Demurrage, transaction fees, storage fees & comparison to commodity money.

[MoonShadow]

This might be better off in a different section, but I'm sure that it's going to go all over the place.

Recently, there has been much discussion in many different threads about how the transaction fees as they exist are a "tragedy of the commons".  Even though they are not such a type of resource commons, those threads seem to have awakened a real future concern about whether or not transaction fees will be enough of a reward for miners once the block reward drops to trivial levels.  The part that really got me wondering is about the lack of "demurrage" in Bitcoin's algorithum.  Many will immediately attack me here as advocating for inflation, but let me say first off that I am not advocating for any change that would actually make Bitcoin inflationary.  However, for those who do not know what demurrage is, it is the intentional introduction of fees into an artificial monetary exchange system intended to replicate the 'storage costs' that a commodity money would have.  For example, when physical gold was the primary medium of exchange between nation-states and silver the primary medium of exchange between private entities; both commodities imposed relatively high transaction fees in the form of transportation costs.  This kind of fee structure Bitcoin replicates well with it's transaction fee schedule.  However commodity money also imposed storage fees upon the long term holder of these mediums of exchange, usually in the form of the costs of building a vault or renting a safety deposit box inside the much larger vault of a bank or other highly secure institution.  It's these storage fees that demurrage is intended to replicate, and for which Bitcoin does not have a corrolary.  I don't know if it could even be done, and still keep everything else straight, but I want to ask for ideas about how demurrage could be introduced into Bitcoin under the following ideal conditions...

1)  Miners should receive the demurrage fees relative to their hashing contributions, as transaction fees and the block reward are now.

2)  Only transactions older than, say 6 months, should be affected by demurrage fees

3)  Demurrage fees should be very small, but should be assessed with each block, or perhaps with each retarget block.

4)  Demurrage fees should favor transactions with large collective amounts, probably the best way to do this is to asses the demurrage fee by each elderly transaction and not by how much it contains.  (a safety deposit box costs the renter the same if there is 100 ounces of gold in it or just one)


The astute observer will immediately notice that this will not likely result in actual demurage fees, but instead in savers periodicly moving their funds to keep them fresh.  This is part of the point, as if it's economicly better for the savers to freshen their funds and potentially pay a transaction fee instead, they still contribute to the security of the blockchain with the side benefit that their old transctions can be pruned from the blockchain once that becomes possible as well as encourage the condensation of many small transaction balances into fewer and smaller transactions.

I'm sure that an exception can be added to the demurrage fee system for the genesis block, so that Satoshi can keep his legacy intact for this heirs.

Any thoughts on how this could be accomplished, or why it shouldn't?  I'm open to being proven wrong about this concern.

[FreeMoney]

I'm sure that an exception can be added to the demurrage fee system for the genesis block, so that Satoshi can keep his legacy intact for this heirs.
 

Fuck that. Somehow you realize it's shitty to do to him, but not the rest of us?

[MoonShadow]

I'm sure that an exception can be added to the demurrage fee system for the genesis block, so that Satoshi can keep his legacy intact for this heirs.
 

Fuck that. Somehow you realize it's shitty to do to him, but not the rest of us?

I don't think that it's "shitty", just that the genesis block is unique. 

[FreeMoney]

I'm sure that an exception can be added to the demurrage fee system for the genesis block, so that Satoshi can keep his legacy intact for this heirs.
 

Fuck that. Somehow you realize it's shitty to do to him, but not the rest of us?

I don't think that it's "shitty", just that the genesis block is unique. 

whatever, all my coins are unique and i'm not running anything that takes them for being old.

[DATA COMMANDER]

This is just a tax on savings. The average early adopter is going to be strongly against this as well they should be). You are suggesting that we impose an artificial storage cost in order to make bitcoins more like a physical commodity as if that were a worthwhile goal. We don't want "digital gold"--we want a good medium of exchange. Why on earth would holders of BTC want to pay to store them when they cost (virtually) nothing to store?

[FreeMoney]

Okay, sorry for the hostility.

