## Will Bitcoin suffer from a mining Tragedy of the Commons when mining fees drop to zero?

64

20

One issue that I haven't seen clearly addressed is the possible Tragedy of the Commons in Bitcoin's future (see this answer). 40 years into the future, the only significant direct benefit for honest miners is transaction fees, as the block reward is almost zero.

The potential problem here is Tragedy of the Commons - it is possible that while the entire network will benefit from miners, the financial incentive for any individual honest miner to mine will be too small, so fewer miners "than necessary" will mine, leaving the network open to 51% attacks. In this scenario, the Bitcoin network will fall pray to frequent attacks and will effectively disintegrate. Also consider extensions like Open Transactions, that might make conducting transactions on the real Bitcoin network unnecessary for most users, who will flock to 2nd level networks and reduce the miner's rewards from TX fees even more.

As Bitcoin is being designed and promoted as a Future Proof system, people have tried to analyse the system's economic behavior in this future.

What analysis has been done on this problem? Is the problem likely to happen (speculative question, I know ... but back up your arguments)? If there is a likely problem in the current implementation, what are some possible way to adjust the protocol in the future to combat this problem?

Update - please see here, I described a possible scenario that leads to network insecurity because of this issue.

Update 3 - Added a wiki entry.

Doesn't the difficulty rate for a block stop if all coins are mined? And then, instead of miners solving blocks, just have the users checking transactions of others whenever they do a transaction? No need for transactioncost. No high power consumption. No threat of centralizing bitcoin. Is that possible? – None – 2013-10-07T22:38:09.600

No, the difficulty has nothing to do with the progress of bitcoin, it increases to balance the hash rate of the network to make the network come up with one block per 10 minutes in average. – Murch – 2013-10-07T23:00:55.013

This is very closely related: http://bitcoin.stackexchange.com/questions/1226/the-end-of-mining I like my version of the question better though, so I'm not voting to close as dup.

– ripper234 – 2012-03-08T13:34:16.137

I think the answer you cite is still very accurate and as of now we've seen no changes in the way bitcoin works that could solve the problem. – Stéphane Gimenez – 2012-03-08T15:03:16.850

This question is good due to poiting out the problem with 2nd level networks. – o0'. – 2012-03-08T16:07:40.280

@Lohoris: Sure, but "yes" is still the answer. – Stéphane Gimenez – 2012-03-08T22:18:11.337

1Consider the TOR network. Now they discourage everyone to relay the TOR network and be more than a client only except if the user sets up a limited liability company at least to mitigate his legal exposure. And TOR is still faster and stronger than ever! – superuser – 2013-01-12T22:21:15.823

20

Mike Hearn just posted about how Network Assured Contracts handle this problem. I don't find an immediate flaw with this.

This is how I understand the proposed solution:

Anyone with an interest in a high hash rate (basically, anyone holding a large amount of coins), can initiate or cooperate on SIGHASH_ANYONECANPAY transactions. Those are an effective way for people to say stuff "I pledge 10 BTC for the next miner to mine a block, provided 100 total BTC is donated in this transaction ... otherwise, I'll get my money back in 5 blocks"?

So, if people or organizations holding a large amount of BTC see that the security is too low for their standard, they can chip in ... provided others do so as well.

Hey ripper234, could you try to explain the solution in your answer (instead of just having a link)? Would be great. – D.H. - bitcoin.se – 2012-03-09T14:59:10.063

@D.H. - added a snippet that explains how I interpret Mike's answer. – ripper234 – 2012-03-09T16:13:19.197

3That's a good approach in that it decouples mining compensation from transactions. But I don't think it solves the fundamental problem. If someone thinks the chance of passing the threshold is close to 100%, he'll have no incentive to offer his own funds. So if people offer funds it means the chance is not close to 100%, which means there is always a risk of the whole thing crumbling. – Meni Rosenfeld – 2012-03-10T17:05:02.890

@MeniRosenfeld - see a response I made to a similar concern: https://bitcointalk.org/index.php?topic=67900.msg793820#msg793820

