Bitcoin enhancement proposal

[zatoichi]

I am unable to find much work regarding what happens after mining ends. A hundred years from now does seem like a long time, but to ignore it is irresponsible. Assuming the community consensus is to shun regulation, here is what I imagine will happen. Everyone will flock to the largest transaction processor since they will likely offer the lowest rates. As their market share grows they will bankrupt all the other processors by offering even lower rates. When the smoke clears they will look around for any other competition like VISA, if they still exist. If VISA has a rate of 2%, they will offer 1.9%. They will be so large that the cost of entry for competition will be prohibitive. This would seem to defeat the entire purpose of the Bitcoin spirit.

The solution: In processing transactions there must be an incentive to maintain an optimally sized network for competition and security of the blockchain. It must be built-in.

[DannyHamilton]

I am unable to find much work regarding what happens after mining ends.

It depends on what you mean whan you say "mining".  The process of adding transactions to blocks, solving for a hash of appropriate difficulty, and claiming the block reward will never end so long as bitcoin continues to exist.

What will eventually end (somewhere around the year 2140), is the block subsidy.  The block reward consists of the sum of the block subsidy and the total of all transaction fees voluntarily paid by all transactions that the miner chooses to include in the block that they solve.

Over time, the block subsidy (currently 25 BTC) is reduced.  It is cut in half every 4 years.  This means that if bitcoin continues to increase in usage in the future, the portion of the block reward that comes from the transaction fees will likely increase while the portion from the subsidy will decrease.  Eventually more of the reward will come from fees than subsidy (though a significant portion will still come from subsidy).  The percentage that is derived from subsidy will taper off until eventually the entire reward is coming from the fees.

Everyone will flock to the largest transaction processor since they will likely offer the lowest rates. As their market share grows they will bankrupt all the other processors by offering even lower rates. When the smoke clears they will look around for any other competition like VISA, if they still exist. If VISA has a rate of 2%, they will offer 1.9%. They will be so large that the cost of entry for competition will be prohibitive.

When you create a transaction, you don't choose the miner (or mining pool) that will confirm your transaction.  You simply broadcast it to all your connected peers.  Miners choose which transactions they want to confirm.  Any transaction that is not confirmed by the "largest transaction processor" ( I assume you mean "largest mining pool"?) will be available to be confirmed by smaller mining pools.  While the smaller mining pools may not solve as many blocks, they won't have to split the revenue among as many mining units and therefore will be able to pay a larger portion of the block reward to each participant.

The solution: In processing transactions there must be an incentive to maintain an optimally sized network for competition and security of the blockchain. It must be built-in.

It is.  It's called the block reward (coupled with an automatically adjusting "difficulty").

[zatoichi]

I'm not sure I understand the economics here. I'm concerned about the final stage which is fee-based. I'm not sure how this fee would get set. Would the fee structure mimic the profitabity of mining for bitcoins? If not, would that drive out a lot of the network perhaps to the detriment of security?

[DannyHamilton]

I'm not sure I understand the economics here. I'm concerned about the final stage which is fee-based. I'm not sure how this fee would get set. Would the fee structure mimic the profitabity of mining for bitcoins? If not, would that drive out a lot of the network perhaps to the detriment of security?

There is limited space in the blockchain.  As such, transaction creators that want their transactions confirmed in the next block will voluntarily offer a higher than average fee with their transaction to provide an incentive for miners to include the transaction in the next block.  Those who want their transaction confirmed in the next few blocks will offer an average fee to provide an incentive for miners to include the transaction before including those transactions that offer a less than average fee.  The transactions with a below average fee will have to waif until the transaction volume is low and there is spare space available in the block.  If the fee offered is less than the cost of increased orphan risk, then most miners will never include the transaction in the block (providing a floor to the acceptable transaction fee if the sender, or receiver, ever wants the transaction to be confirmed).

Bitcoin is an experiment to see if the interaction of these economic forces are sufficient to provide incentive for a secure system.

There are several economic forces all influencing each other:

• Diminishing block subsidy
• Voluntary transaction fees
• Limited block size
• Target difficulty
• Limited total bitcoin supply
• Initial inflationary circulating supply
• Eventual deflationary circulating supply
• Random distribution of "winning" blocks weighted by hashing power
• Perhaps some forces I haven't thought to include?

Only time will tell if the experiment is a success or not.  Either way, it will be interesting to watch.

[msc]

I'm not sure I understand the economics here. I'm concerned about the final stage which is fee-based. I'm not sure how this fee would get set. Would the fee structure mimic the profitabity of mining for bitcoins? If not, would that drive out a lot of the network perhaps to the detriment of security?

Miners can collectively demand a fee as high as people are willing to pay.  And mining will become cheaper with future technology, and will also become cheaper if a lot of people stop mining.  There's not much we can do now, though.  The block reward can't continue forever, so the only reward will be the fees.

