The distinctive value of blockchains is that they can be reliable.
Blockchains can be made consistently resistant to manipulation, censorship, or capture. Their design can make them reliably transparent and accessible. Blockchains do not merely have those properties temporarily - we value them because we have good reason to believe this characteristic hardness will persist.
This deceptively simple property has enabled higher-order features. By combining this reliability with a programmable computing environment, blockchains have enabled digital property (like cryptocurrencies, tokens, and NFTs), complex global financial systems, and other use-cases where we care about reliability, like identity systems and social networks.
When we claim that something is reliable, we are making a statement about the future. Whether something - a blockchain, a bridge, a person - actually is reliable will be determined by future events. We do our best to understand the reasons things might turn out one way or the other (is this cryptography secure? Is the engineering sound?), but ultimately we are making an informed prediction.
This fact can make blockchains hard to explain. Other technologies are much easier to demonstrate, because their distinctive qualities appear quickly. In 1992, to prove that email worked all you had to do was try it once. The proof that someone on the other side of the world received and then instantly responded to your message was right there in your inbox.
But demonstrating the value of a blockchain is not so easy. At first, it might be hard to tell the difference between a blockchain application and a centralized one. But over time, they will diverge. A centralized payment processor might censor your transactions, a media network could deplatform you, and a social network could shut off API access or seize your username. A blockchain application, if designed properly, will not.
This means that establishing the value of blockchains relies more on explanation than other technologies. If the value proposition is “developers can build on this API and it will never be turned off”, you can’t simply prove that value with a tech demo because it depends greatly on future outcomes. You must provide reasons, justifications for why such a claim should be trusted. You must argue for it.
But then something magical happens when groups of people are able to form accurate expectations about a blockchain’s future behaviour. That future reliability reaches back through time into the present, influencing and changing behaviour in the now.
When people know they have a solid foundation to build on, they can invest in and improve their digital property, identities, and commons. Developers can build businesses, protocols, and applications knowing that the infrastructure they rely on can’t be pulled out from under them. The difference between centralized and blockchain ecosystems becomes apparent in the present, because the latter is the site of an outpouring of energy, growth, and investment in both private goods and the digital public commons.
The crucial point is that this does not result merely from blockchains being reliable, but from humans being able to form accurate expectations of that reliability.
When we use blockchains, we are always interacting with expectations about future behaviour, whether we realize it or not.
As a user, your decision to store assets on Ethereum is influenced by expectations about whether those funds will still be there in the future. Your decision to invest time and social capital into your ENS, Lens, or Farcaster profile might be influenced by your expectations about whether the protocol is likely to steal that profile away from you in the near future. As a developer, you have expectations about whether Ethereum will suddenly change the rules of the protocol that you rely on as stable infrastructure for your product or business.
Even people who are least actively thinking about these things are influenced by how others think about them. The most casual or low-information user might use a default chosen by someone else who did think about these things (for instance, someone who uses Ethereum simply because that’s where developers chose to build). Or, they might simply interact with the most “popular” ecosystem - a designation selected by the behaviour of crowds and markets which do factor in beliefs about future behaviour.
But how do people form these expectations, whether they are sophisticated entities or the most casual individuals? Are there commonalities between them? What information and experiences shape our decisions to trust something? And what can we say about that process? Can it be good, or bad? Can it be improved, and if so, how?
Let’s call this Trust Experience, or TX.
TX is the set of experiences that shape and inform our expectations about how a blockchain (or other system) will behave in the future. It is the sum of all external inputs which leads us to believe that it will function in a certain way in the future - to trust it, or distrust it.
If User Experience (UX) is about how a person interacts with and experiences a technology, Trust Experience (TX) is about how a person interacts with and experiences forming expectations about the future behaviour of that technology.
The components that influence TX will vary greatly depending on the user and their needs. These inputs could be very diverse, including:
Reading the code of a smart contract you are going to interact with
Seeing celebrities buy an NFT
Noticing an icon in your wallet that shows you’re using Ethereum
Learning about the security properties of different L2s on L2beat
Looking at a dashboard showing the concentration of Bitcoin mining
Verifying a protocol was audited, even if you don’t understand the report
Thinking that Sam Bankman-Fried is a good kid with impressive credentials
Importantly, TX is not just a list of things we would prefer people use to make these judgements. TX is about what actually shapes trust, not what we wish shaped it.
