memecoins”, “altcoins” and “stablecoins”).

This shared conceptual understanding forms a shared language in high-trust relationships, and a domain-specific common knowledge (the things we all know to be true, and know that we all know, for a given space). The common knowledge tends to cover key areas, such as (i) a description of things in the world (ii) an account of how that world works and often (iii) some moral precepts for what sorts of things in that world are better or worse. The moral precepts typically presuppose some sort of shared game (in the broadest sense of “game”) with specific goals, that members of the subculture are playing, and acknowledge each other to be playing.

I am so confused. What is “neochibi”?  What is the game here? What are they TALKING ABOUT?!

We call this common knowledge culture, and when it is specific to some domain, we call it subculture. Cultures and subcultures include art (in general), specific artistic canons (such as kitchen sink realism or punk rock), political stories (the American Dream), fandoms and much else besides. Perhaps the most important cultures in history are religious cultures. Religion is a particularly scalable form of culture, which was arguably the thing that made it possible to interact and transact with (i.e. trust) other people far beyond your local community. When you can presuppose that a stranger shares your basic understanding of the world and a foundational moral code, trust is much easier.

Common knowledge is critical for high-trust relationships because it’s the factor that enables for information-dense understanding and dialogue. Culture is the tool we use to agree on the facts and norms we all take for granted (or not) within a given context; it is a precondition for intelligible discourse and social behaviour. Without culture, we quite literally cannot understand each other. Consider, for example, the way in which online comment section debate breaks down into people speaking “past” each other rather than “with” each other - because the languages they are speaking use similar words with distinct meanings, or presuppose different games.

3.2. Friends and nemeses

Let’s revisit the friendliness requirement for trustful relationships. Common knowledge is insufficient for high-trust spaces. There are people with whom we share an enormous amount of context, but are not friendly. Think, for example, of a protagonist and their nemesis (Holmes and Moriarty), or competitors in business (Steve Jobs and Bill Gates), or sport (Klopp and Guardiola). Take this class of high-context, unfriendly relationships and let’s call them “nemeses” to distinguish them from high-context, friendly relationships.

Coordination is important, so how is it possible for nemeses to interact with each other in a way that mimics trust? Thankfully humans have invented a whole infrastructure for nemeses to cooperate - we call it contract law. Contracts make it possible for enemies to interact in a predictable, rule-based way, where the representations and commitments of each party in a contract are enforced by powerful institutions. They are also (with the assistance of lawyers) quasi-programmable, in the sense that different parties can encode all sorts of weird representations and commitments into a contract.

3.3. What about strangers?

We’re giving a history of trust infrastructure, but we haven’t yet covered the most common type of relationship that any agent has - that of their relationship to strangers. Here I’ll define strangers as agents to whom Alice has low-context and low-friendliness. This is a really important set of relationships - perhaps the most important - because they are by far the most populous. 

Humans have invented ways to mimic trust relationships with strangers, by creating systems for cooperation that do not presuppose shared context or friendliness. The most powerful is capitalism itself. 

In a free market, price mechanisms translate across vastly different contexts, condensing vast amounts of context-specific information about incommensurable needs and production costs specific to local cultures and languages, into a single quantity: a price. Prices are specific to goods and services at some particular point in a supply chain, but make it possible for strangers across the globe, with no global “contract” and no common culture, to create everything you see around you, not least a pencil. 

Free-market price mechanisms make it possible to trust strangers’ claims, but not all such claims. The subset of claims that can be expressed by price-finding mechanisms are of the form “it was this difficult for me to produce X”, to represent supply, or “I want Y this much”, to represent demand.

Many claims cannot be expressed in these forms, such as the claims that Alice went to a particular school, or has a particular income, or that Biden genuinely won the 2020 election, or that the Pfizer vaccine is safe.

How then, do we come to agree with strangers on these other sorts of claims? Historically this has been achieved by adding a third party to stranger-to-stranger relationships, that both strangers trust. We call these institutions “trusted third parties” (“TTPs”), and the idea is that if Bob is a stranger to Alice, then Alice can still believe that Bob has a particular income, because a TTP that Alice does trust (such a bank or government) says so. TTPs enable arbitrary attestations of any kind, so long as the TTP can act as the insurer or enforcer of that attestation.

