will bitcoin become less secure over time?

[cunicula]

Assume the following:

1) profits from attacking network are a fixed fraction of btc money supply
2) costs of attacking network ~= btc rental price of all mining hardware
3) btc rental price of all mining hardware ~= coin generation rate + txn fee generation rate


Implication: security of network ~ (coin generation rate + txn fee rate)/money supply

If the assumptions are correct ( please correct them if they are bad), then network security should fall over time. The implied txn fee necessary to maintain network security will skyrocket. I doubt then that the present level of network security is sustainable in the long run. Security risks are proportional to market cap not data size in kb. Therefore the future txn fee will have to include a fixed percentage of send amounts in addition to a fee per kb. Otherwise the wealthy will freeride on network security paid for by the btc poor. The tax on the poor will become ridiculously large without this modification.

In terms of network security, bitcoin appears to be a lot like a ponzi scheme.

(for intuition on the wealthy being forced to pay for security, See articles in the Journal of Political Economy by Earl Thompson. Intuition is that those with movable assets provide a temptation to attackers. Therefore fees for collective security should be levied on owners of movable assets. Clearly this is how the private market works (you don't see homeless people walking around with bodygaurds).

[1715095289]

1715095289

[1715095289]

1715095289

[1715095289]

1715095289

[1715095289]

1715095289

[1715095289]

1715095289

[mpfrank]

Seems to me that as the rate of bitcoin generation declines (every 4 years), unless the price also rises, the amount of computing resources dedicated to the network will also decline, unless users pay enough increase in transaction fees to make up for the decreased rate of bitcoin production.  But it's not clear to me what is the incentive for users to pay high transaction fees, unless they are in an unusual hurry to complete their transactions.  So the amount of computing resources dedicated to block hashing will fall, and thus you are right that the cost of an attack will fall, while the value of the money supply stays the same or grows slowly (assuming the price doesn't crash).  So the security of the network may indeed fall, and users' awareness of this issue may put a long-term cap on the price.  So I'd say you are basically right, unfortunately...  Unless someone has a counter-argument.

[myhoho]

 But what we have to do right now?
 The only way which I see is equalization of pools (Deepbit is so mighty that it could easily change something)...

[cunicula]

But what we have to do right now?
 The only way which I see is equalization of pools (Deepbit is so mighty that it could easily change something)...

I think this is a bit of a red herring. The attacker could also use 50 pools all apparently separate, but secretly controlled by one bastard. Point is that any problem now becomes worse over time. I would greatly prefer a system which tried to keep security levels constant (approx.) over time. Front-loading security seems very concerning to me. A little front-loading could be justified by network effects arguments, but i am worried that bitcoin security is far more front-loaded than people realize. It could be that all this security is unnecesary, but then why are we paying miners to provide so much of it?

[ctoon6]

The solution is simple, integrate gpu mining directly into the client, that would easily jack up hash rate if millions of people were using the client. you could also integrate pooling in too. make a new protocol where you select the pool you want in a drop-down box. then you send your BC address to the pool, and the pool automatically sends your coins or when you press send in the client. you could also put in the address of a pool you wanted as long as it is compatible.

[cunicula]

The solution is simple, integrate gpu mining directly into the client, that would easily jack up hash rate if millions of people were using the client. you could also integrate pooling in too. make a new protocol where you select the pool you want in a drop-down box. then you send your BC address to the pool, and the pool automatically sends your coins or when you press send in the client. you could also put in the address of a pool you wanted as long as it is compatible.

Please discuss these off topic issues in another thread. Relevant posts should make comments on the incentives to provide hashing power and the incentives to attack the network and how these are likely to change over time.

[TiagoTiago]

(adding thread to my watchlist; no, it's not the same as clicking the notify button)

[wareen]

The current level of permanent hashing power is probably unsustainable in the long term (due to the declining block reward) but it is not too hard to think of keeping a considerable amount of hashing power on standby, ready to start working as soon as an attack seems to be taking place. Everybody using Bitcoin should have an incentive to contribute some short-term hashing power to secure the network in case of an attack.

If Bitcoin becomes mainstream I could even see a market for some kind of insurance for the Bitcoin network. It would basically only be necessary to contract  some operators of large GPU clusters which are normally used for other scientific or industrial purposes but can be rented short-term on demand for securing the Bitcoin network.

For larger companies (merchants, banks) it could even be cost effective to operate some hashing clusters themselves, using it in part to speed up their own transactions and in part to standby for emergencies.

So even if difficulty will be lower than today, the network could be effectively secured against malicious attacks.

[hugolp]

1. You are ignoring initial investment.

It might not be profitable to buy more hardware because the rate of return is so low that it will take too long to recover, but it might still be profitable to keep the hardware running. Miners will hardly stop mining as long as the income is slightly above the price of electricity. So maintaing a certain hashing power once the network has gotten there is not that difficult.

2. The value of Bitcoin creates incentive both ways.

If Bitcoin value goes up incentivates more cheaters, but it also makes mining more profitable atracting more hashing power. Also, Bitcoin has been created to be used widely, so it will generate a big amount of transaction and therefore transaction fees. The initial compensation for block is a way to jumpstart the system and at the same time distribute the currency.

3. What you are proposing is already posible

The "solution" to the no-problem is already in place. Miners can decide which transactions they process and which transactions they dont. If they feel they are not being payed enough they can decide to not process transactions with a fee that does not met a certain percentage of the bitcoins being transfered, for example a 0.5%. So if someone with a lot of bitcoins wants to transfer them the miners can decide not to unless that person pays a fee that pleases them.


