Miners that refuse to include transactions are becoming a problem

[jgarzik]

There is no economic incentive to including txs.

Incorrect.  Including transactions creates incentives for bitcoin users to use the network (faster transactions).  Slower transactions discourage users from using the network.  This is why deepbit includes as many transactions as it can, and doesn't really worry about fees too much.  Deepbit wants to create incentives for bitcoin use.

Too many no-tx blocks, and users will drift away, causing the value of each bitcoin to drop.

Thus, in the short term, no-tx blocks may make money for the botnetter, but in the long run it will make their (and yours) bitcoins worth less.

The value of bitcoin lies in the strength of the network, and the number of users.

[1714960731]

1714960731

[1714960731]

1714960731

[DeathAndTaxes]

or those users could simply include fees and increase the economic incentive for miners to process them promptly.

If fees were high and the miner was still excluding them I would be inclined to agree.  Still I don't think "mystery" actions are affecting Bitcoin users in any meaningful way.

1 confirm time has increased from 10 minutes to 11.5 minutes but most users look at 6 confirm time.  That has increased from 60 minutes to 61.5 minutes.  A 2.5% slowdown and well within the normal variance of confirmations so that an uninformed users wouldn't even notice tx being "slower".

So I agree there is a tragedy of commons effect but as long as the # of miners excluding all tx remains a minority it is unlikely to have any meaningful effect.

[wogaut]

Personally I actually prefer mining at a pool that includes transactions (even if they don't pay the fees out to miners).

Including transactions is important for the Bitcoin network to be utilized by merchants and users, and that usage is important for the future of Bitcoin.
Right now with 50BTC block reward (or even with 25BTC) transfer fees would need to be immensely high to compete with block rewards. Too high to ensure the use of the BTC network in its early stage. As the number of transfers increase and the block rewards decrease, transfer fees will become more viable. But that requires a high network utilization with a large volume of transactions.

While I don't advocate free transactions (any fee right now however small is better than zero, as it habituates the concept of charging TX fees), including transactions in mining operations is a really easy and cheap way for a miner to show support for Bitcoin. And if you want to see Bitcoin as a future currency that support is certainly worthwhile.

[Brunic]

Some short-sighted miners here it seems...

The real Bitcoin economy would likely not handle a huge increase in transaction fees very well.  Enforcing the minimum would probably be fine.  But doubling or tripling it would likely cause pain.  And even that would get nowhere near the current block reward.

It's not even clear that this miner would include transactions if the fees were higher.  Right now he isn't including the transactions with fees.  He's including zero transactions.  No one knows the reason for this.

Your electric bill is being covered by the block reward.  If you want fees to go up in order to cover this cost, the reward would likely have to come down significantly in order to meet halfway.  Otherwise, the (miniscule) real economy would be hugely affected, and price could plummet.

Be careful what you ask for.

Not right now, I'm not asking for an increase tomorrow. But in around 9 months, miners will have their payday cut by half. I'm not going to do the "good thing for Bitcoin" if I'm stuck paying from my pocket because of low fees. And it will be the same for a bunch of miners there, if they can't profit, they close the rigs. What happen if they close their rigs? Well, solominer is going to have a bigger % of the network and because of the drop of miners, blocks will take a longer time to solve, slowing even more the flow of transactions.

I don't see this situation as a problem, I see this as a wake-up call. It's going to take that long before we see software that manage included transactions in a block, or "optimal" pools that include only transactions with fees.

[payb.tc]

Not right now, I'm not asking for an increase tomorrow. But in around 9 months, miners will have their payday cut by half. I'm not going to do the "good thing for Bitcoin" if I'm stuck paying from my pocket because of low fees. And it will be the same for a bunch of miners there, if they can't profit, they close the rigs. What happen if they close their rigs? Well, solominer is going to have a bigger % of the network and because of the drop of miners, blocks will take a longer time to solve, slowing even more the flow of transactions.

you forgot maybe solo-miner also closes his rigs at this point.

[Brunic]

Not right now, I'm not asking for an increase tomorrow. But in around 9 months, miners will have their payday cut by half. I'm not going to do the "good thing for Bitcoin" if I'm stuck paying from my pocket because of low fees. And it will be the same for a bunch of miners there, if they can't profit, they close the rigs. What happen if they close their rigs? Well, solominer is going to have a bigger % of the network and because of the drop of miners, blocks will take a longer time to solve, slowing even more the flow of transactions.

you forgot maybe solo-miner also closes his rigs at this point.

