The Science Behind Delegation and Why You Should Care

Concordium
Concordium
Published in
3 min readJun 2, 2022

By: Christopher Portmann
cp@concordium.com

In the Sirius update of the Concordium blockchain — scheduled to be released by the end of Q2 2022 — Concordium will introduce delegation, which allows CCD holders to delegate their baking power to bakers (aka. miners or validators) and share in the baking rewards. In this article we give a short overview of some security aspects we considered when designing Concordium’s delegation model.

In a proof-of-stake (PoS) blockchain the power of a baker is proportional to their stake, i.e., proportional to the amount of tokens that they lock in the consensus protocol. For example, the probability of being selected to generate a block or the weight that their signature carries depend on this stake. A consensus protocol in a PoS blockchain will typically be secure up to a certain percentage of the stake being controlled by malicious parties. In many modern protocols this threshold is 1/3 of all stake, and we will use this threshold as an example in the rest of this article. So instead of having to trust all parties with administrator rights — as one does in a centralized system — in a PoS blockchain it is sufficient to trust that parties controlling 2/3 of the stake run the honest protocol, but there is no need to trust anyone specifically.

The next natural question is whether it is reasonable to assume that malicious parties cannot control 1/3 of the stake. There are two main arguments as to why this is the case.

  1. It is expensive to buy enough tokens to control 1/3 of the stake.
  2. Owning tokens means having skin in the game. If a chain is successfully attacked, one can expect the value of the token to crash and for the (locked) stake of the attackers to lose all value.

When developing a blockchain, it is thus not sufficient to prove that controlling 1/3 of the stake is necessary to successfully attack the system, it is also important that it remains expensive to control this 1/3. If a blockchain inadvertently introduces a mechanism making it cheap to control stake, a loophole is introduced which could be exploited by attackers. For example, some recent creative attacks on blockchains involve taking out a flash loan, performing the attack, and reimbursing the loan — all within seconds.

Delegation may open such a loophole. So when designing a delegation mechanism, one has to verify that it still remains expensive to attack the blockchain. In a nutshell, delegation is a mechanism by which token holders can delegate the baking power of their tokens to a party that is running a baker. For example, a baker opens a pool and delegators lock their tokens in the pool. The baker’s weight in the consensus protocol is now proportional to this pool, and so are the rewards. Some of the extra rewards are then shared with the delegators.

Reliable bakers will be favored by delegators (as they generate more rewards), so delegation can be seen as a promotion of good bakers. And it also incentivises token holders to buy and lock tokens in such pools. But if one is not careful, delegation opens a security loophole. A baker with very few tokens can now control a large stake, which violates the two principles outlined above: the baker does not own the stake that they control, so they are not affected by a loss of value of the token, and it is not so expensive anymore to control a large stake as one only needs to be a reliable baker instead of buying tokens with one’s own funds.

In order to plug these holes, Concordium has been working with ETH Zurich to analyze our delegation model. We introduce two specific measures that allow for delegation while still making it expensive to control stake. The first is to allow a baker to only gather tokens in their pool for twice their own tokens. The second measure is to have a fixed reward for delegators, instead of a variable one that is set by the baker and would allow dishonest bakers to undercut honest ones. These measures allow us to get the best of both worlds: a delegation mechanism and strong cryptographic security.

--

--

Concordium
Concordium

Concordium with its Zero-knowledge ID enables the creation of regulation-ready dApps balancing decentralization, security, scalability, and regulation.