TTPs are vital to human progress because we can’t build context and friendliness with everyone. TTPs, such as the New York Times, HSBC, Hinge and Twitter, are the great links of our human experience that bridge strangers together and make cooperation possible. The NYT says that Biden really did win the election, and Hinge tells my prospective dating matches that I really do look like this. 😉

3.4. A framework to understand trust infrastructure

This is my only real disagreement with Arnaud. Arnaud was really talking about creating trustful spaces online, but I want to say that all the above “tools”, including (i) shared culture (ii) contract law (iii) markets and (iv) TTPs should all fall under the moniker of “trust infrastructure”. It’s just that in the case of contract law, markets and TTPs, these tools build trustless relationships that make human cooperation scalable beyond our local communities.

We have a 2x2 grid, so what about that low-context, friendly corner in the bottom-left? These are people with whom we have relatively little shared language but can assume high friendliness. The best examples Goblin Oats could think of were his mom and pop’. It’s nice to give the ‘rents a shoutout in my thoughtboi blogpost, but these could be any individuals with whom Alice is friendly, despite not having much in common with them.

Now that we’ve sketched the history and state of trust infrastructure today, we can point at where it’s letting us down.

4. Why our trust infrastructure sucks

The three areas where I can see room for massive overhaul are:

1. The whole idea of contract law. Contracts (and lawyers) are expensive, have limited programmability, limited interoperability between legal institutions, and rely on sclerotic institutions. Blockchain & cryptography could disrupt this.

2. The whole idea of trusted third parties. Trusted third parties extract monopolistic taxes, are unreliable and they sometimes misrepresent. But most importantly, TTP infrastructure is not modular or user-defined, and therefore can’t be nearly as expressive as we need it to be. Decentralised trust infrastructure could meet our demand for expressivity.

3. The whole idea of online monoculture. Culture works offline, but online we face the nightmare challenge of nurturing subcultures on technical rails that grant global instant access, for everyone, to every space imaginable. Attribute-based authentication could solve this.

The contract law point should be straightforward to understand. On that point, my friend Robin works for an interesting company that sidesteps expensive lawyers as a way to verify “software supply chains”. We’ll get to how to disrupt contract law in a bit.

4.1. What’s the beef with TTPs?

Trusted third parties have been useful but they also impose a really low ceiling for innovating our trust infrastructure. 

The first problem with TTPs is that - particularly online - TTPs tend to become egregious monopolies, thanks to network effects and winner-takes-all economics. This means they can charge huge margins and dominate entire markets for nothing other than being the first to digitalise the social graph in your brain.

With this monopoly power, TTPs also prove themselves to be an unstable foundation for other markets, limiting the size of innovations that can be composed out of the claims TTPs support. A classic example is Zynga. Facebook had an open API up until the point that it was no longer economically rational for them to keep the API open. Zynga and Facebook had a symbiotic relationship where Facebook controlled the social graph that made Farmville so compelling - up until the point when Zynga decided that they couldn’t bear the platform risk of being totally dependent on foundations controlled by a private company.

There are other problems with TTPs, such as the fact that they can occasionally get things wrong, or turn into honeypots.

However, one of the most compelling problems with TTPs is they tend to be poorly designed as their surface area grows. This is for a really simple but powerful reason - just as with markets, TTPs are really just networks. However, networks tend to be poorly designed if they are controlled from the centre rather than at the edges. For the same reason that controlled economies are unproductive, and that Disneyland could never grow as large as New York City, networks function best when they are modular, permissionless, and user-defined from the bottom. We want to grow like bazaars, not cathedrals. 

If you have a bright idea and want to improve the SMTP or any other decentralised protocol such as Ethereum, you can make an improvement proposal to the governing body and go through the process of getting it adopted. However, if you want to improve the Instagram algorithm or network design (such as making your friends list interoperable with other social platforms), your only option is to somehow get hired to work at Instagram (?), and wait until you get promoted into the right role (?), when you can finally make the change you wanted (?!), while hoping it fits the economic rationality of that firm (oof). 

Consider the TTPs we rely on today, from land registries (sclerotic), to mortgage providers (invasive), to social networks (clickbaity and doomscrolly). Contrast them with the original decentralised networks that power the web (DNS, HTTP, SMTP, TCP/IP, TLS). Decentralised technologies tend towards more meritocratic design when dealing with something as large and ambiguous as a network. I want to imagine a future in which our trust-based coordination with strangers can be user-defined, intelligent and expressive, rather than subjected to the whims of our techno-feudal lords.