– ripper234 – 2012-03-10T17:47:42.803

@ripper234: If you mean cunicula's point #5, it's similar but not the same. Vagueness makes sure there are probability estimates rather than waiting games; but the point remains that a player will contribute only if there's a positive probability of failure. – Meni Rosenfeld – 2012-03-10T18:08:26.993

@ripper234: And I guess I'll need to think re the iterated game perspective. I have the feeling that the logic of iterated PD can't be perfectly applied. – Meni Rosenfeld – 2012-03-10T18:20:18.740

@MeniRosenfeld - to make my point more explicit about the iterated version - if the assurance protocol is initiated several times, but nobody can get the required number of donations (or just too few overall donations are obtained) , people will realize they have to "selflessly" participate in order for it to work .. because it really is in their long term interest. – ripper234 – 2012-03-10T23:38:26.750

@ripper234 But the whole point of the tragedy of the commons is that people don't do things in their long term interest: if they did, we would have no problem in the first place. – o0'. – 2012-03-11T10:23:10.357

2@Lohoris - the situation is not identical. In classic TotC there's no efficient mechanism to declare "I will donate to the common good but only if 99 other people also do the same", and really mean it. – ripper234 – 2012-03-11T11:07:44.463

@ripper234 this is a fair point, but I wouldn't be so sure it would be enough. – o0'. – 2012-03-11T11:16:02.010

Please do not move random discussions which are not backed up with some solid theory to this site. – Stéphane Gimenez – 2012-03-11T12:24:29.663

@StéphaneGimenez - the question as asked can have an answer that's verifiable in game theory/economics models. Let's not start a debate here whether this question is a good fit or not (you can open up a meta post if you want). – ripper234 – 2012-03-11T13:13:17.780

– Stéphane Gimenez – 2012-03-11T13:14:11.207

23

There will definitely be a tragedy of the commons problem if things stand as they are now. This was discussed at some length here and elsewhere.

There are some proposed ways to address this and make transaction fees nonzero (block size limit, hardcoded fees, insurance entities, mining cartel, gentleman's agreements which are maintained for fear defection would beget more defection, etc.), but I don't think either results in the combination of competitive transaction fees and a level of hashrate which is enough to offer reasonable security.

This means that there must be some way to improve the network's security at a given hashrate. The only way I know to do that is by augmenting proof-of-work with proof-of-stake, discussed for example here. This is based on the observation that obtaining a majority of the bitcoins in existence is orders of magnitude more difficult than obtaining a majority of the hashrate.

7Proof-of-stake is worse in every critically important way. It not only concentrates wealth letting the rich gain most of the mining. By the same point that the barrier to entry for the monopolist is higher, so is the barrier to removing the monopolist. The monopolist can harm the network without harming himself because it gives him the power to fork the system. Govts have deep pockets and hate anonymity. War on Terror is an example that majority are sheep and will follow the official forks. Once you lose decentralized control, you'll never be able to compete to get it back! THINK PLEASE! – Shelby Moore III – 2013-03-20T20:21:34.487

Wow, very long discussion - thanks! I'll look into that. – ripper234 – 2012-03-09T11:06:29.977

2How about changing Bitcoin (or starting Bitcoin2) and canceling the halving of the block reward? Each block can generate 50 BTC forever. This is inflation ... but controlled inflation. Miners still have some motivation to mine regardless of transaction fees. I'm not convinced it's the right solution because there's no way to know a-priori if a 50 BTC block reward will be valuable enough to incentive miners in the future. – ripper234 – 2012-03-09T12:23:38.960

1

– ripper234 – 2012-03-09T12:33:46.513

Unaccepted - see http://bitcoin.stackexchange.com/a/3129/78

– ripper234 – 2012-03-09T14:41:10.090

1@ripper234: Of course if there's perpetual inflation there will be much less of a problem. Personally I'm not necessarily against the idea, but most people are. Anyway, ideally inflation would be decided on economic principles, and security should be decoupled from it. The coupling of security and inflation in Bitcoin was never designed to be permanent. – Meni Rosenfeld – 2012-03-10T17:25:43.257

