The economic incentives of staking in Serenity


#22

Any thoughts on capping the total number of ether? Doesn’t solve the inflation issue, but does give people a sense of scarcity, like bitcoin. Just seems like it is part of the economic landscape. I know this has been a hot topic in the past, although I haven’t paid too much attention to it, so I’m not pushing it… feel free to dismiss this comment.


#23

I wouldn’t think about this ATM.

Even Bitcoin’s 21M cap is absolutely unsustainable when coupled with PoW, i.e. no state-of-the-art crypto can exist/survive without inflation.

However, as tech matures and becomes more efficient, I think this will probably become a sensical discussion topic.


#24

Full agree with this @MihailoBjelic. Focusing on a cap right now is the wrong thing. Perhaps some day when network fees are truly enough to compensate stakers, then it will be possible. However, it’s a completely unproven theory right now that relies on high price and high network usage. There’s no way Bitcoin would survive with a 21mn cap right now.

The other issue with deflation is it prevents people from wanting to spend it. Targeting “reasonable” issuance rates that guarantee safety is the better way to go IMO.


#25

Is there a specific reason that the current “sliding scale” mechanism was selected? Sure, it may be the best option but I’m just curious. Taking previous notes from this thread and just thinking about it, there seems to be 3 possible (and likely more) options when it comes to staking incentive structures.

  1. Sliding scale for issuance based on total stake

  2. Set hard % and all stakers split. Total ETH created reduces each year

  3. Set actual ETH issuance a year for split. Issuance % reduces each year

I believe @MihailoBjelic mentioned 2 at a point. It’s perhaps a simpler and easy to understand solution. The issue is, you really have to get that number right.


#26

I’m curious as well as to why the sliding scale method has been adopted. It seems to me that getting the sliding scale for rewards correct would be harder than getting a flat number correct. You basically have to decide ahead of time what the right reward is at all parts of the continuum, not just decide one flat amount and let the market dictate the number of stakers. I don’t mean to say that it’s an incorrect approach, but I don’t see why setting a scale for rewards is any easier than finding the flat number for inflation.


#27

Yes, I was talking about this option, but I don’t understand the “Total ETH created reduces each year” part? :thinking: Total ETH created actually slightly increases each year, because we have the same inflation % but the base is slightly bigger each year?

I fully agree, would love to hear other people’s thoughts on this.


#28

Er, yes I have them flipped. Fixed % means total ETH issued goes up every year, Fixed ETH means issuance % goes down each year.


#29

Oh, ok, then we definitely understood each other well. :+1:


#30

What if someone sitting on a large amount of ETH (Joe, Vitalik, etc?) comes along and just stakes it all and immediately gets the majority of the interest. Do we care?


#31

But then which flat number? You have three choices:

  • A target total deposit size (if deposits go below, interest rates keep increasing until deposits climb back up, if deposits go above, interest rates keep decreasing until deposits go back down)
  • The interest rate
  • The total issuance

The first option is too unstable, as it offers no guarantees at all about issuance. The second and third both seem reasonable; the third does better at giving users a predictable deposit size (as there’s a feedback mechanism where interest rates go up somewhat if total deposit size goes down) and a predictable issuance rate, but the second does better at mitigating selfish mining attacks. The inverse square root rule is basically the mathematical halfway compromise between the second and the third.


#32

We do, of course. That’s why there’s a fixed deposit size (32ETH) per validator. If evil Vitalik would want to stake 32M ETH and grab the majority of rewards, he would have to run 1M separate validators (incredibly huge amount of resources and setup and maintenance overhead).

@vbuterin, I believe @haokaiwu’s point was: “whichever the number is, it’s easier to get that one number right than to get five different numbers right”. Do you agree on this? I do.

The third option looks the most simple and the most appealing to me, but could you maybe elaborate on how would the second option work, and how it prevents selfish mining? Also, maybe some details/resources on the inverse square root rule, didn’t hear of it yet? Thanks! :slight_smile: :pray:


#33

The problem with selfish mining is this. If a validator (read: 51% coalition) can act in a way that reduces everyone’s returns, but reduces their own returns less, then they can encourage other validators to exit, and then get a larger share of the pie. With the second approach, each validator’s share of the pie does not increase when other validators exit. The inverse square root rule is a compromise, cutting potential gains from selfish mining strategies by 50%.


#34

Please do not underestimate the ability of people to automate things. Cloning a preconfigured VM instance and changing a few parameters means a server can literally be spun up in a minute using automated systems (ie: Terraform). I know this because I’m already staking another coin. I’ve also heard
Vitalik say that he expects the requirements for hardware to be really low so that anyone can do it, which means any VPS. I could even run many validators on a single host (docker) to lower the costs of overhead.

We also do not need to stake 32m ETH… we only need to stake a few million, right away, in order to get the largest percentage of interest.


#35

On the top of reward / issuance rate policies, I recently updated my writeup on discouragement attacks: https://github.com/ethereum/research/raw/master/papers/other_casper/discouragement.pdf


#36

Sorry, wrong url.


#37

Absolutely, but would you agree that obtaining a large share in the validator pool is an order of magnitude harder/more challenging than in PoW?

To elaborate, in PoW it’s theoretically easily possible for a single entity/miner to have 51% or more of the stake/hash power, there’s no any protocol-level defense against that. To do that in Eth PoS, this entity would need to run tens of thousands of validator instances, make sure they’re all online and protected against DDoS and all other attacks, manage keys for each one of them etc.

Clear, thanks, I’ll have this in mind in future…


#38

In PoW, I have to buy hardware and set it up. That is a pain. I’ve done it, I know.

In PoS, this is all trivial on virtual private servers using automated systems to implement it. In a couple pages of terraform config and a few small shell scripts, it can be done. Even across providers and availability zones. I’ve also done this too, and it is super easy. It is DevOps 101. There is no key management. Wallets can be pre-generated and copied onto the machine during the configuration phase. SSL is automated with LetsEncrypt. Seriously, thousands of instances can be spun up, in parallel, in minutes.

As I said before, Vitalik is even wanting to laptops and phones for this, so the hardware requirements won’t be that much.

You’re also missing the point. All I need to do is be first, in order to get the largest reward percentage. If you’re counting on something being hard to setup for the security of the network, you’re going to be really surprised when someone comes along and destroys that assumption.

You can bet someone somewhere is sitting on millions of ether that would just LOVE to get a large percentage return on it. Vitalik has even been encouraging people to get ready for it on the testnets, so there is no reason why they won’t be ready as well.


#39

Here you are referring to the calculation of the reward quotient?

One thing I haven’t been able to figure out yet is how the “base reward quotient” of 2048 is determined? It obviously swing the whole calculation.


#40

A bit confused by this comment. Being first doesn’t matter. What matters is the current network stake at any given time. If you’re the first staker and are getting 20% return, it won’t last as people will come on right behind you, pushing the rate down for everyone.

At a point, large ETH holders won’t find the risk/reward worth it if they push the rate down too far. Would you want to stake and maintain a huge set of validators for 1-2% interest? Very unlikely.


#41

Ok, that is my confusion then and thanks for clarifying it. I was thinking that the first million ETH will get 12.03%. Then the next 1.5m get 7.61%, etc… indefinitely. This plan works quite well to incentivize people to not stake large amounts then… but at the same time, does it incentivize people enough to stake at all? I guess time will tell.