Engage community on interest rate

[JustinDrake]

Started spreading the word in this tweet. So far everyone who replied seems to agree that interest rates need to be increased.

My suggestion is to have no more than 1% inflation with 2^24 ETH (~16.8 million ETH) effective stake, corresponding to 2^19 validators (~0.5m validators).

[nisdas]

@JustinDrake What would be the basis for 1% , or any other arbitrary percentage. Is there any hard data we can rely on when choosing the interest rate for staking ? Right now it feels like everyone is going on gut feeling when choosing the interest rate, which is not a good way of determining what would effectively be the annual increase in the supply of ether in ETH 2.0 .

[vindberg]

It seems low compared to other blockchains. Have you considered adding a warm-up period E.g. the first 5 years it should generally be higher while the validators industry matures and number of transactions are relatively low.

It might help stabilize the interest for validators if inflation is relative to the combined value from transaction fees (maybe of the last epoch).

[JustinDrake]

What would be the basis for 1%

It's a convenient shelling point (nice round number). Telling the community "inflation is less than 1%!" is an easy sell, IMO.

[djrtwo]

I want to see better models on the costs to run a normal validator machine -- say ~5-10 validator instances. Without understanding the fixed costs of a normal consumer setup, discussions of appropriate interest rate are very difficult.

[vbuterin]

I want to see better models on the costs to run a normal validator machine -- say ~5-10 validator instances. Without understanding the fixed costs of a normal consumer setup, discussions of appropriate interest rate are very difficult.

At present the simplest approximation is bandwidth, and for that we have fairly clear numbers: 16 KB per 6 seconds per shard, so minimum 2.67 KB/s, maximum 2.67 KB/s * 1024 shards = 2.67 MB/s.

[crsCR]

I want to see better models on the costs to run a normal validator machine -- say ~5-10 validator instances. Without understanding the fixed costs of a normal consumer setup, discussions of appropriate interest rate are very difficult.

True - also any(?) 'labour' cost of operation needs factoring in.... (Opportunity cost and risk should be factored in too I guess.)

And I'm assuming you're talking about costs for the setup and running of beacon nodes and not just validator clients - referring to: #157

EDIT: I don't think bandwidth on its own offers much as an indicator of total costs - unless it's v. high of course (and I'd guess only the top tier nodes could require high bandwidth). Connection latency due to contention ratios and network infrastructure should be more crucial...

[0xKiwi]

I'd honestly favor an inflation closer to 1%, maybe even around 2%.

Are there any systems planned for ETH2.0 like the difficulty bomb that will force organized upgrades? If not, I'd really favor a automatically reducing inflation system like Bitcoin's.

[cyber-hokie]

I’d like to see model that accounts for a baseline estimate of cost to run N validators. If it is realistic to run 5-10 validators per “node” (maybe host is a better term here?) and the cost of additional validators is insignificant vs the cost of 1 (e.g. cost of compute resources for 1V is the brunt of the expense), then we should be able to estimate a cost and model out a viable interest rate accordingly.

[econoar]

I ran some different scenarios on the reward scale by tweaking the base reward quotient. Results can be seen her: https://twitter.com/econoar/status/1078694032618844160

Example B seems pretty solid to me if we really want to get to that 10mn number.

Discussion on the topic here as well: https://ethresear.ch/t/the-economic-incentives-of-staking-in-serenity/4157/63

[Swader]

Excellent analysis on the topic of profitability: https://tokeneconomy.co/validator-economics-of-ethereum-2-0-part-one-bc188173cdca

[djrtwo]

Closing for now. Good discussions on twitter, reddit, etc.