Everyone who interacts with a blockchain or blockchain application will have some kind of trust experience, though it will vary widely. A casual user might need only a social proof (“my friend Josh, who is a big nerd, uses Ethereum”), whereas a developer building an application will engage more deeply (“I’ve read the docs”) and rely on other social proofs (“An ex-Facebook executive is building on Bitcoin”). Large corporations, governments, and regulators might look to certain kinds of credentials or other social or market signals, relying on experts to validate technical information.
TX is not unique to blockchains. There are many technologies and systems that we rely on in modern life where future performance is essential. Cloud services where we store our backups, protocols of various kinds, concrete structures we live inside, and legal contracts are all cases where a significant part of the value lies in what happens (or doesn't happen) in the future. We have built around each of these technologies the components of TX that allows people to gain enough confidence in them to use and rely on them.
But it is still early days for blockchain TX. Today, our TX is a lot like early software UX.
Early software UX was built by people who understood how a computer worked, and it assumed a lot of pre-existing knowledge from its users. The term “user experience” was only coined in 1993, and the theory and practice of making software usable by ordinary people was still nascent.
Today’s blockchain TX is analogous. It is mostly created by people who understand how a blockchain works, and it often assumes pre-existing knowledge from its users. Just as the software industry had to invent the practice of UX design, we must figure out how to craft a TX that can move beyond power users and scale blockchains to billions of people.
What kinds of things contribute to TX? We can identify some general categories that are common across blockchains and other systems.
When deciding to trust something, we often try to gain our own understanding of it. In crypto, this might involve reading white papers, code, documentation, listening to podcasts, consulting various other primary sources, and improving our knowledge of the relevant domains (“how does a hash function work, anyways?”). We try to understand, for ourselves, how these things work and why.
Similarly, if you purchase a house you might want to inspect the foundation yourself or look for mold under the floorboards. And if you are deciding whether to hire a lawyer, you might want to understand their motivations and experience, in order to better model their future behaviour.
Sometimes doing our own research might involve trying to understand the incentives of the actors in a system. Does a cloud provider have a strong incentive to not lose our data, because they would suffer legal penalties if they did? Do the builders of an office tower have liability if it collapses? In a blockchain, are there sufficient incentives to pay stakers or miners to support the protocol in the long-term?
Blockchains have a particular advantage with this component of TX. Blockchains are incredibly transparent, relative to other systems. The internal logic - code, specifications - that shapes their behaviour is open and transparent for anyone to read (for those with the knowledge to understand it).
But there are obvious limits to relying on our own understanding of a system to predict its future behaviour. Many systems, including blockchains, are extremely complex. Even people who have expertise in one component of the system (e.g. consensus protocols) might not have the skills to evaluate other components (e.g. a complex solidity application). Few people can assess primary sources beyond their narrow domain of expertise.
We often base our decisions to trust a system on the advice or recommendations of other people. Even when we are doing our own research, we are still relying on the people who created the educational resources we use to learn.
Sometimes this might simply be people we know: our family, friends, or coworkers. We look to relative expertise in our social circle, linking into chains of trust (“my friend Josh, who is a big nerd, uses Ethereum. He learned about it from his friend Tim, who is an even bigger nerd, and spends a lot of time on ethresear.ch”).
In other cases, we are looking to someone further away on the social graph but with deeper expertise. In crypto, this takes many forms: smart contract auditors, following researchers on twitter, listening to podcasts, trying to figure out who is credible.
L2Beat is an example of a contributor to TX that uses experts (the L2Beat team members who inspect and evaluate the real properties of L2s) to inform and educate users about L2s at scale.
Relying on experts has limitations as well. It depends on a healthy ecosystem that produces and maintains expert networks. For various reasons there may just not be a sufficient supply of experts (particularly experts who speak your language!). Or there might be misaligned incentives which produce “experts” who are not trustworthy (see: bond ratings agencies in 2007, or popular crypto twitter influencers in 2014, 2017, 2021…).
This is one reason why intellectual honesty is such an important value for blockchain ecosystems. A community that values intellectual honesty is perhaps less likely to be overrun by grifters or faux-experts. Or at least it is more likely to maintain enough authentic experts to compete against the bad ones.
