Wharton's Kevin Werbach and Nicolas Cornell discuss their research on smart contracts.

‘Smart’ contracts on the blockchain are generating a lot of interest because of their innovative nature and potential to substantially boost efficiency in many areas of law and business. But these contracts — digital agreements that automatically fulfill themselves — come with serious limitations as well.

Kevin Werbach and Nicolas (Nico) Cornell, both Wharton professors of legal studies and business ethics, recently spoke with Knowledge at Wharton about their paper on smart contracts, “Contracts Ex Machina,” which literally means ‘contracts from the machine’ — a nod to Greek tragedy’s deus ex machina plot device. The question they tackled was this: Could smart contracts one day replace contract law?

An edited transcript of the conversation follows.

Knowledge at Wharton: What are smart contracts, and can you give us examples of how they work?

Kevin Werbach: A smart contract is an agreement in digital form that is self-executing and self-enforcing. Let’s say that I want to bet Nico about who is going to win the Super Bowl. I think the Eagles are going to win the Super Bowl, and so we bet $100. Now let’s just assume for a moment that it is a legal transaction — we are in Las Vegas or someplace where you can make a sports bet. So the end of the season comes along, the Eagles do not win the Super Bowl because, of course, they never do, and Nico comes to me and says, “Okay, where is my hundred dollars?” If it’s a normal contract, I might say, “Well, I was just kidding,” or “Well, I actually don’t have the money.” He might have to go into court to sue me to enforce the contract.

With a smart contract, we could do that same agreement digitally, so that the moment the Super Bowl happens and it’s clear who won, the contract is automatically enforced. The money gets transferred. There is no party — neither of us, nor even someone else in the middle, an intermediary — that makes the decision. The contract is just automatically enforced. And it can be applied, this approach of smart contracts, to any kind of agreement. Many things are contracts — the rental agreement that you might use for a home, your employment agreement, your business transactions in a company, purchases. Smart contracts, in theory, could take any of those and make them automatically enforceable.

Knowledge at Wharton: Tell us how else they are different from, say, online agreements that we agree to when we sign up for Facebook or Linkedin, or when we set up autopay for our monthly bills?

Werbach: In the paper, we talk about four different categories of increasingly decentralized and increasingly automated contracts. The first is what you described — what we would call just an electronic agreement. So you go to any website that you sign up for, you click a button, and there is a link there. And you can see, typically, an incredibly long and detailed contract that no one ever reads. But that is a human-readable contract. It’s the same contract you could get on paper. It just happens to be on a screen.

One step from that is what Harry Surden, who’s a law professor at [the University of Colorado at Boulder], calls a “data-oriented contract.” So let us now put the terms of the contract in machine-readable form, which limits what we can do in that contract, but we can do it in ways that computers can at least understand what it means to say “a hundred dollars,” or what it means to say, “purchase this share of stock,” or something. The next step is what Surden calls a “computable contract.” So now we are at the point where the machines can, to some extent, process and enforce the contract. But there is still the fallback of the legal system if something goes wrong.

A smart contract, in theory at least, takes away the legal system entirely. Now there is nothing but that digital agreement. That is the entirety of the relationship, and everything from the negotiating of the agreement, all the way to the full enforcement and clearing of the agreement, happens digitally.

Knowledge at Wharton: Why are people excited about smart contracts? What are the benefits?

Nicolas Cornell: There are two primary reasons why people are excited. The first is about gains in efficiency. Any time you can automate a process and remove the human element, there is the possibility of transformative increases in efficiency and reliability. Just think about what computerization has done for other areas of our lives.

Second, people are also excited about the possibility of removing human institutions, and in particular government institutions, from an important function that they currently play. Right now, contracts depend on a legal system administered by a government. This requires trust in a coercive and fallible authority. So for the same reason that bitcoin enthusiasts are excited about the prospect of currency without a government, smart contract enthusiasts are excited about the prospect of contracts without a government-run legal system.

“The reality is, even though we think machines can render contracts effectively, there are lots of situations where they cannot.”–Kevin Werbach

Knowledge at Wharton: Are smart contracts legally enforceable, and how can one get legal relief after the contract is automatically executed?