Yeah, but other people with the same idea could do the same thing.

[payb.tc]

Not right now, I'm not asking for an increase tomorrow. But in around 9 months, miners will have their payday cut by half. I'm not going to do the "good thing for Bitcoin" if I'm stuck paying from my pocket because of low fees. And it will be the same for a bunch of miners there, if they can't profit, they close the rigs. What happen if they close their rigs? Well, solominer is going to have a bigger % of the network and because of the drop of miners, blocks will take a longer time to solve, slowing even more the flow of transactions.

you forgot maybe solo-miner also closes his rigs at this point.

Yeah, but other people with the same idea could do the same thing.

...and people without the same idea could also jump back in.

nice balancing effect, huh?

[Brunic]

...and people without the same idea could also jump back in.

nice balancing effect, huh?

Yeah, but right now, 95% of the miners include all the transactions, it's the 5% who sort which transactions to include that seems to disturb a couple of people around here.

I don't think solo-miner is a problem, it's more that people think that miners managing the flow of transactions is a problem. We'll see next year, but I'm pretty sure some people will start sorting transactions (like me, if I have the means of doing it, I'll do it, no problems).

[FreeMoney]

I'd also say that more fees also would encourage more honest miners to ramp up their hashing power, diminishing this guy's %.

Indeed, making the definitely wrong assumption that everyone has the same costs would mean that even small tx fees would exclude tx excluders from profiting. I would think the 10% number mentioned as a possible near future amount would do a lot of good.

[Stardust]

Instead of increasing TX fees, we better see this as an attack, and should look for another solution.  The more popular Bitcoin becomes, the more enemies it will have.

[muyuu]

I'd also say that more fees also would encourage more honest miners to ramp up their hashing power, diminishing this guy's %.

Indeed, making the definitely wrong assumption that everyone has the same costs would mean that even small tx fees would exclude tx excluders from profiting. I would think the 10% number mentioned as a possible near future amount would do a lot of good.

Who said excluding? I said lower %. Easy concept really.

There is no way you can exclude dishonest miners from mining on free computing power.

[Haplo]

I'd also say that more fees also would encourage more honest miners to ramp up their hashing power, diminishing this guy's %.

Indeed, making the definitely wrong assumption that everyone has the same costs would mean that even small tx fees would exclude tx excluders from profiting. I would think the 10% number mentioned as a possible near future amount would do a lot of good.

Who said excluding? I said lower %. Easy concept really.

There is no way you can exclude dishonest miners from mining on free computing power.

I disagree, I think it should be fairly easy. In terms of fees, for the moment, the gain from mining coins is larger than the gains from any reasonable transaction fees. After looking over the numbers, the drop in coin production is likely to be met with a corresponding rise in FX rate, which will probably cover most of the cost of mining, and is already taken care of within the BTC framework.

However, I second Theymos in supporting gmaxwell's proposal. I think Gavin's idea would probably lead to a threshold war with the botnet(s). However, in contrast to gmaxwell's proposal:

Make every full node also give a getmemorypool style command on the network port which gives out a set of transactions, along with H(H1||H2).  A botnet that wants to mine without having a copy of the blockchain can trust random nodes to provide the proof of memory, but if it does so it must take the exact set of transactions that node provides.

This seems to me to be basically pointless. As miners become more reliant on fees and less reliant on coin generation, they will not likely want to donate their tx fees to some non-contributing bots. The programming effort implied would be wasted towards placating non-contributors without actually providing any benefit to legitimate operators. BTC would be better off just giving them the stick.

Elsewhere it has been proposed to extend bicoind to allow for remote one:many serving of the blockchain, and if that were implemented it might allow the botnet to contribute legitimately for very similar costs. This addition might also lower the cost of current and new mining rigs since it eliminates the need to store the whole 2GB block chain on every rig, and it would allow users to serve their devices from a BTCBank or from a machine they own. Of course, I don't know enough specifics about BTC to know if this is actually feasible, but it sounds good anyway .

[muyuu]

I'd also say that more fees also would encourage more honest miners to ramp up their hashing power, diminishing this guy's %.

Indeed, making the definitely wrong assumption that everyone has the same costs would mean that even small tx fees would exclude tx excluders from profiting. I would think the 10% number mentioned as a possible near future amount would do a lot of good.

Who said excluding? I said lower %. Easy concept really.

There is no way you can exclude dishonest miners from mining on free computing power.