[DannyHamilton]

The block reward can't continue forever, so the only reward will be the fees.

The block reward (sum of subsidy and all fees) can continue forever, the block subsidy will diminish to 0 between now and block number 6 930 000.

[msc]

The block reward (sum of subsidy and all fees) can continue forever, the block subsidy will diminish to 0 between now and block number 6 930 000.

Right, that's what I meant. 

[DannyHamilton]

The block reward (sum of subsidy and all fees) can continue forever, the block subsidy will diminish to 0 between now and block number 6 930 000.

Right, that's what I meant. 

 

I figured, but just wanted to clarify for anyone who comes across the discussion and might have accidentally thought that you meant that the miners would not receive any reward after 2140.

[zatoichi]

You will have to forgive me but Bitcoin is no longer an experiment. There are probably little old ladies in China who have put their life savings in Bitcoins. If everybody thought Bitcoin is just an experiment the price would probably be back to $12 or under in a heartbeat. If it's just an experiment, this needs to be better advertised.

One theory of transaction cost:

A days transactions must pay for a day's network cost. Of course, after the last bitcoin is mined the network will be monsterously huge as everyone tried to get that last incredibly valuable Bitcoin. No one will be able to afford that fee. A fee advertisement mechanism will have to be used to reduce network size. Then we are back to my original problem of monopolization.

[DannyHamilton]

You will have to forgive me but Bitcoin is no longer an experiment.

Yes.  It is.

There are probably little old ladies in China who have put their life savings in Bitcoins.

Then they have put their life savings into an experiment that is still considered "beta".  This is a risky decision, but it is their decision to make.  Hopefully the gamble will pay off for them.

If everybody thought Bitcoin is just an experiment the price would probably be back to $12 or under in a heartbeat. If it's just an experiment, this needs to be better advertised.

The software says "Version V0.8.5-beta" as it starts up.

Choose "About Bitcoin-Qt" from the menu in the software and it specifically states:
"This is experimental software."

On the bitcoin.org website under the "You Need to Know" page:
http://bitcoin.org/en/you-need-to-know

It states:
"Bitcoin is still experimental
Bitcoin is an experimental new currency that is in active development. Although it becomes less experimental as usage grows, you should keep in mind that Bitcoin is a new invention that is exploring ideas that have never been attempted before. As such, its future cannot be predicted by anyone."

The fact that some people choose to use something without educating themselves about their decision does not change the nature of that decision.

One theory of transaction cost:

A days transactions must pay for a day's network cost. Of course, after the last bitcoin is mined the network will be monsterously huge as everyone tried to get that last incredibly valuable Bitcoin. No one will be able to afford that fee. A fee advertisement mechanism will have to be used to reduce network size. Then we are back to my original problem of monopolization.

I don't understand what you are saying?

What do you mean by "everyone tried to get that last incredibly valuable Bitcoin"  Bitcoins are fungible.  Why would any one bitcoin be any more valuable than any other?

[zatoichi]

If you look at the graph of bitcoin prices (until the China Connection) it seems to be going up and perhaps must since the difficulty increases over time, generally, theoretically. Or maybe it will hit a wall where the next bitcoin is not worth mining. Barring a bitcoin failure, it would not be surprising that the last bitcoin mined would be the most valuable (current market value). I cannot imagine an incentive for the network not to continually increase in size.

[DannyHamilton]

If you look at the graph of bitcoin prices (until the China Connection) it seems to be going up and perhaps must since the difficulty increases over time, generally, theoretically.

There is nothing that requires difficulty to increase. If difficulty gets too high, the miners with the highest costs will stop mining to avoid losing money.  The blocks will then occur less frequently and the difficulty will decrease.  It has hapened in the past, it is likely to happen again in the future.

Or maybe it will hit a wall where the next bitcoin is not worth mining. Barring a bitcoin failure, it would not be surprising that the last bitcoin mined would be the most valuable (current market value).

Perhaps you don't understand the term "fungible".  While a miner who happens to receive the last 0.00000001 BTC subsidy, might choose to avoid using that particular output as an input unless offered something of significantly higher value in exchange, this is a matter of collectible value (such as artwork) and not a matter of useful value.  Regardless, the miner that receives this final subsidy won't have any incentive to require higher fees for the transactions they include in that block.  If they choose not to include any transactions in the block, then the miner that solves the next block (the one without any subsidy) will still receive the transaction fees from the transactions that they include.  As such, there will still be an incentive to mine, and there will still be an incentive for transaction creators to include a market determined fee.

[zatoichi]

A market determined fee? A market implies competition which implies advertisement of competing products or prices. How is the selection done? If it's an auction for the lowest fee then the problem of monopolization remains.