4.2. What’s the beef with online culture?

Culture is the thing that makes high-trust spaces possible, but it’s really hard to build on the web, because the very idea of the internet is at odds with common knowledge generation. For any given subculture, we can talk of that subculture’s “inside” (the core Whatsapp groups, the parties and the “room where it happens”) as opposed to its “outside”. 

Then, subcultures also have filters - the subtle trials and tribulations one must go through in order to make it to the inside. Filters are vital for creating high context and therefore a high-trust space because they ensure that everyone inside the group has high friendliness and high common context, which in turn helps to generate even more common knowledge.

The original idea behind the web was that it would enable instantaneous, global, asynchronous communication. In other words, the web destroys all filters, and opens the door to an Eternal September. As expected, the web destroyed our ability to cultivate nuance in public digital forums, and the result is context collapse. I personally see this with digital monocultures, where digital cultures trend towards information-sparse signals that express the lowest-common denominator of its members, which is to say, everyone. But other people see it on Twitter too. I’m sure the pattern pops up in more spaces than I can recall. It’s the digital analogue for the all-too-familiar hipster-IKEA-WeWork industrial chic airspace that crops up everywhere from Seoul to San Francisco.

We have a cure for digital monoculture, in the form of cosy spaces on the web, far from the harsh sunlight of the clearweb. These cosy spaces are designed to replicate the affordances of meatspace, with its handy walls and doors and locks and spatial distance. 

However, the cosy web, composed of private Whatsapp groups and gated Discord channels, sacrifice the very thing that made the web special and different to meatspace - namely, the possibility of open, global communication. The logical conclusion of the cosy web is that each person belongs to their own digital culture of one, with hyper-specific algorithms and the loss of shared meaning. We’re no longer mods, or rockers, or punks, with our local pubs, secret parties and sense of community. Instead, we each get our own Spotify Wrapped, a hyperlocal and hyper-lonely microculture.

From Venkatesh Rao, via Maggie Appleton

Can’t we combine the best of both worlds - the filters of meatspace with the intelligence of cyberspace? Spaces on the internet should have smart filters - filters that are open enough to programmatically permit entry to the “right” people for that space, without being so open as to get co-opted by low-context mops.

In today’s internet, subcultures do not have smart filters. Today I can see at least two palliatives for this. The first is the adoption of social proof-of-work to signal identity on the web, in the form of people signalling deep knowledge and commitment to some world’s lore, relying on subtle memes and in-group terminology. But this is inefficient and often dumb. We should just be permitted into the spaces that merit our membership, rather than having to grovel to the gatekeepers of some group chat.

Similarly, subreddits and Discord channels promise to create spaces for communities to thrive. But even with these community-first social networks, we once again find ourselves dominated by well-meaning but essentially limited TTPs, who act as the arbiters of our online culture. As mentioned, TTPs are expensive, unreliable and not as expressive as we need them to be. 

This is all looking rather pessimistic when it comes to the prospects for trust infrastructure. However, it’s important to see what we’re missing, in order to imagine what could be. A brighter future is possible for cultivating trust on the web!

5. Towards a better trust infrastructure

Hopefully by this point you’re beginning to see (a) the ways in which trust is pervasive (b) the critical role of trust infrastructure and (c) how the internet throws it all into jeopardy. It’s a bit like looking at the sun - it’s so big and bright that it’s hard to see. These patterns crop up in every relationship, saturating every element of your lived digital experience.

I can point to specific examples that make sense to me, but you will have your own lived experience that I will fail to completely understand, or sufficiently articulate. Notice those problems and write them down! They’ve been so pervasive for so long as to lie completely undetected unless you develop a certain smell and sensitivity.

Thankfully, mathematics and engineering point towards how we can radically overhaul our trust infrastructure and prepare the ground for new species of flourishing relationships that we don’t have words for yet. History tells us that it’s hard to talk about the future, because the future tends to look weirder than we can imagine. Let’s try anyhow:

5.1. A better alternative to TTPs

As I hope you can see, TTPs have always only ever been a second-best solution to generating trustless claims between strangers. Thankfully, there are new advanced decentralised technologies that create user-defined, modular and maximally-programmable attestations. The core idea, as Arnaud says, is that with new cryptography, we can make arbitrary proof-based attestations about pretty much anything, such that Alice can trustlessly believe a given claim without reference to a TTP.