@ripper234 That is actually decreasing inflation, percentage wise. – PyRulez – 2016-05-09T16:46:35.923

Proof of stake does not concentrate wealth in the rich. It moves wealth to those people that run software to process transactions (which is desirable). All else equal, if you own X% of all the coins in existence, you will continue to own X% of all the coins in existence as long as you are staking. Doesn't matter if the "reward" rate is 1% or 100%, This rate is effectively just the size of the penalty of holding without staking. – hedgedandlevered – 2018-10-16T18:36:58.193

8

I've read some of the linked discussions, and it seems some of the participants fail to understand the basic economic theory of the marginal cost. In any high fixed-capital business, the net present value (NPV) determines the ROI (IRR) and determines the opportunity cost where investors apply their capital. Thus, in normal functioning markets, the lowest marginal cost provider does not expand their capacity to take all of the volume at their cost, rather the market price is set by the highest marginal cost provider and the lower marginal cost sell at that price to maximize ROI and thus investment. Realize that not all investors have the same opportunity cost (some have better competing opportunities than others), thus the supply of investors at any given ROI is finite.

However there is a risk. The fundamental change facing Bitcoin's future is the shift from a forced highest marginal cost price structure— because all miners get the same price for mining a block— to a free market transaction fee pricing structure where dumping or subsidies could be used to destroy the higher marginal cost miners and gain a monopoly, thus also compromising the security of the system. However, a monopoly strategy could be applied in the current Bitcoin by dumping or subsidy to force difficulty rates up thus destroying higher marginal cost miners, and/or simply being a greater proportion of the total system hash computing power.

A solution to this problem is simple and is necessary for another reason. A market-based transaction fee does not guarantee any level of hash computing power, unless the user is motivated to pay more for more security and/or transaction processing speed. Either this is incentivized by having the recipient pay the transaction fee (which is a good idea any way), or the transaction fee should be set globally by the system to maintain a desired level of hash computing power (which means transactions that don't need it are penalized but at least it is consistent). Thinking through this further, market-based fee even if incentivized as stated, may not guarantee enough profitable business to offset the determined monopoly strategy, thus a global system transaction fee is safer.

However, a sufficiently subsidized monopoly strategy can still drive the difficulty up and/or capture a larger proportion of total hash computing power.

One potential counter-measure is LiteCoin's Scrypt, as it should make it more expensive to monopolize and reward more CPUs to be miners thus decentralizing stake, although it does open the botnet risk. If debasement is paid back to users by encouraging them to run full nodes and leveling the playing field between ASICS and CPUs using Scrypt, then it is the world's processing power versus the monopolist. The user who is not gaining from mining and thinks they are running a full node, could be advised to run a test for presence of a botnet. This plays well with my conceptualization that perpetual debasement is necessary (of which some might disagree so this answer is not making that claim). Btw, I think Proof-of-Stake moves in the opposite direction by centralizing stake.

A potential means for someone without deep pockets to monopolize Bitcoin, would be if ever there was a way to borrow sufficient Bitcoin to purchase the necessary 51% computing power, then destroy Bitcoin forcing an inflationary spiral as holders stampede to sell out, then repaying the borrowed Bitcoin at a fractional of the original value.

The current block size transaction limit doesn't accomplish anything other than cap throughput, because it doesn't move even if hashing capacity expands by either of the simple mechanisms in the prior paragraph.

Given my prior two answers got negative votes, I am also expecting this answer to fly over the heads of some voters. Apparently some of the participants either don't have a very good grasp of economics (or have a vested interest or I am wrong and they didn't comment to tell me why). I would appreciate if downvoters would at least try to defend their logic with a comment below my answer. That gives me an opportunity to debate them and show them why I think they are wrong (or to admit my mistake). The point is to make sure we collectively have the correct logic.

