Unit 410 supports efforts to further decentralize the Polkadot network by growing the validator set. Through on-chain governance, the community has previously elected to do this gradually rather than through quick increases, which can introduce risk. We’re supportive of this methodology, which is why we will vote nay on Referendum 16 but support more conservative measures like Motion 58.

Some Context: A History of Polkadot’s Validator Set

The number of validators securing Polkadot has grown substantially since the protocol’s May 26, 2020 Proof of Authority launch. At that time, the network consisted of 20 nodes, all operated by the Web3 Foundation. When it was deemed stable in June, it transitioned as planned to Nominated Proof of Stake (NPoS) consensus and was gradually scaled to 197 validators by sudo, the network’s super user which has since been removed.

Referendum 2 was the first governance proposal whose goal was to expand this validator set. It employed the set_validator_count call to immediately adjust the validator count from 197 to 236, an increase of roughly 20%. However, discussion surrounding how to best add validators took another direction and Referendum 2 ultimately failed.

Referendum 7 took the more conservative approach of increasing the validator set by 1 at the end of each era (14,400 blocks, or ~1 day) for 100 eras using the schedule_after call. This referendum reached the simple majority necessary to take effect, but ultimately failed on-chain due to a misconfiguration. Referendum 9 fixed the issue and succeeded in implementing the gradual growth of the validator set after being fast tracked by the Technical Committee and Council Members.

Shortly after that, the Thousand Validators Programme was launched on Polkadot to further drive the decentralization of the network and encourage validators with smaller stakes to join. This same program has been running successfully on Kusama, which now boasts 900 active validators and over 400 in waiting - an awesome feat.

In short: the Web3 Foundation, Parity, and all of the Kusama/Polkadot/Substrate ecosystem have rightfully demonstrated a strong commitment to decentralization. When given the choice, stakeholders have elected for deliberate, controlled measures over swift changes and we strongly support steady movement towards increased decentralization.

Where We Stand Now

As of this writing, there are 297 active validators, 439 in waiting, and a decision on how to best continue. Referendum 16, the current referendum, uses the same set_validator_count call as Referendum 2 to more than double the validator set to 599, which would make this the most dramatic increase thus far.

It may be easy to look at Kusama as justification for rapid expansion in Polkadot, but we believe decentralization is difficult, and to be undertaken cautiously. It’s especially hard with a network as large as Polkadot with qualities that cannot be easily replicated. Scaling the validator set increases bandwidth requirements, makes concurrency more difficult to maintain, and requires a growing number of capable teams.

Kusama is an unparalleled canary network. It informs Polkadot’s direction while driving substantial value all on its own (as evident by its $1.5B+ market capitalization). But even Kusama cannot uncover all the challenges that come with operating a network of Polkadot’s scale. Look no further than the graph of validator nominations becoming too large to fit within a single block in October 2020, forcing nominators with fewer than 200 staked DOTs to be trimmed from this graph. We don’t know what other gotchas, if any, lie in waiting.

That’s why it is critical to proceed with caution, and we’re encouraged to see the community consistently signal that they feel the same. Token holders have overwhelmingly voted nay on Referendum 16, a performance report on key network metrics has been made available, and a new Motion (58), which takes into consideration the recommendations from the performance report, is working its way through on-chain governance.

Informed Decision Making

Performance Report I: 300 Validators Milestone, Polkadot Metrics and Next Steps is excellent in that it gives the community guidance for decision making and clearly communicates which metrics must be monitored moving forward to maintain network security. Important points include:

  • Lowest/AvgStaked Ratio: The ratio between the lowest validator’s stake and the average validator’s stake becomes problematic for network security if it’s below 50%. The network is at approximately 77%, though this figure has dropped as the validator set has expanded.
  • Bandwidth Costs: Each additional peer uses 15-20 kbps of bandwidth, which is small but at scale can make running a validator more costly.
  • Time to Finality: So far increasing the number of validators hasn’t decreased the amount of time to reach finality, but this metric deserves to be monitored moving forward.

Parachains will be enabled on Polkadot in the coming months and, as noted in the report, require additional consensus logic. As a result, a limited subset of validators will be randomly selected each session for parachain consensus. The team developing parachains has estimated the total set of validators can gradually increase to 500 before reevaluation.

In an ideal world, each network stakeholder would be actively doing this type of analysis on their own, but most parties simply don’t have the time or knowledge to examine network performance in this level of detail. In lieu of that, reports such as this one are exemplary for how to educate interested parties, encourage sound decision making, and nurture interest in burgeoning communities.

Next Steps

Education and sound governance will each play a key role as we look to expand Polkadot’s validator set and the community at large.

Performance Report 1 closes by highlighting the need for educational resources for new nominators so they can wisely choose validators, while calling on validators to produce reports of their own. The Thousand Validator Programme uses educational materials as a key tool in onboarding new stakers. We’ve recently added to this body of knowledge by documenting our approach to account security and open sourcing software for secure multisig signing.

Unit 410 will join the majority by voting nay on Referendum 16, but we’re supportive of proposals which take cautious steps to expand the validator set such as those laid out in Performance Report 1. Motion 58, which expands the validator set 1 slot per day for 100 days, should be the next referendum which meets that criteria.

We are so excited about the future of Polkadot, the addition of parachains, and to cast our vote for steady, steadfast growth of the Polkadot ecosystem.

We’d like to thank Will for his review and assistance with sourcing information for this post.