This idea is a few years old, but there is recent and ongoing work to massively increase the expressibility of proofs. Right now we can prove statements that can be represented as an “arithmetic circuit”, which doesn’t amount to a whole bunch out of the box. Thankfully, tools like ZK email and TLS notary enable us to attest to anything that gets delivered via email, or over the web via TLS, respectively. This blows open the expression space for attestation.

For example - and here, we are just scratching the surface - agents can make proof-based attestations, in any given context, that they belong to a particular Facebook group (outside of Facebook), that they have a particular income (without revealing it), or that they have a given number of followers on Twitter.

In such a way, decentralised attestation technology is maximally expressive, user-defined and modular in a way that TTP-attested claims could never be. In a decade or few, we will look back at the era of TTP-attested claims and see it as incredibly antiquated and inflexible, much as we look back at batch processing today.

It’s really important to understand why this time is different when it comes to overhauling our trust infrastructure. There’s an old graveyard of decentralised attestation projects, ranging from local identity wallets, to verifiable credential protocols - and despite decades of effort, they never seem to get off the ground. One of the critical barriers preventing such projects from reaching widespread adoption is that our identity already lives inside TTP systems, such as Google and Facebook. This forms a chicken-and-egg scenario that forces ingenious entrepreneurs to solve a bootstrapping problem - there will be no usecase for decentralised attestations without adoption, and no adoption without a usecase.

Projects like Worldcoin give a financial incentive to solve the bootstrapping problem. Fully onchain games force so much activity onchain that the bootstrapping problem is effectively dissolved. ZK email and TLS Notary essentially hijack our centralised identity providers, inheriting their security without their centralisation. So we have new abundant fonts of decentralised claims, kickstarting the flywheel for attestation infrastructure.

5.2. A better alternative to contract law

Maximally expressive attestations are handy in-and-of themselves. However I’m yet to mention the dreaded B-word! Blockchain is such a radically different idea for how computers work that it has many different explanations. Today I’ll introduce blockchain as a machine for creating credible commitments. That is, blockchain is a tool for strangers (and indeed nemeses) to make particular representations and commitments to future condition-sensitive action in a trustless way. A blockchain is a tool for enforceable, programmable promises. If this looks similar to contract law, it should!

☝🏻 desperately trying to find images to represent a “credible commitment machine”

Really, programmable blockchains like Ethereum are just ways for agents to credibly say to each other that “if X, then Y”, where X is something the blockchain can observe, and Y is something the blockchain can enact. In today’s world, the expression space of X and Y is relatively small, given that blockchains can only observe and enact activity that already takes place within the blockchain. In most cases, this is limited to financial transactions, and not a huge bunch else.*

However, by posting proof-based attestations to the blockchain, the blockchain can essentially observe any arbitrary claim. This means that the blockchain can enforce all sorts of contracts, with all sorts of representations and commitments. It does so without the need for expensive lawyers, and it does so in a way that inherits the user-defined and modular affordances of decentralised attestations. This sweeps away the whole question of interoperability between legal jurisdictions and brings greatly-enhanced programmability to contracts.

In crypto circles, we sometimes hear utopian talk of the great merge between real-world assets and onchain security, or the meatpace-cyberspace singularity. Attestations are a realistic, pragmatic way to bring expressiveness to the blockchain and disrupt (among other things) the very idea of contract law. Attestations and blockchains are perfect technologies for each other that magnify the power of each. They promise to overhaul our existing infrastructure for trust and generate all sorts of trustless interactions that were otherwise impossible or egregiously expensive.

*unless you’re also into autonomous worlds, which essentially solve the meatspace-cyberspace reconciliation challenge by bringing a world’s entire state onchain, and therefore into the expressive capture of a a blockchain. At Tonk we think that autonomous worlds will play a vital role in the adoption and manifestation of widespread trust infrastructure for that very reason.

5.3. A better alternative to internet monoculture

Similarly, attestations are a powerful way to generate smart filters for online cultures. Above I said that the whole problem was that filters are either too strict or too open, and that we’re in search of a way to programmatically filter agents into high-context spaces, as they deserve.

Attribute-based authentication is the basic idea that you can permit entry to particular trustful spaces based on potential members’ attestations, simply by defining those attributes, rather than particular members. This promises to grow subcultures to internet-scale, but sidesteps the question of lore-signallling and does so without relying on TTPs. Again, since proof-based attestations are user-defined and modular, they can be infinitely more expressive and interoperable than the trust infrastructure developed by centralised community-oriented TTPs such as Reddit or Discord.