There is a sense in which is already baked in. Miners can 'charge' whatever they want based on whatever they want. It isn't unreasonable that they could require higher payment for tx with dependencies that were way back in the archives, right? Is it actually a tiny bit costlier to look farther back?

Even if there isn't extra costs, miners can still charge for it if they think that is a good idea.

[DATA COMMANDER]

If bitcoins that have been held for a long time actually cost more to spend, then there's no need to create an artificial storage cost. Miners can simply charge higher transaction fees for older coins.

[FreeMoney]

If bitcoins that have been held for a long time actually cost more to spend, then there's no need to create an artificial storage cost. Miners can simply charge higher transaction fees for older coins.

That's what I meant.

[MoonShadow]

This is just a tax on savings. The average early adopter is going to be strongly against this as well they should be). You are suggesting that we impose an artificial storage cost in order to make bitcoins more like a physical commodity as if that were a worthwhile goal. We don't want "digital gold"--we want a good medium of exchange. Why on earth would holders of BTC want to pay to store them when they cost (virtually) nothing to store?

Long term storage of capital is not free for the network, even though it might seem that way.  The network does suffer an uncompensated cost.  Namely, the ongoing replication of those deep transactions as new clients bootstrap and the ongoing disk storage costs, however small those might be individually, multiplied by the number of nodes.  What I'm suggesting is an incentive for capital accumulators to consolidate their holdings into fewer/newer transactions, allowing the network to 'compact' the blockchain.  Currently there is no incentive for early adopters, or anyone else with any substantial holdings, to spend their oldest transactions first.  This is the default action of the client, and this might be enough, but sooner or later someone is going to mod the client to allow users to spend newer coins first, because the deeper the transactions are the more secure they are.  There does need to be a cost for that kind of long term security, particularly if those holdings are spread across numerous transactions that cannot be pruned.  I do like the proposal of a miners' choice and rear-loading the fees.  It permits the well heeled bitcoiner to contribute to the security of the blockchain in a less direct manner, for the only miner that is likely to accept his old transaction without the demurrage fee is one that he owns or is otherwise closely associated with anyway.  If he is owner of a bitcoin bank, his old transactions can be spend or freshened without fee only if his own bank is doing the processing; implying that his financial actions directly benefit the security of the blockchain because that would have to be true for his transaction to be accepted for free in any reasonable period of time.

I have a proposal.  Using the rear-loaded, miners' choice model.  A minimum fee rule for any new transaction with inputs that are older than a year (in blocks) will have an alternative minimum fee based upon demurrage of one Satoshi ( .00000001 BTC) per retarget cycle (2016 blocks) for every input that exceeds one year since it's transaction was recorded.  This means that each input is charged for demurrage from it's inception, not the end of the first year, and the minimum fee for a new transaction with a single transaction exactly a year old would be at least .00000026 BTC.  This isn't much at all, but would still incentivise some savers to either freshen their savings once each year, consolidating their many transactions down to one while doing so, and potentially paying a transaction fee for the effort; or resolve to pay for the network storage costs upon release of funds.  This even gets demurrage upon off-network transactions wherein the private keys are traded instead, because no one saves money to never spend it, so sooner or later that has to happen.

This rule need not go into effect until the block reward is cut to 25 coins, giving the early adopters plenty of time to plan out their best course of action, most of whom will be consolidating numerous 50 BTC transactions into a single (bytewise small but BTC-wise large) transaction; permitting the network to prune even the block reward transactions from the old blocks, perhaps all the way down to the headers alone.

Also, this rule would be an alternative minimum fee, so if some other rule required a higher mimimum fee, those fees would not be additive.  It's just whichever minimum fee is highest that is required; or a miner willing to process your transaction for free.

Alright, I'm ready.  Tell me what you think, but please leave my mother out of it.

[FreeMoney]

I heard your mother charges a minimum fee.

[FreeMoney]

Isn't the minimum fee just a default setting too? Won't people mod around that just as easily as the oldest first rule? Or do you mean to make the default client consider blocks with tx that have less than the minimum fee invalid?

