Incentives in blockchain protocols
View/ Open
Date
31/07/2021Author
Stouka, Aikaterini-Panagiota
Metadata
Abstract
Bitcoin is a digital cryptocurrency supported by the blockchain protocol proposed
by Nakamoto in 2008. The blockchain protocol offers a public transaction ledger,
organized as a sequence of blocks of transactions. The sequence of blocks is maintained in a distributed way by a set of peers called miners.
The main novelty of
the Nakamoto’s protocol is the use of a proof of work scheme in which miners
expend computational power to get a chance to produce a new block and in turn
they get a reward for each block they produce. The success of Bitcoin prompted
a large variety of other blockchain protocols that attempt to improve on various
aspects of the original protocol. Two examples that are related to the present
thesis are (i) variants of the original proof of work, longest chain protocol that attempt to improve its resilience characteristics with respect to adversarial behavior
and (ii) proof of stake blockchain protocols according to which each participant is
elected to produce a block with probability proportional to its stake, rather than
computational power. In this thesis we examine blockchain protocols from a game
theoretic perspective. This means that we consider participants as rational utility
maximizers as opposed to being divided between honestly behaving and adversarial. This thesis mainly focuses on answering the following three questions: (i) do
miners have incentives to follow the blockchain protocol, when all the other participants do so (this is related to the notion of Nash equilibrium) (ii) how can we
design a reward mechanism that promotes decentralisation, by disincentivising
the formation of undesirable large pools in proof of stake blockchain protocols?
(iii) given such a reward mechanism, how can we disincentivise existing pools to
create a cartel and censor other pools’ registration in the blockchain with the
aim to avoid competition? In order to answer the first question we propose a
suitable notion of Nash equilibrium, called “coalition-safe equilibrium with virtual payoffs (EVP)”. This notion allows us to provide (i) a unified picture of the
incentives in the Bitcoin blockchain protocol when the participants are rational
and try to maximize various utilities based on the rewards and the costs, and (ii)
novel results regarding incentives in a fair variant of the Bitcoin protocol called
Fruitchain [PODC 2017, Rafael Pass et al.]. The motivation for the second question that this thesis answers is the following: although Bitcoin was designed to
be executed in a decentralised way without a trusted party, participants tend to
avoid participating directly in the protocol. Instead, they tend to create teams,
called pools, which are managed usually by a single participant, called pool leader
and they follow pool leader’s instructions in order to get paid. For example, very
few pools may have collectively the majority of computational power, something
that could be dangerous for the security of Bitcoin if the operators of these pools
collude. In order to answer the second question we examine how participants in
a proof of stake blockchain protocol should be rewarded so that in a Nash equilibrium they form k pools where k is a parameter. To be more specific, we define
what a reward sharing scheme (RSS) is and we propose an RSS that achieves the
following level of decentralization: (1) it incentivizes participants to form k pools
and (2) it mitigates Sybil behavior [IPTPS 2002] that in our case is related to
how many independent entities are the actual pool leaders of these k pools. In
addition, we provide a formal analysis regarding the equilibria that arise from a
system using this RSS. We discuss at some length also the deployment of such an
RSS in a proof of stake system. We remark that the reward mechanism that was
implemented in the incentivised testnet and the “Shelley update”’ launched by
the company IOHK (Input Output) on the Cardano cryptocurrency was based
on our results. The third question we answer thoroughly and formally in this
thesis relates to a serious concern that arises in the deployment of an RSS and
relates to censorship of transactions. In a proof of stake system in order for a
pool to be registered it should create a special transaction and this transaction
should become part of the ledger in order to be actionable. However, the existing
pools that run the blockchain protocol may not be willing to add such a transaction, i.e., engage in censorship. We provide an anti-censorship mechanism and
we prove the favorable equilibria that arise when such a mechanism is utilized.