The problem with transaction fees

[ptd]

Consider a world where mining only pays out transaction fees. Almost all miners charge a transaction fee of t and all users pay a transaction fee of t on their transactions.

One miner decides to increase his profits by charging a smaller transaction fee. He publicly announces his intention and many users reduce their fee to be on par with the new fee he has announced. The miner increases his profits (because he processes more transactions) and the users save money. All other miners are forced to reduce their fees as well in order to make any transaction fees at all.

This process continually repeats itself until transaction fees are reduced to their smallest possible value and very little mining is done. This leaves the blockchain venerable. One solution is continue to pay out a block reward forever. What do people think?

[1713918983]

1713918983

[1713918983]

1713918983

[gigabytecoin]

This process continually repeats itself until transaction fees are reduced to their smallest possible value and very little mining is done. This leaves the blockchain venerable. One solution is continue to pay out a block reward forever. What do people think?

Yes, but this is not a problem. On the contrary, it is the magic of a truly free market at work. Thus benefiting the entire world with the lowest possible transaction fees. You should read "the wealth of nations" by Adam Smith to really understand what I just said.

One solution is continue to pay out a block reward forever. What do people think?

That is not a viable solution. It would completely ruin what Bitcoin currently is.

Keep reading the forums, there are lots of great posts about those topics in here!

[casascius]

Consider a world where mining only pays out transaction fees. Almost all miners charge a transaction fee of t and all users pay a transaction fee of t on their transactions.

One miner decides to increase his profits by charging a smaller transaction fee. He publicly announces his intention and many users reduce their fee to be on par with the new fee he has announced. The miner increases his profits (because he processes more transactions) and the users save money. All other miners are forced to reduce their fees as well in order to make any transaction fees at all.

This process continually repeats itself until transaction fees are reduced to their smallest possible value and very little mining is done. This leaves the blockchain venerable. One solution is continue to pay out a block reward forever. What do people think?

I agree that this is a problem.  This is a negative feedback loop.  I believe that mining for transaction fees is a non-equilibrium unsustainable solution that will lead to most people giving up mining because it's not worthwhile.  This is not "the free market at work", it's an example of "tragedy of the commons".

[BitLex]

I believe that mining for transaction fees is a non-equilibrium unsustainable solution that will lead to most people giving up mining because it's not worthwhile.

if most people give up mining, because it's not worthwhile, less people mine more blocks and get more fees in a shorter timeframe, what again makes it worthwhile to them.
i don't see a tragedy here.

[casascius]

I believe that mining for transaction fees is a non-equilibrium unsustainable solution that will lead to most people giving up mining because it's not worthwhile.

if most people give up mining, because it's not worthwhile, less people mine more blocks and get more fees in a shorter timeframe, what again makes it worthwhile to them.
i don't see a tragedy here.

Other than the strength of the network is depleted because of economic incentives to leave it.

Part of the recipe for making Bitcoin strong is a critical mass of mining.  An attacker must exceed the honest miners in CPU power to take over the network.  By giving most honest miners a reason to stop mining, as suggested by the OP the network is made vulnerable to takeover.

[ryepdx]

By giving most honest miners a reason to stop mining, as suggested by the OP the network is made vulnerable to takeover.

My thoughts exactly. The problem with free market theory, aside from assuming that people are rational actors who can be counted on to do what is in their own best interests, is that it does nothing to prevent crashes of things that can only afford to suffer one crash. I do not doubt that it would take only one hostile takeover of the bitcoin network to leave it in smoldering ashes, especially if that takeover occurs after bitcoins have reached a wide degree of acceptance.

A good way of handling this would be to have an agency (or two: this is the free market, after all) keep tabs on how much computing power needs to be in the network to keep bitcoins secure from various levels of attack. I do think that the majority of people are ethical and interested in doing what is best for them and those they are close to, so making known the state of the network as well as what, if anything, needs to be done to secure it, would probably be a good idea.

[Hal]

Consider a world where mining only pays out transaction fees. Almost all miners charge a transaction fee of t and all users pay a transaction fee of t on their transactions.

One miner decides to increase his profits by charging a smaller transaction fee. He publicly announces his intention and many users reduce their fee to be on par with the new fee he has announced.

If I post a transaction with a fee that would be accepted by only one miner, I will have to wait much longer for my transaction to get into a block. If he controlled 1% of the mining power, I'd have to wait for 100 blocks on average. That may not be attractive to many users.

[casascius]

If I post a transaction with a fee that would be accepted by only one miner, I will have to wait much longer for my transaction to get into a block. If he controlled 1% of the mining power, I'd have to wait for 100 blocks on average. That may not be attractive to many users.

How many miners would forego an additional personal gain for the good of the commons?  Especially when nobody is watching?  99%?  Forever?

I would be willing to bet far more than 1% would accept any transaction that came to them, either because they didn't understand how doing so ruined it for everyone else, or they just plain didn't care.