Cornell: We think the answer is yes, that smart contracts are legally enforceable. There is no reason to think one cannot make a contract in computer code or electronically. We have been doing that for quite a long time now. There is a little bit of a complication because we normally think of contracts as agreements that are intended to be legally enforceable, and smart contracts are, by their nature, intended not to be legally enforced. But we still think that they are changing the rights and obligations between the parties. And that is the important thing for something to be a contract in the legal sense.

How can we get relief afterwards? There are a bunch of legal mechanisms to reverse transactions or disgorge wrongfully obtained funds, but these are legally quite different structurally and potentially more difficult. So by reversing who stands as plaintiff and who is trying to legally change the situation, we really do end up shifting the complexion of the legal dispute that might arise.

Werbach: This is a good example of partly what got us, as legal scholars and professors, interested in this. Smart contracts are a technical innovation, and the enthusiasts, a lot of the engineers, say, “Well, this has nothing to do with the legal system and legal enforcement.” The reality is that it actually forces us to take a closer look at just what the legal system does. And it creates all these kinds of fascinating new issues.

Knowledge at Wharton: In your paper, you list some limitations of smart contracts. What are they?

Werbach: If you read a lot of the breathless pronouncements, smart contracts have basically solved the problem of contract law. We no longer need the legal system or government, as Nico talked about. We don’t have any uncertainty anymore, because contracts are automatically enforced. Well it is, of course, not that simple.

There are two broad classes of challenges that we talk about in the paper — one is a set of legal doctrinal issues. Contract law has all sorts of requirements. For example, we cannot make an illegal contract. I talked about the betting contract before. Let us say we have a contract to kill my mother-in-law. No court is going to enforce that. It is an illegal contract, even if we might write it down and sign it and seal it and so forth. There are other limitations. For example, a contract has to have what is called “consideration.” Someone has to give you something in return for the promise. Contracts cannot be unconscionable, and so on and so forth.

It turns out if you go through all these legal doctrines and say — ‘Well, how does that apply in a system where, again, the contract just enforces itself on this distributed network, with no one making a decision in the middle?’ — you get lots of confusion and lots of problems. So those are the doctrinal issues.

“Smart contract enthusiasts are excited about the prospect of contracts without a government-run legal system.”–Nicolas Cornell

Then there are a huge set of practical issues. The reality is, even though we think machines can render contracts effectively, there are lots of situations where they cannot. Some of those are in situations that are fairly obvious — for example, if a contract says something like ‘you’ll use best efforts,’ which is frequently included in human contracts. But what does that actually mean? How do you reduce that to computer code? Even things that seem simpler to apply, it turns out that people cannot necessarily foresee what is going to happen in the future. So when you put it in that automatically executed digital form and just let the machine run, it creates all sorts of problems in situations we’re already starting to see.

The final thing is, when people negotiate a contract in real life, a traditional contract, there is still the possibility they can re-negotiate it. Maybe at this point in time we do a deal, and we’re certain we want it to be enforced. But then a month later, things change and we both say, ‘All right, no, it’s in our mutual interest to change the contract, or maybe even to get out of the contract entirely, because things have changed.’ The smart contract is not smart enough to do that, unless the parties build into the smart contract code at the beginning the possibility for that modification. Those are just a few examples of the kinds of problems that come up.

Knowledge at Wharton: You also mentioned in your paper the presence of software bugs. We all have experienced computer errors whenever we log into our computers, and certainly smart contracts are not an exception.

Werbach: Absolutely. There’s a whole range of situations, from the classic kind of bug, to things that are uncertain — where humans intended one thing in the contract, but the machines interpreted it a different way. Or they can’t distinguish what people really intended, even if the language seems clear. One of the functions that courts provide in the regular legal system, in contract enforcement, is you can get testimony. You can interpret and look at the terms of the agreement and figure out what it really means.

Knowledge at Wharton: There was a specific example you had in your paper about Ethereum on the blockchain. Can you tell us what happened there?

Werbach: This is a fascinating example. There is this something called the DAO, which stands for Distributed Autonomous Organization. It was a virtual crowdfunding system like Kickstarter, where people contributed currency — in this case crytpocurrency, the virtual currency of Ethereum called ether. People put up about $150 million worth of ether at the time into this system. And the idea is they would be able to vote on projects that would get funded and so forth.

Someone found a bug in the code. They were able to exploit the code in a way that they could siphon off about $60 million of that ether into their own DAO, and then they were going to be able to take it and use it for whatever, or convert it into other currencies. The problem is that, according to the code of the system, it was a legitimate transaction.