I disagree, I think it should be fairly easy. In terms of fees, for the moment, the gain from mining coins is larger than the gains from any reasonable transaction fees. After looking over the numbers, the drop in coin production is likely to be met with a corresponding rise in FX rate, which will probably cover most of the cost of mining, and is already taken care of within the BTC framework.

However, I second Theymos in supporting gmaxwell's proposal. I think Gavin's idea would probably lead to a threshold war with the botnet(s). However, in contrast to gmaxwell's proposal:

That still doesn't exclude them. You diminish their gain some %. Exactly what I was pointing out since the beginning.

[DeathAndTaxes]

That still doesn't exclude them. You diminish their gain some %. Exactly what I was pointing out since the beginning.

Exactly.  If tx fees are significant enough a botnet will include them. 

Today excluding all tx costs a miner roughly 0.06% of potential revenue.  He likely is thinking OH NOES I only get 99.94% of potential revenue.  Whatever will I do.

If hypothetically tx volume doubled over the next year and fees increased to just 1 bitcent per tx on average after the subsidy cut excluding tx fees would cost a miner 6% of potential revenue. 

Is that enough?  Who knows but I would imagine 6% gross loss vs 0.06% gross loss at least makes a miner think what could I change to collect some of that uncollected revenue.

Every miner (even "mystery") has a tipping point. 

[Haplo]

That still doesn't exclude them. You diminish their gain some %. Exactly what I was pointing out since the beginning.

Gavin's proposal would reduce their % by delaying the empty blocks such that most of them expire.

gmaxwell's proposal would require them to maintain the list of tx in order to have their blocks accepted at all. That is, if they don't have access to the primary blockchain/tx data, they cannot even construct valid blocks. That would be very hard (although not necessarily impossible) for a botnet to get around.

In the case that they did somehow get around it, Gavin's proposal could be used on top of gmaxwell's to delay "cheap" blocks and reduce the total impact on the network. I don't think that would even be necessary, and the more miners depend on tx fees, the less of a difference it will make.

However, in order to produce a similar income to coin mining, at 80tx per block, it would require a fee of around 0.5BTC, or US $2.50, which is pretty exorbitant even for an ATM withdrawal. While that would certainly give miners an incentive to boost their capacity, it doesn't actually solve the freeloader problem and it would create a disincentive for BTC users. Such a rate hike would be far above the design of the currency.

That still doesn't exclude them. You diminish their gain some %. Exactly what I was pointing out since the beginning.

Exactly.  If tx fees are significant enough a botnet will include them. 

Today excluding all tx costs a miner roughly 0.06% of potential revenue.  He likely is thinking OH NOES I only get 99.94% of potential revenue.  Whatever will I do.

If hypothetically tx volume doubled over the next year and fees increased to just 1 bitcent per tx on average after the subsidy cut excluding tx fees would cost a miner 6% of potential revenue. 

Is that enough?  Who knows but I would imagine 6% gross loss vs 0.06% gross loss at least makes a miner think what could I change to collect some of that uncollected revenue.

Every miner (even "mystery") has a tipping point. 

Again, that behavior is part of the design of the currency, and should not be a concern. It's also not an appropriate solution to the problem at this stage of the currency's development.

[DeathAndTaxes]

However, in order to produce a similar income to coin mining, at 80tx per block, it would require a fee of around 0.5BTC, or US $2.50, which is pretty exorbitant even for an ATM withdrawal. While that would certainly give miners an incentive to boost their capacity, it doesn't actually solve the freeloader problem and it would create a disincentive for BTC users. Such a rate hike would be far above the design of the currency.

This is a strawman.  Fees don't need to be EQUAL to the block reward they just nee to be relevent.

If you make $100K a year and I say "hey man for a couple hours of setup work I can show you how you can make a WHOLE PENNY more next year" you likely won't be interested.  However if I said "hey man for a couple hours of setup work I can show you how you can make $10K more a year you might be.

$10K < $100K but it is at least relevant fraction.
$0.01 < $100K and is so tiny as to be immaterial.

To influence behavior fees don't need to exceed block subsidy they just need to be a relevant FRACTION of block subsidy.

As I showed above after the subsidy drop if avg fees were a mere 1 bitcent per transaction it would be a relevent fraction (~6%) of potential revenue and that would be hard for any miner (botnet or not) to ignore.

Again, that behavior is part of the design of the currency, and should not be a concern. It's also not an appropriate solution to the problem at this stage of the currency's development.

