“It basically makes all rewards inside of ethereum. So this would be current Proof of Work (PoW) rewards and it also makes a commitment to include all future rewards – including Casper voting rewards, sharding block rewards, and any thing like that – all be proportional to the maximum supply minus current supply.
That basically means that eth’s supply, instead of just going up linearly, the growth would exponentially decay and the supply would exponentially converge towards the maximum.
At least until we introduce things like rent and partial fee reclaiming. At which point, we’ll have a kind of balance where the supply probably ends up being constant somewhere below the max, and coins getting burned – coins getting created would roughly match each other.”
Daniel Nagy says that for now smart contracts have been pretty diverse in their rate of burning eth when they wanted to. But if you have a limited supply that would mean you’d have a formal way of burning eth to remove it from the formal calculator of current supply.
Buterin said there are many ways of doing that. If it was to be done in a more standardized manner, you could just burn eth by sending it to a zero address. Then rewards could even be paid out of the zero address in proportion to the amount of money in the zero address.
“I think capping supply is a bad idea,” Nick Johnson, an ethereum developer, says. “You can fund security either through inflation or through transaction fees. Funding through transaction fees encourages holding and discourages an active ecosystem.
Which I think is a bad idea in general and can also lead to a deflationary spiral where costs go up because transactions are fewer and that leads to fewer transactions and so forth.”
“It’s really worth noting that transaction costs in ethereum in the long run are not proportional to eth fees,” Buterin said. “They’re proportional to supply and demand.”
“In general, it seems like they go up when eth goes up because when eth goes up demand and adoption goes up, but really, if the price of eth was to go up for some exogenous reason – which in general is likely to happen if a proposal reduces supply – then that increases the price without increasing adoption.
So it is quite likely that will lead to a reduction of the transaction fee people pay. So I don’t see a reason to expect in the long run a hard-link.
Another point regarding the deflationary spiral, from a macroeconomic perspective, that stuff tends to apply much more when, first and foremost, the thing that is deflationary is a unit of account for an entire economy, and, second, when it’s something that’s deflationary where long term sticky prices are being set in.
In the past it was definitely the case that eth gas prices were sticky, but now we have dynamic fee calculations for pretty much everything. So I don’t really expect eth prices themselves to be sticky. Like, eth is hyper-volatile to begin with, so I don’t expect we’ll notice any difference in the price movements as a result of fixed supply.”
“I did not mean deflationary spiral necessarily in the traditional economic sense,” Johnson says. “I mean in the sense that if we assume people are going to adjust their behavior based on how much a transaction is going to cost, and if we assume there’s some number of minimum fees we have to pay to secure the network, then high transaction costs could lead to fewer transactions which leads to higher transaction costs.”
“I’m not necessarily suggesting higher transaction costs,” Buterin says. “The main place in the long run where money would come from in this change is from lower revenue paid to a participant that provides security. So, in the short term miners and in the long term casper validators.
In the long term I do think casper validators revenue should be roughly equal to the amount people pay in transaction fees.”
“There’s no closed loop though there, is there,” Johnson says. “Because presumably there is some level of incentivization we need or some level of staking we need to secure the network and there is nothing that ensures transaction fees match up to that.”
“There’s no minimum, there’s like a level,” Buterin says. “So basically if we have a smaller reward, instead of having 20 million eth staking we might have 10 million eth staking.
On the other hand there is a risk that if we have a cryptocurrency which is inflationary, that could lead to its value dropping. Which by itself leads to less capital securing the network.
It is kind of hard to figure this all out in the long run, but I personally do think there is evidence that transaction fee levels are capable of providing enough revenue to secure a blockchain. And in the long run, if they’re not, then there’s the question of how valuable is the system that we’re building in the first place.”
“Personally, I don’t think transaction fees have to be the thing that supports the blockchain,” Johnson says. “If we need some amount, x eth a day to incentivize miners or stakers, you can take that from inflation or fees.
I think it makes for a more useful system if you take it from inflation because it imposes the cost on everyone who is invested in that system, not just those who are transacting.”
“I used to think this way,” Buterin says. “The problem is, as Vlad keeps pointing out, that if you do that, basically every ERC20 token becomes a better store of value than eth.
If eth becomes this unique token inside of ethereum that has the anti-privilege of being inflated to pay for security expenditure, and you have the ability to just print ERC20s on top of ethereum and market them, and these tokens don’t have this disadvantage, then it may well be the case eventually there is going to be this tragedy of the commons where, even though eth is necessary for network security, no one wants to support eth.”
“That seems a bit of a stretch for 2% a year inflation to me,” Johnson says. Trustnodes note, the standing proposal is for 500,000 eth a year. That initially means 0.5% inflation, but as it is a fixed supply increase, while the total supply keeps cumulatively increasing, that fixed inflation constantly goes down to insignificance and very much practically zero.
“Two percent a year inflation is a lot,” Buterin says. “Think of it this way. In the long run people expect to earn something like 4% a year return on, I believe, from the stock market. If you take 2% off of that, then what you’re basically saying is the amount of money someone needs to retire goes by a factor of 2 because it goes down from 4% to 2%. So in the context of financial markets returns, 2% is a huge deal.”
Johnson and Hudson Jameson wonder what would be required at a technical level for implementation.
“My personal preference for implementation is that we basically just pre-mine max-cap minus current supply into some particular address,” Buterin says. “This could be address zero, and we could just say that whenever we’re paying rewards we subtract the balance form that amount, that account. And that account by itself is used as the current supply factor.”
“That’s roughly three orders of magnitude simpler than what I was thinking, so sounds good,” Johnson says.
“It sounds like there’s actually a good amount of support around this,” Jameson says. “This would require a hardfork, so is this something that could potentially be in Metropolis?”
“I personally see two possibilities,” Buterin says. “If the community really wants it, then in phase two of Metropolis. The other possibility is to just wait a bit and incorporate it as part of Casper FFG change.”
Dealing with process, Jameson says there would need to be an Ethereum Improvement Proposal (EIP) and someone to champion it. Vitalik, are you championing this, Jameson asks.
“I’m kind of waiting on I guess like waiting to see more community feedback,” Buterin concludes.