Blockchains, Trustlessness, and Game Theory

Blockchains are often described as “trustless”:  this doesn’t mean that they can’t be trusted- but instead that the users don’t have to trust each other. In many applications, this is what sets them apart from conventional technologies- the lack of a centralized figure like a bank, government, or utility.

Many people in the blockchain and bitcoin community love the idea of removing a central arbiter of trust, as it resonates with their own political and ideological goals- whether libertarian futurists, or an anarchist open-sourcers. This idealism may have driven many of the early adopters to mine or buy bitcoins, but I argue that the long-term value of blockchains will be dependent on enabling new types of trustless interactions.

Instead of focusing on removing the central figure from existing businesses –an incremental improvement– the bigger impacts may come from creating new markets and applications  where trust issues currently block any interaction.  To understand this, we need to dig into trust issues in more detail.

Because blockchains are immutable, any transaction becomes part of permanent record- no callbacks, no cheating. And because each transaction is cryptographically signed, a user can always be sure that the transaction will be validated on the network.  At a conceptual level, this means that a merchant can sell something to somebody on the other side of the globe, without ever meeting them and without knowing whether they are an honest customer or a criminal.

This is different from conventional payment networks -Visa, MasterCard, Paypal- which allow for  chargebacks when a customer says that the good never arrived. And here we land at the nexus of the trust issue: the customer must trust the merchant to deliver the good, and the merchant must trust the customer not to issue a spurious chargeback. The payment network stands in the middle, trying to mediate the trust issues.

Let’s examine this by using a tool from game theory: a payout matrix for each of the players, where we see the benefits or losses resulting from each strategy. This article has a more in-depth discussion of using game theory for managing trust issues.   We’ll consider two strategies for each side: the merchant may ship the product or fail to ship the product, and the customer may either pay for the product, or pay and issue a chargeback request.

The merchant incurs the cost of production \(C_p\), the transaction cost \(C_t\) of the payment network, a chargeback penalty \(C_c\), and the expected penalty if they do not ship the product and are fined for doing so \(E(C_{ns}\). The customer pays the product price \(P_p\), gains value \(V_p\) from receiving the product, and faces the expected penalty \(E(P_{fc})\) for making a false chargeback. We can form the payout matrices for the vendor (in green) and the customer (in blue), represented below:

\(\)

Vendor
Ship Don’t Ship
Customer Don’t
Chargeback
\(\color{green} P_p – C_t – C_p\)
\(\color{blue} V_p – P_p\)
\(\color{green} P_p – C_t – E(C_{ns})\)
\(\color{blue} -P_p\)
Chargeback \( \color{green} -C_t-C_p-C_c \)
\( \color{blue} V_p – E(P_{fc})\)
\( \color{green} – C_t – C_c)\)
\(\color{blue}0\)

Comparing the outcomes for all scenarios, we see that whenever the expected penalty \(E(P_{fc})\) for making a spurious chargeback is less than the cost of the product, the customer will prefer to file a chargeback. In view of this, the vendor will prefer not to ship the product or to exit the market altogether, as there are no outcomes with net benefits. This scenario is played out in many countries with weak enforcement of wire fraud laws, where vendors will not participate because of the preponderance of false chargebacks- situations where Bitcoin payments have become popular.

For the vendor, this is clearly a bad deal- and you can see that it’s only through a careful balance of incentives and enforcement that we’re able to coax people to behave honestly and move them towards the economically productive outcome! This is a perfect illustration of the type of trust issues which blockchains seek to address.

Now, let’s consider a blockchain-based payment system with a method for monitoring and controlling the delivery of the good (easiest to imagine for music, e-books, or other digital goods). We’ll assume that there is still a transaction cost \(C_t\) which is paid by both sides, but that non-payment will block the good from being delivered, and non-delivery will automatically result in the customer being refunded. The customer can inspect the blockchain and make sure that they trust the delivery mechanism, and the vendor can make sure that the customer has sufficient funds to cover the transaction. Once the agreement is signed, it is immutably written to the blockchain- no chargebacks. We’ll assume that the The payout matrices now look different:

Vendor
Ship Don’t Ship
Customer Don’t
Chargeback
\(\color{green} P_p – C_t – C_p\)
\(\color{blue} V_p – P_p\)
\(\color{green} – C_t \)
\(\color{blue} -C_t\)
Chargeback \( \color{green} -C_t \)
\( \color{blue} -C_t \)
\( \color{green} 0 \)
\(\color{blue}0\)

Both parties now are incentivized to behave honestly, shifting equilibrium to the upper-left quadrant of our payout matrix. In a nutshell, this is the power of blockchains for creating new markets- whether for direct transfers of money (Bitcoin), escrow accounts to enable delivery of digital goods (Bithalo), or smart contracts granting access to hardware assets (Slock.it).

A few observations:

  • When the two parties interact with each other repeatedly, they will learn the trustworthiness of their counterparty: this provides a strong incentive not to cheat. This can be formalized by comparing the expected value of cheating with the net present value of profits from future transactions.
  • Some trust issues can be resolved by providing side payments that reward compliance incrementally. This pay-for-performance approach is common in industry, and could work well with many internet-of-things applications… without ever requiring the complexity of a blockchain or smart contract!
  • With the rise of online markets, we’ve seen well-recognized vendors such as Amazon and Alibaba serving to address trust issues by mediating payments, managing chargeback investigations, and sometimes even handling fulfillment to ensure that payment and delivery are fair. This is another way of getting around trust issues.

Blockchains are a fantastic tool for addressing trust issues- but they are not the only tool, and successful brands have historically grown by successfully navigating these challenges without blockchain technology. However, where the trust issues are complex, the transaction volumes low, or the profits small, blockchains can offer a low-cost way of creating trustless transactions.

 

Thanks to Giancarlo Bigi, Andrea Bracciali, Giovanni Meacci, and Emilio Tuosto for their excellent article “Validation of Decentralised Smart Contracts Through Game Theory and Formal Methods” which outlines a framework for assessing smart contracts using game theory.

Leave a Reply

Your email address will not be published.