Polkadot Governance Model

Owned by Former user (Deleted)
Last updated: Jul 01, 2019

Polkadot uses a sophisticated model of on-chain governance with specifications targeting some of the inadequacies in the system to create a more fair playing ground for its participants. To do this, it uses techniques such as referenda with adaptive super-majority thresholds and batch approval voting.

Governance on Polkadot is centered around stakeholder voting. Their core principle stands:

All changes to the protocol must be agreed upon by stake-weighted referendum; the majority of stake can always command the network.

But, unlike most on-chain voting systems, Polkadot’s governance model recognizes and anticipates that voter-turnout is rarely 100%. In order to counteract this and ensure ‘sensible’ decisions prevail, Polkadot uses three techniques:

  • A council;
  • Adaptive Quorum Biasing; and
  • Lock-vote multiplying & delayed vote enactment.

Council

To represent passive stakeholders, Polkadot introduces a Council, an on-chain entity comprising a number of actors each represented as an on-chain account. Polkadot proposes a fixed number of 24 people to comprise the council all with a fixed term of 12 months. Members to the council are elected via approval voting.

The election for the council is continuous. All stakeholders can sign their approval to a registered candidate. When a seat becomes free, votes are tallied based on total stake. The most approved candidate then takes the seat. Runner-ups may keep their place for the next tally, but other unsuccessful candidates are removed from the roster.

An elected member can be removed only by a referendum.

The council has two governance purposes:

  • Proposing sensible referenda; and
  • Cancelling uncontroversial referenda that are considered dangerous or malicious.

Uncontroversial is defined as a unanimous vote by the council. A controversial vote is considered a vote in which at least one person dissents.

In order to propose a referenda, a majority of the council must be in favor with no members exercising a veto. A member may only exercise their veto once on a single referenda; if it is re-proposed at a later date, that same member may not veto it again.

In order to cancel a referendum, the council must vote unanimously. Since unanimity is difficult to achieve, it is assumed that the vote will be uncontroversial. This functions as a last resort if there is a significant issue found in the proposal.

Referenda

Referenda are stake-based voting schemes. Each has a specific proposal associated with it in the form of a privileged function call in the runtime. Proposals are voted on during a fixed period, then votes are tallied and if approved, the function call is made.

Referenda may be proposed one of three ways:

  • Publicly submitted proposals;
  • Proposals submitted by the council, either through a non-veto majority or unanimously; or
  • Proposals submitted as part of the enactment of a prior referendum.

All referenda have enactment delays. These are the periods of time between the referendum ending and, if approved, the proposal being enacted. For publicly submitted proposals and proposals approved by the council, the enactment delay is fixed at two weeks. For proposals as part of a prior referenda, the time may be set as desired.

Public Referenda

Proposals for a referenda may be submitted by anyone, but the vote on that referenda does not occur immediately. When submitted, the proposer must stake a deposit above a fixed minimum of $50, then the proposal enters the queue. Once the proposal is in the queue, any person may ‘second’ the motion any number of times by placing a deposit of the same size behind the initial deposit.

At a fixed interval, proposals with the most stake behind them will be removed, the deposits will be released back to the original owners, and a referendum will take place on the proposal. Those who placed a deposit behind the proposal will be marked in approval.

Vote Weighting

All votes are weighted on two criteria:

  • Tokens: the amount of token under ownership by the voter.
  • Time: the amount of time the tokens will be locked after the referendum has ended. This is measured in multiples of enactment delay and bounded between one and six.

Therefore, in this system, in order to vote a voter must lock their tokens up for at least the enactment delay period beyond the end of the referendum. This is Polkadot’s way of curbing vote selling, by ensuring some minimal economic buy-in to the result.

The two values, tokens and time, are multiplied to provide the total number of votes per voter. For example, a voter with six tokens locked up for the minimum duration of time would have the same vote weight as a voter with one token locked up for the maximum amount of time.

Vote Counting

There are three ways to count votes in this system:

  • Majority-carries: a simple comparison of votes, with a threshold of 50% + 1.
  • Positive turnout bias: a heavy super majority of aye votes is required to carry at low turnouts, but as turnout increases towards 100%, this system reverts back to majority-carries.
  • Negative turnout bias: a heavy super majority of nay votes is required to reject at low turnouts, but as turnout increases towards 100%, this system reverts back to majority-carries.

If there is a majority consensus in the council, we use majority-carries voting. When the council is unanimous, Polkadot assumes that the decision is uncontroversial. In this case the burden of proof lies on those against the motion, and negative turnout bias is adhered to. For publicly proposed referenda, the status quo is assumed, so positive turnout bias is used.

These decisions are made based on Adaptive Quorum Biasing, which is a means of altering the required simple majority based upon turnout. With low turnouts the majority needed increases, so as turnout approaches 100%, it approaches 50%.

Controversies

So far as I can tell, Polkadot has not experienced signification controversy regarding its governance model. Its unique design as a blockchain makes the protocol forkless, which removes enfranchisement from users that they would typically have on other blockchains: the decision to fork, or not. Updates to the software, based on the governance model, are implemented automatically.

Zamfir has publicly spoken against this model, and other on-chain governance models like it. While many believe that Polkadot’s model offers much needed improvements compared to other blockchain governance models, Zamfir argues that on-chain government enfranchises the coin holders too much while devaluing the input of node operators and other ‘stakeholders’.

Gavin Wood, founder of Polkadot and former Ethereum Dev, responded to Zamfir’s criticism stating:

Stakeholders have a very clear and broad incentive to do what’s right for the network, which essentially means driving the price up. It’s also unreasonable to believe that node operators are somehow experts on protocol changes.

Current Council Members

Polkadots coin, DOT, is holding private sales of tokens but is not publicly available for trading and purchasing yet. Therefore, the governance model of DOT has not gone into effect.