Expecting the mining community at large to follow a "transaction fee honor code" when nobody is watching is like expecting the populace to never cheat on their taxes when nobody is auditing.  It's just not likely to work for very long.  It's a non-self-sustaining, non-equilibrium scenario.

[caveden]

This problem has been discussed before. I agree it's tricky, and it also worries me a bit. This transaction fee thing does look like a tragedy of the commons.

One solution would be to have a maximum block size limit that automatically adjusts itself. It can be made so that it's always "tight", possibly pushing fees up. There's a thread for this: http://bitcointalk.org/index.php?topic=1865.0

I do worry not even that would be enough, though. I mean, I can easily imagine a scenario where the greatest majority of transactions (>95%) are not done on the block chain, precisely to avoid fees. Most people would use bank-like services for their transactions. Transactions between different banks could be dealt by them, for example, bank A has an account on bank B and vice-versa, and they transfer to these accounts, avoiding to use the block chain. They would only need a block chain transaction once in a while in order to transfer the balance. This could be done daily or even weekly, depending on the level of trust these institutions have on each other.

Anyway... we're discussing a scenario that will take long years to happen. Impossible to know what will be of bitcoin by then.

[ptd]

This process continually repeats itself until transaction fees are reduced to their smallest possible value and very little mining is done. This leaves the blockchain venerable. One solution is continue to pay out a block reward forever. What do people think?

Yes, but this is not a problem. On the contrary, it is the magic of a truly free market at work. Thus benefiting the entire world with the lowest possible transaction fees. You should read "the wealth of nations" by Adam Smith to really understand what I just said.

Indeed, the fact that transaction fees are eliminated by competition is not the problem. It's the fact that difficulty is also eliminated, it makes double spending attacks too easy.

One solution is continue to pay out a block reward forever. What do people think?

That is not a viable solution. It would completely ruin what Bitcoin currently is.

Note that keeping the block reward at 50 forever is not going to be a big problem. 50 btc will become a gradually smaller proportion of the total number of bitcoins until eventually it becomes equal to the rate of loss (which is going to be proportional to the total number of bitcoins in circulation).

Could you explain your thoughts more?

[ffe]

Consider a world where mining only pays out transaction fees. Almost all miners charge a transaction fee of t and all users pay a transaction fee of t on their transactions.

One miner decides to increase his profits by charging a smaller transaction fee. He publicly announces his intention and many users reduce their fee to be on par with the new fee he has announced.

If I post a transaction with a fee that would be accepted by only one miner, I will have to wait much longer for my transaction to get into a block. If he controlled 1% of the mining power, I'd have to wait for 100 blocks on average. That may not be attractive to many users.

I agree. In fact I assume there will be a range of offered fees and a range of miners accepting certain minimum fees resulting in a range of wait times for transactions to post.

[ffe]

This process continually repeats itself until transaction fees are reduced to their smallest possible value and very little mining is done. This leaves the blockchain venerable. One solution is continue to pay out a block reward forever. What do people think?

Yes, but this is not a problem. On the contrary, it is the magic of a truly free market at work. Thus benefiting the entire world with the lowest possible transaction fees. You should read "the wealth of nations" by Adam Smith to really understand what I just said.

Indeed, the fact that transaction fees are eliminated by competition is not the problem. It's the fact that difficulty is also eliminated, it makes double spending attacks too easy.

One solution is continue to pay out a block reward forever. What do people think?

That is not a viable solution. It would completely ruin what Bitcoin currently is.

Note that keeping the block reward at 50 forever is not going to be a big problem. 50 btc will become a gradually smaller proportion of the total number of bitcoins until eventually it becomes equal to the rate of loss (which is going to be proportional to the total number of bitcoins in circulation).

Could you explain your thoughts more?

My two cents worth; A reward of 50 charges everyone equally for the cost of making a new block (by debasing the currency a little). Now this may not "be a big problem" but why is it more fair than charging the creators of new transactions directly for the cost of those blocks in the form of a fee.

Think of the 50 coin reward in the early days as a reward for creating a small fraction of the β21M coins we want created. We all have an interest in making those and we all pay. Once the β21M coins are created I could care less if this guy's transaction goes in fast or slow. Let him pay for the service. I don't want my coins debased to pay for fast service for someone else.

[eMansipater]

You all seem to be forgetting the force that operates in the opposite direction--spam.  If bitcoin is at all successful, the number of free transactions will grow beyond the rate of overall processing and effectively force a minimum fee.  This minimum fee multiplied by the number of transactions in a block provides a hard bottom to mining income.  And no matter what it is, it will always have the vast majority of non-spam transactions being sent at that level.  The high load means that because of random spikes and valleys there is not a guarantee that paying the minimum fee will get your transaction in right away which means that miners within their own processing will always process the highest-fee transactions first.  So the paying-a-fee-to-get-priority doesn't rely on a set proportion of miners.  It works for all miners because who would not try to be the first to process the high fee transaction?

