For all it is worth, it is pretty close to being accepted upstream into the SDK: https://github.com/cosmos/cosmos-sdk/pull/9245
Indeed, the argument for minimum commission is not straightforward. It is tied to the reason why Cosmos SDK supports an upper limit on the number of validators.
The limit is there to help find a balance between enough validators to support/decentralize the chain and maintaining chain performance. The official docs state âa Tendermint blockchain gets slower with more validators due to the increased communication complexityâ.
The primary performance factor on Tendermint chains is latency. First, validators need to rent or purchase fast hardware (CPU and Disk) and secondly, locate it where there are less hops/lower latency to reach other validators.
Thus this upper limit exists exactly to ensure that validators have enough profit to afford such HW and network quality. When the upper limit is raised to the point that one can validate with a couple of thousand LUNA at stake, you are most likely bringing in âvalidatorsâ running servers on consumer HW, from home. If 1/3+1 validators run suboptimal configurations, the chain will suffer (this is not the case ATM).
Too large a number of max validators, you end up with ppl with basic technical skills running their own validators for fun, with the main goal of not paying validation fees on their own stake. They donât care if they get more delegations, but they wouldnât hurt, so hey they set their fee to 0%.
20% of the validators run at 0% at this time. Of course, not all of them are running suboptimal setups, but there is another 13% running under 3% commission, which most would agree is not enough to break even on costs (forget profits). So there we have 33% of validators, and probably most of them are running suboptimal setups.
In addition to amateurs, you also have operations like ours, where we are a very small team validating on over a dozen chains, and we do not have the bandwidth to be constantly present on social media, network with big investors, setup a booth at a conference, host an AMA on telegram, but we do invest in infrastructure (including HSMs and enterprise grade disks at Tier III/IV DCs). We have to charge 0% or close to 0%, and subsidize our costs from other chains. After we started on Terra and charged 0% we reached #8 or #9. As soon as we raised our fee to just under 3.5%, we dropped to a position in the 80s⌠Yep, from top 10 to 80 something because of 3.5% (still not enough to cover costs).
Cosmos chain validation is already not a free market - there is an upper limit on the number of validators. Setting this limit too large you end up risking 1/3+1 of validators running suboptimal HW, and a lot of validators, if not most of them, operating at a loss. This is not sustainable. Having no commission floor is a lose-lose situation. Or, depending on how you frame it, benefits a minority at the expense of the majority.
PS: the thesis stated above can easily be tested if a well placed validator (voting power and infrastructure) starts tweaking their timeouts so only 2/3+1 of validators can sign their blocks. this has been done in the past on Cosmos and indeed, every block was on the cusp of 66%+1.
PS2: I suspect the performance requirements do not rise linearly with the number of validators. It rises exponentially.
3 Likes