Goblin and I find it useful to visualise the future web as a conceptual space stretched over a billion dimensions, mapping out all possible interests, attributes and claims. You can imagine distinctive zones existing within this latent space, such as “lives in Brighton; enjoys indie pop and left-wing politics”. However, the physics of this latent space are not controlled by any TTP, and our ability to traverse the latent space and gain entry to particular zones does not rely on the feudal railways implemented by TTPs. Instead, we can all have our own user-defined, maximally-expressive vibemobiles to find our spiritual homes online, and connect with people we would never otherwise discover.

We’re into the foggy future now. However, it also makes intuitive sense to me, that when we accelerate the creation of smart-filtered trustful spaces on the web, we can repurpose that same decentralised trust infrastructure to so that spaces can make attestations about those very spaces, and present attestations to other spaces, creating “metaspaces”. For example, your “neighbourhood treasury” space could prove its financial strength to a committee of credit scorers and qualify for “neighbourhood credit”. 

Or perhaps bands can sidestep the perils of record labels and establish a direct relationship with their fans, reserving exclusive experiences to those who can prove they received a particular email receipt of gig attendance. Using ZK email, this could even be autonomously created by the fandom without requiring the cooperation of the ticket company. In that world, there’s no need for TTPs like Patreon to support creators. Artists and fandoms would be self-contained, self-sovereign, wealthier and more culturally productive. 

Talking about the future is lossy, and I give these examples (inevitably somewhat skeuomorphic) because it’s important to illustrate how attestations essentially give us a way to resolve the paradox of online culture generation.

Here are a few more examples of what would be possible with new trust infrastructure (can you see how?):

• Proof of adhering to payment schedules to disintermediate creditors.

• Shipping attestations to disrupt the entire “letters of credit” financial system that intermediates global shipping.

• Proof of authentic review to disrupt commoditised, Amazon-style e-commerce.

• Proof of achieving some outcome in a virtual in-game economy to claim rewards on a programmable, user-defined bounty system.

• Proof that a particular photo was taken in a particular place at a particular moment. 

What else can you see? Where do you see TTPs extracting overwhelming taxes or otherwise prohibiting particular relationships? Where do you see the legal system ignoring problems that are too big to solve or too small to care about? Where do you see online subcultures breaking down due to context collapse? Where can you spot creative souls scattered across the globe, unable to connect with their creative soulmates? How much squad wealth lies latent under the existing web? How much great art, or great communal flourishing? How much new trust is possible, when we overhaul our trust infrastructure? 

6. Why I personally get excited

Crypto, and advanced decentralisation technologies in general, are in search of an endstate metaphor, on a par with “AGI” or “the personal computer”. My proposal is that trust infrastructure is a perfect candidate. It also helps that Gen X’ers and people outside the crypto religion seem to grok it when they hear it. Legibility is important, people. 

There are so many questions remaining, such as how trustful spaces will find each other, how they will be generated, what the role of our personal AI daemons will be, who the early adopters are and how autonomous worlds will incubate trust infrastructure. 

I get very excited about these, all the more because trust infrastructure is perhaps our only answer to creating an expressive and liberal internet that achieves the same goals of illiberal social credit systems without creating a Google Internet Corp. that controls the entire web.

Trust infrastructure is also our best candidate as the killer app for decentralised private compute. That’s one of the reasons I’ve stopped describing Tonk as a decentralised private compute company, and more as a trust infrastructure company. In general, it’s better to frame projects in terms of the problem they solve rather than the technology they use.

7. A call to action

Can you see the trust infrastructure around you? It shapes your life. It needs an overhaul. It will be hard, but at the same time it’s never been easier, or more urgent. This is too big for any one team, or any one company to tackle. Decentralised design is smarter when dealing with internet-scale networks. If you want to fix the web, don’t just ameliorate its symptoms. Instead, transplant another dimension onto the web. Extend the internet right out of your screen and into the rich cacophony of our shared world, with the new dimension of trust.

8. Acknowledgements

My thinking here is thanks to people who have taken the time to answer my questions and articulate their thoughts. I’m particularly grateful to Goblin Oats, with whom I’ve had a complete mind-meld, so this is as much his thinking as mine. This blog post started as a summary of the most recent trust infrastructure working group session led by Arnaud, and informed by Guiltygyoza, Achilleas, Mathcastles, Kara, Jaymo.

May 2024