There are several other forces acting in the opposite direction to a tragedy of the commons, such as the fact that the cheaper transactions become the more transactions will be sent.  Or, if the difficulty falls too low it becomes a financial incentive for someone to load a lot of processing power into mining and set a fee for near-immediate processing.  For large enough markets these will all tend to keep difficulty too high to launch a cheap attack against.

As bitcoin gets bigger, though, we can start to be more intelligent about detecting double spend attempts as well.  Double spend attempts are actually quite obvious if you are looking for them, because they have to contain two separate, legit, signed transactions; they're only conceivably profitable for very large transactions; and unless the receiver is a fool the attacker has to "come from behind" after multiple blocks have already been generated.  This produces an incredibly distinct signature which could be used to increase the cost of a double-spend beyond a temporary 51% surge.

How and why?  Well, there is a financial incentive for a large mining operation to fight double-spend attacks because successful double-spends wipe out already mined blocks and thus cut into their profits.  So, a mining operation can temporarily bring in a "surge" of computing power at a slightly higher cost to fight off a double-spend attempt--merely having the power available means that it shouldn't have to be typically used so this isn't a huge overall cost compared to the security it provides the mining operation, and the surge still earns them blocks, just at a slightly narrower profit margin.

There are also other techniques that might be worthwhile in a simpler sense, because if double-spend attacks are feasible the value of bitcoins is impacted.  So, a miner which doesn't have access to extra computing power for the surge tactic can change its behaviour when detecting a double-spend attempt for this reason alone.  Rather than switching to the second chain as soon as it takes over, the miner could enact a "wait and see" strategy to prolong the length of time the double-spender must overpower the network and thus increase the cost of attack.  There would be particular incentive to do this if the miner had found a block in the chain that was being overtaken by the attacker.

tl;dr:Bitcoin has lots of room for expansion and toughening without changing the rules of the protocol.  The questions being raised are worth answering, but they don't present a significant threat to bitcoin itself, which is worth making a point of.

[kiba]

Some of your writing here could be turned into a worthy article for The Bitcoin Weekly. 

[eMansipater]

Some of your writing here could be turned into a worthy article for The Bitcoin Weekly. 

Just give me a topic .  I've done some professional writing before.

[mndrix]

This process continually repeats itself until transaction fees are reduced to their smallest possible value and very little mining is done.

Slight adjustment: Fees will be reduced to their smallest possible value and only very efficient mining is done.  Inefficient miners depart, efficient ones grow and the network is still strong.

Of course, block rewards for mining will continue for many years.  During that time, Bitcoin prices will remain the dominant factor in mining profitability.  It will be interesting to see how much prices move upward once the block reward drops to 25 coins.

[deadlizard]

so the fees fall and the proffesional miners leave then the difficulty falls until CPU mining is viable again and every client is able to collect fees and the network becomes P2P again until the proffesionals realise they can make a quick buck by rowing the difficulty. entering during low difficulty and leaving when the difficulty rises.

[FreeMoney]

I think there is a problem, but it hasn't been precisely identified.

Think about what is valuable that you get for free. Obviously for now transactions are being paid for by new coins. In the future a fee will be paid, but this fee you pay is for your first confirmation only. The second, third, fourth confirmations are valuable, but free. This is odd. Another valuable-but-free thing is having other miners build off of blocks that you have found.

I am pretty sure that in the future miners will 'forward' part of their fees to anyone who builds off of their blocks. This will also have the effect of spreading received fees over the multiple confirmations that a transactor wishes to 'buy'.

If there is a clean way to do this (and I think there is) then it will take care of this "commons" problem.

[Syke]

Consider a world where mining only pays out transaction fees. Almost all miners charge a transaction fee of t and all users pay a transaction fee of t on their transactions.

Stop right there. First, miners don't charge fees. They either include txs or they don't.

One miner decides to increase his profits by charging a smaller transaction fee.

For this to even be possible, there would have to be so many txs that they don't all fit in a block. That would be thousands of txs every 10 minutes. That is in the very distant future. And again, miners don't charge fees. If there are so many txs that they don't all fit in a block, they'll just pick the txs with the largest fees. This "smaller transaction fee" situation makes no sense.

[ptd]

Consider a world where mining only pays out transaction fees. Almost all miners charge a transaction fee of t and all users pay a transaction fee of t on their transactions.

Stop right there. First, miners don't charge fees. They either include txs or they don't.

One miner decides to increase his profits by charging a smaller transaction fee.

For this to even be possible, there would have to be so many txs that they don't all fit in a block. That would be thousands of txs every 10 minutes. That is in the very distant future. And again, miners don't charge fees. If there are so many txs that they don't all fit in a block, they'll just pick the txs with the largest fees. This "smaller transaction fee" situation makes no sense.

Suppose I run a miner that only includes transactions that contain at least a 0.01 btc fee, regardless of what is already in the block. That is "charging a fee".