RE: HF21: SPS and EIP Explained
Witness funding was and is not intended to fund projects after it was cut 80% (I believe in HF12).
Some witnesses do get involved with projects as part of their campaign for votes, and given surplus funds under some conditions, but that's not a core part of witness rewards based on how the budget was designed.
By contrast, witness rewards were intended to support projects prior to being cut 80%. That bundling was not seen as a good approach and was a good part of the motivation for the huge 80% cut. It was always envisioned that: a) existing reward funds could be used to fund projects via voting for posts, and b) something like SPS could be implemented later to more directly fund projects without the somewhat messy process of doing it with posts. It took almost three years to get there, but better late than never.
And you think the current inflation allocation setup is the right way to go for witnesses? Their costs are fixed in USD, which means a tanking steem price could make it un-economical to run a witness node at all... which would put the entire network in jeopardy.
In this system witnesses secure the network, in a POW system the miners secure the network. Miners are not able to mine at a huge profit for long before the system corrects itself. We have no such correcting mechanism with steem...
Instead, network securers continue to get inflation regardless. Fundamentally, that doesn't sound right to me.
With the recent cost reductions to run nodes it also makes some sense for there to be revenue reductions as well and this was the perfect opportunity for that to happen.
There are no cost reductions to run witness nodes as yet. Over time the costs have simply grown, though to some extent this was expected. The MIRA version (with its associated cost reductions) isn't recommended for witnesses.
I don't think the witness reward mechanism is perfect, but if we are looking narrowly at the reward amount, I think the analysis that was done when it was reduced by 80% and set to 10% of total inflation was reasonable and hasn't changed significantly, therefore the split shouldn't change. Proposals to revamp the system in a more fundamental way can be considered. I'm open to looking at it.
Why is it not recommended? That was one of the selling points sold to us by Steemit,Inc... That this new version would help decentralize the platform as it would make running witness nodes cheaper.
I would guess it will eventually get there, but not yet. The main priority was reducing the costs of Steemit's own expensive RPC nodes, and they did that.
MIRA can reduce expenses of running a node (full RPC, seed, witness) but at the cost of a much slower replay. On my witness test node, a replay usually takes 20-22 hours without MIRA but when I enabled MIRA I gave up after 3 days of replay still not completed. For a top witness, it is critical to be able to replay as fast as possible in order to get the node back on track on events such as a HardFork where a replay is required.
MIRA will help with reducing the costs of other nodes but for witnesses nodes it's not recommended for the reason above.
Interesting. Thanks for the explanation. Was that just smoke and mirrors then from steemit,inc saying that MIRA was going to help reduce the costs for running a witness node and thus making steem more secure and more decentralized?
It will help for new comers who don’t have enough votes to produce enough blocks at first. They can do with waiting few days for a replay. After a replay it’s all good. Also good for secondary witness nodes. So if a top witness runs multiple servers then yes it will reduce costs, just not recommended on the primary witness server.
Posted using Partiko iOS
I see. Thanks.