Behavioral Crypto-Economics: The Challenge and Promise of Blockchain Incentive Design

In 2009, Satoshi Nakamoto designed Bitcoin to align the incentives of computers. In 2018, we are building multi-billion dollar blockchain projects, incentivizing humans using the same principles. We are assuming the efficient market hypothesis and the wisdom of the crowds. But what happens when the crowds are not that wise?

Elad Verbin
Mar 16, 2018 · 10 min read

1. Introduction: Bitcoin, Behavioral Economics, and Cryptoeconomics

Let’s start from the origins, Bitcoin. The concept of incentive design in blockchain originates from the original Bitcoin whitepaper by Satoshi Nakamoto, and is well summarized by Andreas Antonopoulos in his book and videos. Nakamoto used incentive design to achieve a previously-unattained goal: a scientifically-solid, secure, decentralized digital currency. Nakamoto’s design incentivizes miners to secure the network and disincentivizes defection from the protocol’s proper operations. Furthermore, it reasonably aligns the incentives of all stakeholders: miners, users, and developers contributing to the ecosystem. Its open-source nature ensures that an organized attack is not very lucrative, by enabling stakeholders to recognize the attack and defect to other chains. Seen another way, Nakamoto found a clever game-theoretic solution to the classic Byzantine Generals’ Problem, by paying the generals a salary as long as they act honestly, but garnishing that salary if they are caught trying to cheat.¹

  • Gnosis, Augur and other prediction markets attempt to predict the future using a price discovery mechanism: incentivizing users to profit by trying to form accurate predictions of the future, and betting according to these predictions.
  • Steemit incentivizes users to post interesting tidbits, and/or truthfully vote on the quality of other people’s posts. Other reputation systems incentivize users to “upvote” reputable actors, thus creating a Blockchain analogue for humans’ de-facto reputation systems. (None have proven themselves thus far.)
  • Numerai incentivizes data scientists to devise good algorithms for trading in financial markets
  • Futarchy incentivizes users to stake good decisions
  • Ocean incentivizes users to stake good datasets and to provide added value to existing datasets (think Numerai meets Gnosis)
  • Polkadot incentivizes stakeholders to make honest decisions in the network (“validators” and “collators”), to look for bad actors (“fishermen”), and to decide who is trustworthy (“nominators”).

Incentive design is considered one of the killer features of blockchain systems

Overall, incentive design has spread to many exciting applications and is considered one of the killer features of blockchain systems. (More accurately, the killer feature is the ability to implement a fine-grained incentive system in a highly-scalable way, which supports tiny and large incentives alike). This is captured in the spirit of writing by the most prominent blockchain innovators. Trent McConaghy writes in a recent blog post:

Figure 1. Qualitative scatter plot of decentralized systems by their automatibility and size of their Action Space

2. From Satoshi to Steemit

We will now analyze some ways in which incentive design has been extended since the original Bitcoin mold. In Figure 1 we qualitatively map the current landscape of highly-extended incentive design on a two-axis system.

  • The “size of action space” axis describes how many possible actions must be explored in order to maximize reward. In Steemit the possible actions are as wide as the number of possible good posts, while in Bitcoin the effective action space consists of just one action: “mine and validate honestly”. In Numerai, a human is asked to design a good algorithm, and gets rewarded according to its performance.

Blockchain incentive design needs
public policy experts as well as scientists.

We thus see that incentive design has been extended to incentivize intrinsically-human actions (e.g. writing blog posts) in vast action spaces.

3. Building Robust Incentive Systems?

So far, we have established that the design of complex cryptoeconomic systems is a difficult and not-yet-understood task. Thus cryptoeconomic systems will mostly start out broken, and have to undergo repeated iterations of improvements. But what’s wrong with that? Isn’t that the way technical systems have always worked? From the invention of fire and writing, through state-building and the space race, and onto computing, software and the internet — all of these start out deeply flawed, and become better over time.

Footnotes and Edits

(placed on a different page)

Berlin Innovation Ventures

Berlin Innovation Ventures is an early stage venture fund run solely by R&D experts, investing into globally ambitious algorithmic projects out of Berlin.

Elad Verbin

Written by

Berlin-based Computer Scientist. Algorithms developer. Investing in early-stage algo-tech: ML, blockchain, zero knowledge, infrastructural algos. Partner @ BIV.

Berlin Innovation Ventures

Berlin Innovation Ventures is an early stage venture fund run solely by R&D experts, investing into globally ambitious algorithmic projects out of Berlin.