Hi everyone,
Ross here from a16z. This is a needed post from GFX Labs and Millie detailing realistic considerations for a transition to on-chain voting. We are supportive of a transition away from off-chain voting for a number of reasons. On-chain voting typically results in increased transparency, efficiency, and decentralization. Here are a few additional considerations after observing a number of other protocols implement on-chain/off-chain voting schemas:
Off-chain + on-chain voting convolutes the governance process. A number of other protocols have governance schemas that require multiple rounds of off-chain voting before progressing to on-chain voting. This has two effects, which should be balanced against one another:
(1) Efficiency is decreased due to repeated voting rounds prior to any action, which is desirable in some cases e.g. modifying core code. This is undesirable for more routine operations.
(2) The true level of proposal support is obfuscated by off-chain voting, sometimes resulting in unexpected outcomes on-chain. Because a significant percentage of voting power belongs to entities that cannot participate in Snapshot voting due to technical constraints, there is often a disparity between Snapshot polling results and actual on-chain votes.
There are many possible avenues to further reduce the cost of voting on-chain. As a starting point, GFX Labs’ estimated cost of $0.14/vote at 30 Gwei is significantly lower than L1 voting. The coming implementation of EIP-4844 will further reduce this cost by up to two orders of magnitude. As we’ve previously discussed in other threads, there are a variety of economic incentive designs to completely alleviate the cost of voting and actually pay ecosystem participants to vote and/or delegate. Additionally, as MakerDAO has demonstrated, the rise of a professional delegate class can also negate the cost of voting for the subset of participants most impacted by voting costs. With all of these factors in mind, we do not think the cost of voting on L2 should materially influence any decisions made regarding on-chain voting mechanics.
Finally, we think it’s important to take a deliberate, thoughtful approach to governance design. Because governance systems tend to ossify over time, it is far easier to “get it right” upfront rather than rapidly implementing an on-chain voting design and then subsequently seeking to overhaul that governance design later on. Optimism is also in an advantageous position to learn from and build upon the governance designs of numerous other protocols that have already explored various on-chain voting mechanics.
Looking forward to everyone’s thoughts and contributions as Optimism moves toward on-chain voting!