1Please humor me with more anonymous downvotes. Bitcoin going forward will require transaction fees. How will it compete against a no transaction fee model with perpetual debasement? Since current investment in Bitcoin hinges on expectation of rising demand from future transaction volume, such a competitor pops the ponzi bubble. Also due to lack of scrypt, it is no longer feasible to earn some Bitcoin with the excess CPU power idling wastefully on your PC to drive demand for transactions. – Shelby Moore III – 2013-03-20T21:55:05.340

1The bitcoin borrow and destroy option is an interesting one. Could also be used to leverage all of the crypto markets as it would affect confidence across all markets. – MarkWalls – 2014-08-03T15:37:49.357

4

Public Service Mining seems quite plausible to me. In 40 years time one would hope that Bitcoin would have a large number of stakeholders who have a very strong incentive for the financial system to keep running smoothly. Those stakeholders I would expect to put non-zero effort into mining even for near-zero direct reward.

4

I now see a much more urgent and catastrophic threat vector potentially circa 2016.

Difficulty will tend to rise and fall such that the system-wide cost of mining is nearly the value of the coins mined (not perfectly but on a smoothed basis roughly true).

Thus the lowest-cost marginal miners will earn the highest profits and rates-of-return. Thus, as Satoshi Nakamoto anticipated, mining will become centralized among those with more capital who can buy the highest performance hardware.

This is more accentuated by the designed volatility in the price of Bitcoins, because a) marginal cost producers can go through periods of negative profits or no production, and b) more capital means more inertial reserve, economy-of-scale access to capital funding markets, and access to debt to delay insolvency so as to smooth volatility in market conditions.

EDIT: Relevant excerpt from linked question above which was deleted:

since Bitcoin is such a minuscule portion of the economy and can't create such global "network effects" on real productivity, appreciation in FX price is the main mechanism for evidence of the creation of value. Appreciation in price does create some value because it encourages the teaching, investment in, and thus spread of the currency. However, when the appreciation in FX price moves exponentially faster than any human calculable time-preference for opportunity cost, and exponentially faster than growth in spending for goods & services can (due to its connection to real world of profit & loss that adapts more slowly than speculation), then the price is dominated by wild guessing, i.e. emotional speculation.

So to capture a majority or maybe even most of the mining, an entity probably merely needs to subsidize on a large-scale.

The government or powerful elite has the funds to do this, either in public view using taxes (or deposit insurance on P2P currencies) or in documented black budgets.

Is there a scenario where the public would clamor for the government to fix a problem by doing this?

Imagine the public became heavily invested in wild "dot.com 2.0" run of Bitcoin price appreciation to say $1000,$10000, or $100000 (as some people expect). All bubbles eventually crash. Tangentially, the creator of Bitcoin would probably be a$billionaire.

So with an exponential crash in price, there could be a drastic reduction in miners, joined with a public wanting to blame something or someone. The government could probably find massive support for deposit insurance.

Thus I conclude that the computational Proof-of-Work design of Bitcoin is a serious threat to anyone who views Bitcoin as being independent of the control of the government. Noting that correlation isn't proof yet via Occam's Razor it is ostensibly designed for this outcome and thus to fall into the lap of the government.

And we are not talking about 2040. The pricing froth is building now so this could happen soon. And we also have another halving of the debasement payout rate in late 2016.

Which Btw, is when the repeating-throughout-all-recorded-history Economic Confidence Model expects this current bounce of capital flows into the USA dollar economy to turn hard down again a la 2008— exactly when we will need a P2P currency to help us avoid chaos of capital controls and liquidity collapse. Bitcoin could fail at precisely the time when we will need it most. That is what concerns me so much.

I would very, very much appreciate if any downvoters sign with a comment, in case this answer becomes prophetic we will be able to know how prescient you were.

this answer is very speculative, I think it'd be worthwhile improving this with some references. – rny – 2017-10-27T01:21:25.157

1

The "40 years" time limit isn't relevant at all. Since the reward for mining gradually decreases, miners will have to adjust along the way much before that time; actually they'll have to adjust every time the reward halves, unless they are operating with profits so high they will still be gaining money after the drop.

Since most of the miners are part of a mining pool, the tragedy of the commons can be easily avoided: the problem with the tragedy of the commons is that many individuals do not coordinate and can only act short-term instead of long-term. However a pool is a group of many individuals, hence it is much easier for a pool to decide to accept only transactions with a relevant fee.