“The only way to know that that was theft and not a real transaction was for a human to look at it. The machines had no way of knowing.”–Kevin Werbach

The only way to know that that was theft and not a real transaction was for a human to look at it. The machines had no way of knowing. So in order to fix that problem — not to have people lose $60 million that they had put up — the developers of Ethereum had to basically break the whole system, not just the DAO. All of Ethereum had to do what’s called a “hard fork,” to basically make it as if the transactions had never happened. And that potentially undermines the whole point of smart contracts and of these blockchain systems, which is the notion of immutable distributed trust.

Knowledge at Wharton: You say that smart contracts are both committing to something in the future and yet not exactly making a promise, either. Can you explain that?

Cornell: You can think of there being two ways that you can make a commitment. Imagine I say, ‘I’ll meet you at the bottom of the hill, I promise.’ That’s one way of making a commitment, and it binds me by creating a certain kind of obligation. A different way is I might say, ‘I’ll meet you at the bottom of the hill’ and then just throw myself tumbling down the hill. That is also committing myself to seeing you at the bottom, but it is a different way of doing so, and it is a little blunter. It may be more efficient, but it is not exactly a promise, and it serves a very different function. That’s a rough analogy, but I think the commitment in smart contracts is more like that, and less like the promise.

Knowledge at Wharton: Kevin, do you have any other thoughts?

Werbach: In the paper, we talk about different notions of what ‘contract law’ actually means. And this is partly why this is really interesting to us as scholars: Smart contracts force us to re-examine just what contract law is actually doing. Is it basically prospectively creating new rights, or is it retrospectively — as we argue — creating a mechanism for remediation when something goes wrong? And even though we’re looking at what is in existing contract law, smart contracts actually help us unpack that.

Knowledge at Wharton: What do you think are perhaps the best applications of smart contracts, if you see their limitations?

“A lot of people think the point of contract law is just to make sure that people actually do what they say they are going to do.”–Nicolas Cornell

Werbach: It is hard to say exactly because the potential of this technology is so broad — and certainly situations where it is easier to implement the agreements in digital form. The initial use of smart contracts is for bitcoin. And bitcoin is very strictly limited in what you can do with smart contracts. It allows you to basically transfer bitcoin back and forth, using the smart contract mechanism to enforce that, and it is tightly limited.

These newer smart contract platforms open up a tremendous amount of new possibilities, but that also leads to more things that can go wrong. So really, I think, what we need to see is the development of best practices, development of things like templates — ways to basically create hybrids where you can actually build into the smart contract. For example, I talked about the opportunity to re-negotiate or potentially the opportunity to bring in dispute resolution, whether through a formal court or more informal, say, arbitration mechanism.

Knowledge at Wharton: Given all the shortcomings of smart contracts, as well as their benefits, what role do you see them playing, and how can we make them better?

Werbach: It’s hard to say. I think the reason that we are stumbling is that the opportunity here is just immense. And contracts are everywhere in the world today, everywhere in our life — and especially in the digital world, where almost every interaction is a contract. We clearly are going to see smart contracts get used more and more widely.

And so it is really less of a question of deciding what segment of human activity smart contracts will be used for, but more thinking about how they can be implemented in a way that’s sensible, that addresses these kinds of issues, and that understands the legal system still plays a vital and valuable role.

If what we want is the thing that the legal system does, the thing that courts do relatively effectively, let us still use that mechanism. There is no reason to get rid of it. If what we want, though, is something different — we want that definitive commitment, what Nico was talking about — then smart contracts are great. We are just at the early stages of figuring out how those things get divided.

Cornell: Just to follow up on what Kevin was saying about the different ways of thinking about contract law. Smart contracts are an important way or access point for understanding what contract law does for us. A lot of people think the point of contract law is just to make sure that people actually do what they say they are going to do. Or they see it as a mechanism to ensure that people either perform or pay some sort of damages.

If you think of contract law in this way, as just a mechanism to ensure a certain kind of result, then smart contracts will look like they can serve this function and maybe serve it more efficiently. But this all assumes that the point is to ensure a certain kind of result takes place. If what we want to do is regard our transactions ex post and think not about creating repercussions but adjudicating retrospectively what happened, what went wrong, and how we are going to fix it, then it seems like smart contracts are just not designed to serve that function. That’s what we want the legal system to do.