Or if you mean the transaction originator (spender) includes a fee, is there a guarantee that this will find a processor at less than competing services (e.g. VISA at 2%)

[DannyHamilton]

A market determined fee? A market implies competition which implies advertisement of competing products or prices. How is the selection done? If it's an auction for the lowest fee then the problem of monopolization remains.

It's a public auction for the highest fee.

Everyone voluntarily chooses what fee they want to pay for their transaction.

There is a financial incentive for miners to choose the transactions that offer the highest fees per byte (since they get to keep that fee).

There is a temporal incentive for transaction creators to offer a competitive fee per byte (since having their transaction chosen by a miner means faster confirmation).

There is a financial incentive for miners to fill all the bytes of a block if there are enough fee paying transactions available to do so (since each transaction increases their revenue).

All the transactions that are included in a block are publicly listed on every copy of the blockchain.  This allows everyone to see how many transactions were included in the block and what the total transaction fees were.

Simple division allows the user to determine what the average fee per byte was in the most recent blocks.

A reasonable assumption is that a fee higher than this average will increase the likelihood of your transaction being included in the next block, and that a fee less than this average will increase the likelihood that you will have to wait many blocks before your transaction will be confirmed.

The maximum blocksize is the competing "product" that is advertised and well established.  The transaction fees are the "prices" that are publicly visible and offered by the competitors desiring space in the next block.

[zatoichi]

Still the fees must cover the cost of running the network or no one will process transactions and be less that competing services (VISA?). Is there an optimum network size that guarantees security? Obviously a network of one is the cheapest but guarantees no security.

[DannyHamilton]

Still the fees must cover the cost of running the network or no one will process transactions and be less that competing services (VISA?). Is there an optimum network size that guarantees security?

I suppose that is what this experiment is intended to discover.

Obviously a network of one is the cheapest but guarantees no security.

Clearly.

The incentives have been created by the protocol.

The experiment will play out, and the free market will determine how big of a network it is willing to pay for.  Over time, mining will likely migrate to the locations that are most favorable for profitability.  Since the revenue per hash is fixed for any given network size, this means that profitability is increased by the reduction of mining costs.  I think most of the ongoing costs of mining are electricity, cooling, and internet access.  Therefore, it would seem that the mining operations that manage to minimize those costs are most likely to eventually push out the operations that have higher costs associated with those assets.  Perhaps mining will eventually move to the poles (where cooling costs are minimized)?  Or areas with cheap access to renewable energy sources (such as hydro-electric or geothermal)?

[zatoichi]

Well I have to stand by my theory that unfettered market forces favor monopolization and without a control this is Bitcoins future. Perhaps what is required is something like the reverse of difficulty where the fee goes to zero if the network goes to one on some kind of sliding scale with maximum return at a level of ideal security. Easy to implement.

[DannyHamilton]

Well I have to stand by my theory that unfettered market forces favor monopolization and without a control this is Bitcoins future.

You are welcome to your opinion.  At this time, the market seems to be favoring the success of bitcoin, but only time will tell if the current market beliefs are correct or if you are.

Perhaps what is required is something like the reverse of difficulty where the fee goes to zero if the network goes to one on some kind of sliding scale with maximum return at a level of ideal security. Easy to implement.

Easy to say, doesn't mean easy to implement.  If it is so much better, and so easy to implement, feel free to create your own crypto-currency based on your beliefs.  If yours is truly better, then it should win out in the long run.

I suspect that you'll discover that such a system isn't as easy to implement as you seem to believe it is, and that it very well may not be better.

[zatoichi]

I am actually surprised that you would not be behind an enhancement for increased security. As for implementation all you would need would be a single field in the header showing how much of the fee can be recovered based of the network size at time of creation (e.g. network size=1, recovery ratio=0). Some work might have to be done to determine optimal network size and the exact nature of the curve. I have asked in multiple places how big the current network is but got not a single answer. Of course it changes but not even a ballpark figure? I mean come on, count the nodes in the blockchain. Unfortunately I am not that technically proficient.

[DannyHamilton]

I am actually surprised that you would not be behind an enhancement for increased security.

Oh, I'm certainly interested in increased security, I'm just not convinced that your suggestion could be implemented, nor that it would increase security.

As for implementation all you would need would be a single field in the header showing how much of the fee can be recovered based of the network size at time of creation (e.g. network size=1, recovery ratio=0).

And in a peer-to-peer system, how would you determine "network size"?  How would you know when the transaction was created? What would you do with the unrecovered portion of the fee? There are so many details about how the protocol works that you are choosing to simply overlook.  Overlooking details is a great way to break a system.

Some work might have to be done to determine optimal network size and the exact nature of the curve. I have asked in multiple places how big the current network is but got not a single answer.

That's because it is a peer-to-peer decentralized system.  There is no reliable way to know what the "network size" is, and there is no reliable way to know if the netowork size will be smaller or larger a few seconds from now.