Experts need to gain their expertise from somewhere. This underlines the importance of (1), because making it easy for anyone to do their own research might produce a larger number of credible experts who can contribute to the overall TX of a blockchain ecosystem.
We often look to the past to understand the future. This is captured by the “Lindy effect”, the idea that the future life expectancy of a technology is proportionate to its current age. It’s no surprise that in the crypto ecosystem, we often talk about things achieving “Lindy” status.
When we don’t have precise historical analogs, we look to similar systems. Has this type of plane crashed before? Has any blockchain ever failed?
This is a component where blockchain TX tends to be weak, given that blockchains are so new and still evolving relative to other sources of hardness.
We are social animals, accustomed to using social cues and group behaviour to inform our own beliefs. What do my friends think? Does my tribe like this chain, or that one? What does the market say?
These are very powerful components of TX. In some cases they can result in a powerful default that is difficult to change. This can be a positive thing, because it might indicate a justified confidence in the reliability of a system. But a poorly chosen default - or a default that is not updated when the facts change - can lead people to use unsafe systems.
We’re very familiar in crypto with how these components can create terrible TX for a blockchain ecosystem. Celebrities, famous investors, and many “experts” were invested in Terra before that blockchain fell apart. And markets can sometimes be badly wrong, as they were with Enron and FTX.
One of the core goals of the Ethereum community should be to improve the TX of the Ethereum ecosystem.
Blockchains are novel, strange, and unintuitive. People, cultures, and societies have reasonable questions and concerns about relying on them as global infrastructure. What are their limits? Will they fail or become corrupted? Are they safe? For blockchains to succeed, those questions need to be answered, not just once but over and over again. And the way we do that is by creating the best possible TX for our ecosystem.
But how should we approach that task? That project is much bigger than this blog post.
But as a starting point, here are three principles for good blockchain TX:
First, TX needs to create accurate expectations.
TX that consists of an echo chamber of simple, clear, and pleasing explanations about a blockchain, but which ultimately creates false expectations, is bad TX.
Good TX must always be striving to create accurate perceptions of reliability and not just sell the most idealized story. Often, that means being honest about limitations, weaknesses, and risks.
This is again why both transparency and intellectual honesty are such crucial properties for blockchain communities. There are many obvious incentives for actors in crypto ecosystems to deceive others or conceal real limitations or risks in protocol design. Good TX means pushing back against those corrosive influences and holding our ecosystem to a higher standard.
Second, TX is a property of an ecosystem, not an individual product.
If you are an application developer, you have control over the four corners of your application. This gives you significant power to shape your user’s TX.
But your user’s perceptions and expectations about the cryptocurrency or NFTs they use in your application, or about Ethereum as a whole, or about the idea of blockchains in general, largely depends on things outside of your control. The broader ecosystem is what supplies the experts, educational resources, memes, or crowd / market behaviour which shapes a large part of your user’s TX. This is true in non-blockchain cases, as well. The TX of your chequing account is influenced by your expectations about the fiat currency stored inside of it, the legal system of your jurisdiction, the behaviour and reputation of the banking industry at large, and many other factors.
The upside is that applications get to inherit good TX from the ecosystem they are part of. It is easier for a protocol like MakerDAO or Farcaster to make credible claims to their users about how it functions when it is built on an underlying network that has both hardness and great TX.
TX is also a lot more credible when it is not sourced from a single party. Experts might be more credible if they are independent from each other, and even disagree to some extent. Maybe a slightly adversarial ecosystem is more likely to find truth, and prove credible over time.
All of this means that when we’re trying to assess TX, we need to look at the whole ecosystem and not one narrow application. It also means that good TX is not the responsibility of any single actor, but something that we can only achieve together as an ecosystem.
Third, TX needs to scale both up and down to meet people where they are.
Ethereum’s TX needs to serve an extraordinary range of users. Individual users, developers, large corporations, government agencies and regulators, from every nation and culture on earth, all need different things to help them come to an informed understanding about Ethereum’s real properties and reliability.
This means that good TX should meet people where they are. Good TX does not mean naively assuming every ordinary user will read the yellow paper or inspect smart contract code. For many users, good TX might just mean that the defaults provided to them (by the market, their tribe, their regulator, their friends, their wallet) are good enough, and the process that chooses that default is informed by a deeper engagement with blockchain TX.