@ripper234 updated, considering mining pools – o0'. – 2013-04-14T09:22:55.080

Please see my update for a detailed scenario. I'm trying to understand whether Bitcoin can have a steady state where mining rewards are essentially all from TX fees. – ripper234 – 2012-03-09T06:54:42.710

1

There is always an incentive to mine honestly

At present every block comes with a built-in financial reward through the creation of bitcoins. Over time, through scarcity and use, these bitcoins can be expected to go up in value. Therefore it is possible that the act of mining can be somewhat sustained purely by the value of the reward (shared out through an efficient pool) until that last bitcoin is mined.

However, that is not guaranteed by any means so we have difficulty and transaction fees to act as safeguards. As miners shut down their loss-making equipment the difficulty reacts until it becomes profitable again. Remember, in the very early days bitcoins were almost valueless but people still mined effectively because of the very low difficulty.

Most importantly, there is the transaction fee that which is the final, and most reliable, safeguard of all. This ensures that if a Bitcoin transaction is important enough to the client then they will pay that fee to get it mined - probably paying more for faster mining. This is especially true of second level operations (like OT) that will rely on Bitcoin for efficient reconciliation. There are only so many "paper transactions" that can occur before a margin call is made.

Assuming that Bitcoin is successful due to it's utility, it is reasonable to further conjecture that high value transactions will mix in with low value ones in a fair manner. Distributed mining pools will continue to share out the work and the subsequent fee reward based on contribution.

Overall, the combination of these forces will always tilt the balance in favour of honest miners getting more reward than dishonest ones.

To that end it seems unlikely that a Tragedy of the Commons (where individual gain is paid for by the group until a resource is used up) will apply.

You seem to be forgetting that if the difficulty drops too low then the network is vulnerable to 51% attack. It seems entirely possible that the transaction fees offered voluntarily won't be enough to encourage miners to keep the difficulty high enough to avoid this attack. Ok, you're probably not forgetting, but your answer doesn't seem to explain how the difficulty can be kept high enough. – Highly Irregular – 2012-03-15T22:16:57.797

1

Adjust difficulty per transaction in the block?

As I understand it a big part of the problem is that the fees people are paying might turn out to be too low.

[UPDATED] As Nick mentions in the comments of this answer: The only reason a miner would exclude a non-zero fee transaction is to save space on the block, thereby hoping that in the event of a another miner finding the block at the same time their smaller block will reach more nodes faster and be accepted. Of the top of my head this seems like it might not be incentive enough, especially if communication technology keeps increasing.

My idea is to link the difficulty required for the solution of a specific block to both the current network difficulty as usual and to the number of transactions (or total size of) the transactions in the block. Meaning transactions now have to compete to be worthwhile to add to a block.

I have no idea on feasibility, or if it's redundant with existing mechanisms but I assume that a change could be made to the protocol to make it so that when checking a blocks validity it takes this rule into account? Probably only if the block reward has reached or neared zero and over a certain minimum threshold number/size of 'difficulty free' transactions (which could be variable based on the utilization over the previous time period?).

I haven't sat down and though this through or looked at the bitcoin source code however. :)

"Because any miner has nothing to lose by including every transaction with a non-zero fee" Not true. Having a larger block means that it will take longer to be relayed, and thus is more likely to be stale. – Nick ODell – 2013-04-06T19:33:46.543

Ah true. This means there is some incentive for the sender to out-price other transactions on a per byte basis, I wonder if it's significant enough. I'll amend my answer, thanks for the correction. – Jack Casey – 2013-04-08T01:04:54.437

0

One way to counter such a problem might be a miner that would be forcing some fee from each transaction it puts in a block. When the reward for generating a block will be entirely gone, a miner might decide not to even start generating a block until some financial incentive appears. If most miners would follow the same reasoning, soon everyone would have an incentive to pay a fee to get their transactions into a block. If there wouldn't be sufficient coins to be earned, it might be more cost effective for a miner to "sit it out" and wait for the transactions to accumulate.