So you say, I disagree.  If 1 cent per tx kills Bitcoin then it was going to die anyways.   I am not saying it needs to be part of the protocol we simple need to get enough miners to exclude non-paying transactions to force freeoaders to invest in the success of Bitcoin.  If you think 1 cent is too much to be able to transfer an unlimited amount of value anywhere in the world instantly then you simply have unrealistic expectations and we may need to start working on that.

[Haplo]

So you say, I disagree.  If 1 cent per tx kills Bitcoin then it was going to die anyways.   I am not saying it needs to be part of the protocol we simple need to get enough miners to exclude non-paying transactions to force freeoaders to invest in the success of Bitcoin.  If you think 1 cent is too much to be able to transfer an unlimited amount of value anywhere in the world instantly then you simply have unrealistic expectations and we may need to start working on that.

Ok, in the example you gave, you're saying around 10%. Each block yields 50BTC, equiv $250 USD, at 10%, that would be $25 USD, or 5 BTC. 5BTC/80tx = .0625BTC = $0.31 per tx. Still pretty high for a tx, especially if it's for small amounts.

If $0.31 causes there to be less than 80tx per block on average, that means even higher fees will be required to maintain 10%, which is a further disincentive to making tx. Until the number of coins generated drops significantly, and until the number of average tx increases significantly, this is still not enough of a disincentive for the botnets to stop mining empty blocks, nor is the incentive high enough for them to take on the whole 2GB blockchain on every one of their nodes.

While this isn't a huge problem right now, if nothing is done to deny their entry into the network, it could eventually become a 51% attack, or could be used as a stepping stone for one. Artificially beefing up the miner's pocketbooks would be a far more expensive and less effective solution than simply changing the code to exclude freeloaders. Depending on how much BTC appreciates, the opportunity cost of freeloading might continue to be justified even after 2017. It also depends on the cost of running dummy miners vs miners w/ full blockchain. If the savings is significant enough (apparently it is for the moment), then direct disincentive will be required to prevent them from competing with legit miners or else the performance and the security of BTC will continue to be compromised.

[FreeMoney]

So you say, I disagree.  If 1 cent per tx kills Bitcoin then it was going to die anyways.   I am not saying it needs to be part of the protocol we simple need to get enough miners to exclude non-paying transactions to force freeoaders to invest in the success of Bitcoin.  If you think 1 cent is too much to be able to transfer an unlimited amount of value anywhere in the world instantly then you simply have unrealistic expectations and we may need to start working on that.

Ok, in the example you gave, you're saying around 10%. Each block yields 50BTC, equiv $250 USD, at 10%, that would be $25 USD, or 5 BTC. 5BTC/80tx = .0625BTC = $0.31 per tx. Still pretty high for a tx, especially if it's for small amounts.

If $0.31 causes there to be less than 80tx per block on average, that means even higher fees will be required to maintain 10%, which is a further disincentive to making tx. Until the number of coins generated drops significantly, and until the number of average tx increases significantly, this is still not enough of a disincentive for the botnets to stop mining empty blocks, nor is the incentive high enough for them to take on the whole 2GB blockchain on every one of their nodes.

While this isn't a huge problem right now, if nothing is done to deny their entry into the network, it could eventually become a 51% attack, or could be used as a stepping stone for one. Artificially beefing up the miner's pocketbooks would be a far more expensive and less effective solution than simply changing the code to exclude freeloaders. Depending on how much BTC appreciates, the opportunity cost of freeloading might continue to be justified even after 2017. It also depends on the cost of running dummy miners vs miners w/ full blockchain. If the savings is significant enough (apparently it is for the moment), then direct disincentive will be required to prevent them from competing with legit miners or else the performance and the security of BTC will continue to be compromised.

He was (clearly) talking about a time coming soon when total tx fees may make up 10% of the reward miners get for mining. Not 10% tx fees.

simply changing the code to exclude freeloaders

Go for it. Or maybe you could just read the discussion, that might be more your level.

[DeathAndTaxes]

So you say, I disagree.  If 1 cent per tx kills Bitcoin then it was going to die anyways.   I am not saying it needs to be part of the protocol we simple need to get enough miners to exclude non-paying transactions to force freeoaders to invest in the success of Bitcoin.  If you think 1 cent is too much to be able to transfer an unlimited amount of value anywhere in the world instantly then you simply have unrealistic expectations and we may need to start working on that.

Ok, in the example you gave, you're saying around 10%. Each block yields 50BTC, equiv $250 USD, at 10%, that would be $25 USD, or 5 BTC. 5BTC/80tx = .0625BTC = $0.31 per tx. Still pretty high for a tx, especially if it's for small amounts.