But making TX as widely accessible as possible does not mean underestimating people or only dumbing things down for them. We should not let a desire for familiar Web2 UX trump the interests of accurate and informed TX.
First, because there are many people who require a high standard for their TX - they want to verify before they trust. Blockchain TX must also scale to meet them.
Second, because good TX - just like good UX - will actually teach people things over time and improve users’ ability to understand complex systems.
Modern UX relies on many conventions, symbols, or conceptual metaphors that were invented and introduced over decades of UX innovation. The interface you are using to read this blog post contains components developed and introduced over decades (a keyboard, GUI, a desktop, a mouse cursor, a scrolling document, a touch screen…). We take them for granted now because technologists were successful at teaching them to us. But they were strange and foreign to people once. Good TX must be able to accomplish something similar.
Stray observations, addendum, caveats:
TX helps make sense of why the crypto community is dominated by debate and argument. We are all trying to reason about what will happen in the future, trying to answer questions that do not have empirical answers: Will Bitcoin’s 21 million supply cap hold? Will Ethereum’s PoS lead to censorship or concentration of power?
TX helps clarify how and why we talk about “education” so much in crypto communities. We implicitly understand that “education” is about more than educating developers about how to build dapps. There is an implicit idea that everyone should be able to learn how and why Ethereum works, even if they are only a very casual user.
TX is of course not the only thing we should care about with blockchains. A blockchain with great TX, and which is very reliable, but which cannot scale to meet demand for blockspace will have limited impact on the world. A blockchain with a terrible developer experience & ecosystem will not attract useful applications.
The elephant in the room is that the “expectation” many blockchain users care about most is the price of some digital asset. There is a speculative mania around crypto, which will likely continue for some time, that can distract people from the underlying value of these platforms. But that underlying value really does exist, and it will still be there when the mania ends. TX is obviously not about future financial performance of some asset, but instead about these more fundamental properties.
One barrier for blockchain TX is that people’s standards for software TX is very low. People have an expectation that software is ephemeral, that you can always hit “undo”, and that the companies that control that software will rise and fall over time. Most people don’t understand that the TX of the digital world can be a lot better, that software actually can have hardness, and so this is not yet a differentiating factor.
Our relation to TX is probably modified or proportionate to our time preference. A blockchain that can meet expectations for 1 year before collapsing is an outright failure. But a blockchain that can meet expectations for 30 years is obviously valuable for many use-cases. Many people use “cheap but centralized” blockchains for the occasional transactions, but would never leave funds on them.
There is an obvious connection here to my article Atoms, Institutions, Blockchains which introduces the concept of “hardness” as the distinctive feature of blockchains which allows them to perform functions otherwise limited to institutions (like law, or money) or nature (like gold). You could think of TX as a framework for how people experience and reason about hardness once it exists. Although you don’t have to read or care about AIB to make use of this post.
There is a popular (though misguided) idea in crypto that the Right Way to gain blockchain mass adoption is to completely hide or abstract the “blockchain” from the user (let’s call it the “veil of abstraction” thesis). TX is intended as a rebuttal to this idea, and maybe a path forward. By framing TX as analogous and complementary to UX, maybe we can find more practical ways as an ecosystem to make crypto more approachable without hiding the essential distinctive properties that separates blockchains from other software, and helping users make informed choices. Good UX and good TX are not incompatible - this is just a design problem at an early stage of a new technology. All problems are soluble.
Dan Finlay’s recent post The Protocol Seeking Protocol contains some complementary ideas, in particular relating to how people (and groups of people) discover and then learn to rely on new sources of hardness. I think this fits neatly into the “TX” framing and Dan’s ideas go deeper than this post on how we do this in practice.
One of the most obvious holes in the blockchain ecosystem’s TX is the lack of any reliable and informed resource that analyzes, measures, and compares key security metrics of different L1 chains. How does the “economic security” of Ethereum compare to Bitcoin, or Tron, or Solana, or Cardano? How should we make those comparisons? What are the metrics we should look to? Who will build the “L2beat” for L1s?
Thanks to Rachel, Tim, Danny, Liam, Josh, Jesse, Dankrad, Justin, Trent, Jason, and ST for their comments on earlier drafts.
Old UX image of Norton Commander from Yuri Samoilov