Of course it changes but not even a ballpark figure?

Impossible to know.  Perhaps there is just a single miner doing all the mining?  Perhaps there are a few trillion individuals all mining with CPU power?  How would any one node tell the difference?  The only thing any node knows about the network is the number of nodes it is connected to, and it can't even trust that since a single node could connect from multiple IP addresses.

I mean come on, count the nodes in the blockchain. Unfortunately I am not that technically proficient.

There are not any nodes "in the blockchain".  The blockchain is just a history of transactions and a headers.

[Cryddit]

I think I agree that the way Bitcoin is structured it must eventually become centralized, and that centralization will be the end of the era of bitcoin as we knew it.  Once centralized, it will inevitably be controlled by governments.  Once controlled by governments, it will inevitably become just a payment network for fiat currency.  If its structure did not favor centralization, we would not have mining pools in the first place.  

That is sad, but it's not going to happen immediately.  In order to prevent it we would need a fundamentally different reward structure and means of securing the block chain.  But we're not going to get that in Bitcoin.

Also, we shan't have to wait 127 years for the end of mining.  You said it yourself; the reward is halved every four years.  That means that it goes from 25 to 12.5 to 6.25 to 3.125 to 1.6  in just 5 halvings, or 20 years.  In another 5 halvings, forty years from now, it'll be less than 0.025 BTC, which is close enough to zero that it doesn't really matter after that.  Mining, insofar as it continues to be relevant given the threat of centralization and takeover by governments, will sink or swim on fee support in our lifetimes, not those of our descendants.

[zatoichi]

I realize that the suggestion is in its infancy but I am hoping for people to see that perhaps it is Bitcoin that has failed in the details. I wonder if the inventor pushed the final date way into the future so he wouldn't have to rip his eyes out when it crashes. I mean why not 20 years? Of course it may crash way before 2140.

Doesn't the blockchain show where the new bitcoins are awarded? Isn't the node a party to that transaction? If so, you should be able to get a minimum count (just remove duplicates).

[DannyHamilton]

Also, we shan't have to wait 127 years for the end of mining.  You said it yourself; the reward is halved every four years.  That means that it goes from 25 to 12.5 to 6.25 to 3.125 to 1.6  in just 5 halvings, or 20 years.  In another 5 halvings, forty years from now, it'll be less than 0.025 BTC, which is close enough to zero that it doesn't really matter after that.

You are mistaken.

Block subsidy is issued in integer units (no decimals).

The original subsidy was 5 000 000 000 integer units (commonly referred to by the nickname "satoshi").

The current subsidy is 2 500 000 000 integer units.

In 5 halvings the subsidy will be 160 000 000 integer units.

In an additional 5 halvings (forty years from now), the subsidy will be 5 000 000 integer units.  That is significantly more than 0 (by 6 orders of magnitude).

Whether those 25 mBTC will matter will depend largely on what the exchange rate of bitcoins is at the time.

If bitcoin exchange rate is $10 000 per mBTC, then 25 mBTC will be $250 000 per block.

[DannyHamilton]

Doesn't the blockchain show where the new bitcoins are awarded?

No.  The only information the blockchain has about where the bitcoins are awarded is the transaction outputs of the coinbase transaction.  These outputs simply encumber some value(s) with a requirement of a digital signature from a particular private key (or set of private keys) to be reassigned elsewhere.  There is no "node" information at all.

Isn't the node a party to that transaction?

There are never any "nodes" that are party to any transactions.  Transactions are a list of unspent outputs (as inputs to the transaction), a set of signatures providing proof that the transaction is authorized to spend those previously unspent outputs, and a list of new unspent outputs encumbered with a condition that must be met to spend those new unspent outputs (typically a signature from a private key that is associated with a bitcoin address).  This is entirely "nodeless".  Transactions are relayed from node to node with no knowledge of where they originated.  The only difference with the coinbase transaction is that there is no list of previously unspent outputs or signatures, only new unspent outputs encumbered with the condition that must be met to pend them.

If so, you should be able to get a minimum count (just remove duplicates).

Understand yet why there is no way to know how many nodes (or miners, or mining pools) exist?

[zatoichi]

I recently realized that knowing the network size at all times is not necessary. What is necessary to know at all times is whether the count is greater than the "ideal". The exact number does not matter until it falls below the ideal. Obviously the larger the network the more secure it is but the ideal would be the lowest number that constitutes a specific threat level considered acceptable. What would be nice about this would be that the largest processors would have an incentive to keep their rates high enough so competition exists. Their penalty for reducing prices below competitive rates is an decrease in the recovery ratio.

Where are Bohm and Jacopini when we need them? They figured out the minimum number of programming constructs that could satisfy any algorithm. They came up with three. I'm sure the number of nodes for adequate security would be much larger but it might also be easily confirmed as existing at any one time (a simple ACK?)