No I said with TRANSACTION VOLUME DOUBLED
WITH SUBSIDY CUT TO 25 BTC
AND A 0.01 BTC FEE.

and that was only an example

25 BTC subsidy
160 tx * 0.01 = 1.60 BTC
1.60 / (25 + 1.60 )  =  6%

A 1 bitcent tx fee (avg) on 160 tx per block after the subsidy cut would make fees contributing 6% of miner's total reward.

While this isn't a huge problem right now, if nothing is done to deny their entry into the network, it could eventually become a 51% attack, or could be used as a stepping stone for one. Artificially beefing up the miner's pocketbooks would be a far more expensive and less effective solution than simply changing the code to exclude freeloaders. Depending on how much BTC appreciates, the opportunity cost of freeloading might continue to be justified even after 2017. It also depends on the cost of running dummy miners vs miners w/ full blockchain. If the savings is significant enough (apparently it is for the moment), then direct disincentive will be required to prevent them from competing with legit miners or else the performance and the security of BTC will continue to be compromised.

I disagree.

First 51% and 1tx block have nothing to do with each other.  51% attack can happen even if botnet includes all txs.

So lets keep it on the topic of 1tx blocks.  As shown in that example above even a modest fee after the subsidy cut results in fees contributing a SMALL BUT SIGNIFICANT PORTION OF REVENUE.

Hashing power follows price.  More correctly hashing power follows block compensation.  Fees adding 6% to block means 6% rise in hashing power.  That makes "mystery" share smaller.  Mystery sees revenue decline due to subsidy cut and then sees it decline even further due to tx fees (which he is excluding) making up greater share of total rewards. 

Unless mystery is an idiot he is going to start saying ... "I would like 6% more revenue lets see how I can get a cut".  If he doesn't then maybe 8%, 10%, 15% eventually drives him to the tipping point.

Someday fees will need to support 100% of the network.  We can start the ball rolling at > 0.03%.


I haven't seen a single proposal that would either work or wouldn't have significant side effects and open the network up to "gaming" and malicious attacks.  Protocol changes should be the last choice.  Miners requiring higher fees is a simple strategy to try first.

[Haplo]

First 51% and 1tx block have nothing to do with each other.  51% attack can happen even if botnet includes all txs.

It does, depending on how the botnet is handling tx. If the botnet is running off a few servers providing the blockchain, then that botnet would increase the amount of computing power needed to hit 51%. As it stands, since the botnet is blind, it can be incorporated by an intelligent attacker to make up 51%. At the current botnet size, it gives about 15% leverage either way, although this may increase or decrease in the future.

Unless mystery is an idiot he is going to start saying ... "I would like 6% more revenue lets see how I can get a cut".  If he doesn't then maybe 8%, 10%, 15% eventually drives him to the tipping point.

Someday fees will need to support 100% of the network.  We can start the ball rolling at > 0.03%.

I haven't seen a single proposal that would either work or wouldn't have significant side effects and open the network up to "gaming" and malicious attacks.  Protocol changes should be the last choice.  Miners requiring higher fees is a simple strategy to try first.

Well, in 2013, when that becomes relevant, then we can talk about what a reasonable fee is. 1 bitcent isn't bad, but it can't be predicted how the value of BTC will respond to the drop in subsidy, or what kind of losses that will mean to the miners (or not). If the value of 1 BTC goes to $10, then an even smaller fee would suffice, since technically the miners wouldn't be losing anything.

Let's just say mystery saves, say, 25% of costs by not running the blockchain, and the value of BTC doubles, then he will still have no incentive to play nice, and may even decide to take a larger chunk. The only way that would not be the case is if bitcoin's value appreciated more slowly than the subsidy decrease, or if fees became fairly exorbitant.

The main problem is that, for the time being, the BTC community is fairly vulnerable, and it is precisely during this time when that 15% of total processing on the network would be the biggest liability to BTC's survival. If someone with the will and the resources figures that out, we could have a big problem. Even if not, that's still 15% of all BTC produced going towards someone who isn't investing in the actual function of the network. Ideally, the means should be provided so that tx's can be mined properly without the cost of a full blockchain, then mystery would be adding 17.5% or so to the total processing power of the network rather than wasting 15% running in place.

Also, I don't see how gmaxwell's proposal, with my revisions, would open up the BTC network to "gaming". Every block must be cryptographically valid as well as providing proof that it's running on correct blockchain information. It's pretty hard to make an attack within those constraints.