How Decentralized is the NNS?

A quick review of recent governance voting

As with all things crypto, there’s a lot of discussion around decentralization of the NNS (the controlling DAO of the Internet Computer). So, how decentralized is the NNS?

There’s many ways to measure decentralization, so let’s explore a few.

Vote Control

The NNS allows its voting members (called neurons) to follow other neurons on voting measures (similar to a representative democracy). A few organizations’ neurons have achieved the status of “named neuron”, which means their neurons are identified by named (instead of an id number) and selectable as a followee from the NNS user interface. The current named neurons are DFINITY, ICA (Internet Computer Association), ICPMN (of which I am a voting member), cycle_dao and (recently) ICDEVs.

Governance proposals give a clear indication of how voting power is distributed amongst named neurons. Recently, a typical governance proposal will receive approximately 182 million votes. Those votes are divided as follows:

• 89.6M, or 49.2% of recent votes, from DFINITY (although this number will decrease to 59M [32%] per DFINITY’s new voting strategy).

• 33.6M, or 18.4%, from neurons following ICPMN or cycle_dao (probably split close to even between the two groups, although it’s challenging to say since cycle_dao has followed ICPMN on recent proposals. As an aside - perhaps cycle_dao could clarify their stance on this topic?).

• 9.2M from neurons following DFINITY or ICA.

• About 1M from following ICDEVS (this was prior to ICDEVS becoming a named neuron, so this is sure to increase).

• 49M in manual votes (very impressive!!)

The NNS has made a lot of progress as a DAO in terms of decentralization, but there is still have some room for improvement. Specifically, I’d like to see:

• More named neurons from organizations with a clearly defined objective and voting strategy. More named neurons equals more diversity in opinions and approaches.

• More active voting participation from the 230M in voting power that is currently not engaged in governance voting. A goal of the NNS should be to achieve greater than 90% of total voting power actively engaged in voting and governance deliberation. Currently, that metric stands at 43.9%. Read on for more information on this…

Genesis/non-Genesis

Within the Internet Computer community there’s a lot of discussion between genesis neurons (which tend to be large ICP holders) and non-genesis (which tend to be more community-owned and smaller ICP holders). I discussed the breakdown of who is and isn’t voting in an article a few weeks ago. The table below (taken from the article) shows that about 47M of the 182M votes in proposal 47098 came from genesis neurons, mostly DFINITY (Genesis Unknown in the table below). The remaining ~135M votes come from non-genesis neurons.

230M non-voting

Of the 230M voting power in non-voting neurons, at least 52M votes are in neurons dissolving, almost all of it with dissolve delays less than 3 years. This is a great opportunity to replace non-voting ICP with voting ICP.

In addition, I estimate there is about 20M in ICP (~30M in voting power) within Early and Seed round neurons whose owners have not claimed their neurons (through KYC) or have claimed and not taken action on their neurons. This is voting power and ICP that will remain locked and not voting until their owners’ complete KYC.

There is at most 150M in voting power among neuron holders who are locked and claimed, but not voting. It would be great for the NNS if these owners could take an active role in the governance of the Internet Computer.

DFINITY’s Neurons

As stated above, DFINITY has recently announced that they will only vote with their neurons with ids between 2000-2055, which accounts for 59M votes. These neurons all have half a million ICP in them staked and locked for 8 years. DFINITY will no longer vote with neurons 4018 - 4037, which have a current voting power of around 30M votes. Neurons in the 4xxx range are currently dissolving, so their voting power is gradually decreasing, so even if DFINITY decided to reverse course, these neurons are continually losing voting power until they would have 0 voting power in November of 2023.

DFINITY is currently not merging maturity on neurons 2000-2055 (although it can’t be determined whether they are spawning maturity). Given that, long term, DFINITY will only own neurons 2000-2055, and assuming they never merge maturity with these neurons, DFINITY will have at most 70M voting power long term. Therefore, it is likely that as the NNS attracts more participants and total voting power increases, DFINITY will long term (3+ years) have less than 15% of total active voting power on the NNS. The previous statement does not include neurons following DFINITY’s neuron or actors whose incentives are exactly identical to DFINITY’s incentives.

Conclusion

The NNS is making strides towards decentralization and any metric used to measure this progress has improved in the past few months. There is no one person or organization that can control the approval of a proposal on the NNS. However, there is opportunity for continued progress which is important because it currently is possible for a few people/organizations to pass any proposal. These improvements could include:

• More named neurons by groups with clearly defined objectives

• More ICP locking on the NNS to increase total voting power

• More voting participation by the 230M vote power that is not voting

My hope is that readers of this newsletter will become leaders in the NNS community and take measures to improve the decentralization of the NNS.