[MoonShadow]

Isn't the minimum fee just a default setting too? Won't people mod around that just as easily as the oldest first rule? Or do you mean to make the default client consider blocks with tx that have less than the minimum fee invalid?

I mean put the rule into the default fee structure for inclusion into the fee paying section of the block.  Any transaction that pays less than the minimum is just a tip and still must wait to be included into the free section, if at all.  Not that a transaction that expects a minimum fee fails a validity check and fails to propagate.  Although that could be a form of enforcement of demurrage.

Yes, the minimum fees can be ignored by users, but it is not in the self interests of miners to accept transactions that should be paying a minimum fee based on published fee schedules agreed to by miners.  Some will always accept the transactions regardless, but they are a charity, contributing to blockchain security without expectation of compensation.

[MoonShadow]

I heard your mother charges a minimum fee.

And your mother said you couldn't afford it.

[DATA COMMANDER]

This entire idea is based on the premise that the market can't price transaction fees "properly" for some reason. What is that reason?

[smooth]

This entire idea is based on the premise that the market can't price transaction fees "properly" for some reason. What is that reason?

The premise is that unless static holders pay something for the service they are receiving from the growing block chain, the resources allocated to maintaining the system will be less than the social optimum.  It's a good idea.

[DATA COMMANDER]

The premise is that unless static holders pay something for the service they are receiving from the growing block chain, the resources allocated to maintaining the system will be less than the social optimum.

Miners are free to charge more to process older coins. Besides, that "premise" is actually an assertion, and a vague one at that. (Does "social optimum" have a concrete definition, or is it just a rhetorical device?)

[Raulo]

You touch only one aspect of this "storage fee" problem: disk space cost.

But there is subtle and much more important problem completely absent in commodity money. The BTC value depends on the safety of Bitcoin transaction system which depends mostly on difficulty. There is no analogy in gold. It's just like the value of gold depended on how well guarded are other's people gold.

A hoarder that nicely keeps all his coins in one place and costs the whole network just a few hundred bytes benefits from large difficulty. But pays nothing to keep this difficulty high enough. Without any mechanism for paying for this protection, the difficulty will be set on a level that is too low to protect this collective wealth.  And since nobody has any motivation to voluntarily pay for this protection (because you cannot pay for protection of just your money, you can only pay for protection of everyone's money) the Nash equilibrium will be such that nobody pays and everybody expect everybody else to pay. And one cannot expect that a few billion worth of BTC (let's be optimistic) will be properly guarded by a difficulty level corresponding to a few million dollar compute system.

[smooth]

Miners are free to charge more to process older coins.

Yes, that was part of the suggestion.

Besides, that "premise" is actually an assertion, and a vague one at that.

Yes, that's what "premise" means.

(Does "social optimum" have a concrete definition, or is it just a rhetorical device?)

Yes, it does.  It is the allocation of resources where the sum is of individual returns is maximized.  Tragedy of the commons is well known (simple) example where the equilibrium is a not a social optimum.

In this particular case, imagine that all transactions are priced equally, but the volume of current transactions slows for whatever reason (say BTC2 replaces BTC in active usage).  Holders of BTC would like to see mining continue so the chain can stay ahead of attackers, and they would pay miners to do it, but there needs to be some mechanism by which they collectively compensate miners, otherwise it won't happen.  

[benjamindees]

Does "social optimum" have a concrete definition, or is it just a rhetorical device?

It means no externalities.  You're not depending on me to do anything for you, and vice-versa.  Raulo hit the bigger issue, though, which is that your storage of Bitcoins depends upon constant energy usage by the network, which isn't free and which isn't covered by transaction fees.

But it should be easy for miners to impose a higher fee on older coins, right?

It would be interesting to come up with the scale of device needed just to store one person's wealth and process that person's transactions throughout her lifetime.  Saying that something like a 10w solar panel, rechargeable battery and $20 FPGA could replace your banker would be an interesting talking point.

[gim]

Won't your proposal have the same effect as a small inflation?