This situation would decrease the difficulty as the average amount of computing power would decrease. It might also encourage miners to continue mining for their own version of a block after another miner has already solved it just to earn those coins. Both of those things would make it easier for a 51% attack, but by then everyone will know if Bitcoin is here to stay. If the market is the size it is today after such a long time, one might not care to go after or protect Bitcoin. If there is sufficient money to be had, there will be people mining it

All in all, approach to mining probably will change by the time Bitcoin is scheduled to reach maturity and there will be new challenges to address, but there is also ample time to prepare for that.

Please see my update for a detailed scenario. I'm trying to understand whether Bitcoin can even have a steady state where mining rewards are essentially all from TX fees. People are investing in Bitcoin as a "currency to replace all currencies". If that is not the case, I think it's best to try to analyze and detect this now, rather than wait 40 years. – ripper234 – 2012-03-09T06:55:51.927

-3

I propose that the payment processing activities shift away from miners and be placed on the retailers accepting bitcoin transactions. What if every retailer accepting bitcoin transactions for more than an average of approximately \$5 USD was asked to run a small scale, highly efficient mining rig at their place of business? Or even a highly efficient raspberry pie miner at every retailer accepting bitcoin? The cost of the hardware and electricity for the retailers to dedicate hashing resources to the community would be much less than the current payment processing fees charged by centralized banking entities. This would keep payment processing decentralized, the hashrate acceptably high, and eliminate the need for miners. I do not see miners adding any value to the bitcoin project now or in the future. In fact, I think the entire project is at risk due to the current mining structure, and the power of the mining pools. Retailers would voluntarily participate knowing they are doing their part to keep the network safe, secure, and bitcoin transactions affordable. The alternative is to pay whatever mining pools demand as transaction fees.

1

Your argument isn't sound: "and eliminate the need for miners" No it wouldn't. You yourself just proposed that merchants would become miners instead. –– "The cost of the hardware and electricity for the retailers to dedicate hashing resources to the community would be much less than the current payment processing fees charged by centralized banking entities." ← citation needed. –– "Retailers would voluntarily participate" ← citation needed. –– "I do not see miners adding any value" see e.g. What Value does mining provide?

– Murch – 2016-01-30T10:14:14.750

-4

Since you presume that a 51% attack would become beneficial, this will incentivize everyone to attempt a 51% attack. This is similar to an arms race, only the results of anyone doing this in 40 years would be financially cataclysmic. I suspect there will be supercomputers waiting in storage ready for anyone attempting a first strike with a strong retributive response. There will probably also be other forms of retributive attacks on the network leaving nothing off the table.

Sorry, but I was hoping for some more serious analysis of the problem. – ripper234 – 2012-03-08T13:33:48.100

1Your question begs an answer to problems 40 years in the future. I'm not sure how you expect a engineering answer to technologies that haven't been invented yet. Speculation at this point would be science fiction. – cbeast – 2012-03-08T13:42:59.447

1there is concrete analysis that can be, and has been made about this. Just like if I'd ask "how can Bitcoin scale to handle VISA like volume", this question also can have a good answer even at present day, proven by rational economic arguments. I'm not driving at speculation here. – ripper234 – 2012-03-08T13:51:04.227

@ripper234 Master Charge is 40+, a lot has changed in finance since then. It's likely Satoshi chose the 40 year curve for Bitcoin reward just for this reason. Technology will change beyond our current paradigms. Give me a hint as to what kind of analysis you would like to have? Autonomous agents? A new algorithm? Historically, war has been used for long term financial resolution. Bitcoin will dramatically change even before it gets released and will continuously evolve and adapt to our contemporary paradigms and technologies. Anything in 10 years, let alone 40 years would be pure speculation. – cbeast – 2012-03-08T15:27:53.700

Simple answer: Society cannot survive without commons and they are always manageable. Because Bitcoin is decentralized, the free market will find a solution. Already FPGA and ASIC processors are solving the electricity cost factor of mining. Hopefully, we will not legislate a central bank to take over control of the network before then. – cbeast – 2012-03-08T16:11:38.730