And last, this has been discussed already. Why keep opening threads discussing the same again and again?

[cunicula]

The current level of permanent hashing power is probably unsustainable in the long term (due to the declining block reward) but it is not too hard to think of keeping a considerable amount of hashing power on standby, ready to start working as soon as an attack seems to be taking place. Everybody using Bitcoin should have an incentive to contribute some short-term hashing power to secure the network in case of an attack.

If Bitcoin becomes mainstream I could even see a market for some kind of insurance for the Bitcoin network. It would basically only be necessary to contract  some operators of large GPU clusters which are normally used for other scientific or industrial purposes but can be rented short-term on demand for securing the Bitcoin network.

For larger companies (merchants, banks) it could even be cost effective to operate some hashing clusters themselves, using it in part to speed up their own transactions and in part to standby for emergencies.

So even if difficulty will be lower than today, the network could be effectively secured against malicious attacks.

"Everybody using Bitcoin" has a strong incentive to free ride - both now AND in the event of a crisis. The community is already two or three orders of magnitude too big for that to work. It will only get bigger (well maybe it will).

Where does the fund come from to pay for the clusters in case of attack? Voluntarism?

If voluntarism is sufficient for security, then why incentivize miners to burn fossil fuels and hash money. It is a tax on everyone holding bitcoin (not small, around a 50% rate this year for example).
I have no problem with taxation if it goes to something useful, but the idea of taxing for security now and relying on voluntarism later is problematic to say the least. It seems highly improvident to spend down almost all of the effective security funds in the first few years of operation.

[ercolinux]

Hashing power could decrease in future but is not really probably. There is point of force in the whole system:
if bitcoin will be still there in 6-8 years its base of users will probably be of some millions of users. Due to the limited number of coins their value must be high enough to permit exchange and transactions. If we arrive to 5 millions of users (0.25% of the 2 billion of internet users or 1.3% of the users who are on facebook every day)  and we consider 500$ as a average value of BTC that a single user wants spend in BTC in one year that brings a value of at least 80-100$ for every BTC (the same BTC can be used more times in a year but some are probably lost or used as reserve/speculation). And the transctions fees that now are really low can become intersting for miners (with 5millions of BTC users  there will be at least  8/1000 transctions for blocks, putting an average 0.001BTC of fee on it and you have near 1BTC for block of fees). So mining a 13,5 BTC block gives you even high revenue than today (50x14$ -> 13,5x80$, but even if we consider another 5 years with blocks at 7BTCx80$ mining still gives revenues).

On the other side if you consider only 500.000 nodes with the client running on CPU/APU (100MHash/sec in 5/6 years will probably be normal value for them) you have 50THash of power (enough to keep half of the net sure at the today power with few GPU miners)

If we can't reach this numbers in 6/8 years probably there will not be BTC any more.

[Vandroiy]

And last, this has been discussed already. Why keep opening threads discussing the same again and again?

Because every time it has been actually discussed, the result was Bitcoin security NOT functioning properly after minting, and NO consensus on a choice of block size limit being in place. One cannot reduce a two-variable problem (difficulty, fees) to a one-variable problem (fees) without any explanation as to why this can be done.

Neglect block size limit, and see what you get. All but the cheapest miners will be pushed out of the market, reducing difficulty, which in turn creates competition among the cheapest miners. It's a Tragedy of the Commons, with only the non-profit miners remaining in the end, and difficulty converging to zero.

I have had discussions in lengthy threads about it, and the bottom line was fairly clear. People arguing that current Bitcoin protocol would work don't even have a model they agree on! Some want to remove limits, some don't, some think cartels will do the job, some just neglect the Tragedy of the Commons... and each argument showed significant flaws. The only model that really gets backing has difficulty not only de-coupled from possible attacker size, but also converging to very low values unless limits on transactions push fees to arbitrary numbers.

See the following thread for the discussion, or the linked post in which I tried to summarize models. It's a mess, but if you take some time to think about it, it becomes clear that there is something wrong.

Thread on difficulty equilibrium after minting:
http://forum.bitcoin.org/index.php?topic=6284.0

Post in which I try to show that people who don't see a problem hardly ever agree on a model:
http://forum.bitcoin.org/index.php?topic=6284.msg111735#msg111735


The only thing that looks promising is the block message rule 13: "Reject if timestamp is after the median time of the last 11 blocks", as this could be used to prevent chain splitting after a certain amount of time. But I think things should be formal and proven secure, not just happen to make difficulties for an attacker.

[cunicula]

1. You are ignoring initial investment.

It might not be profitable to buy more hardware because the rate of return is so low that it will take too long to recover, but it might still be profitable to keep the hardware running. Miners will hardly stop mining as long as the income is slightly above the price of electricity. So maintaing a certain hashing power once the network has gotten there is not that difficult.

This is irrelevant. Two reasons:
a) Attackers would probably rent hardware by establishing many different pools and buying bitcoin to pay off participating miners. Miners would not need to know they are participating. The rental rate is the appropriate metric of attack cost. Sunk costs (such as the initial cost of hardware are irrelevant).

b) Even if attackers purchase hardware rather than rent it, old hardware will be broken or poor quality after a few years. In this unlikely scenario, overcapacity due to past investment only postpones a severe security situation, it does absolutely nothing to solve it.

2. The value of Bitcoin creates incentive both ways.

If Bitcoin value goes up incentivates more cheaters, but it also makes mining more profitable atracting more hashing power. Also, Bitcoin has been created to be used widely, so it will generate a big amount of transaction and therefore transaction fees. The initial compensation for block is a way to jumpstart the system and at the same time distribute the currency.

This is accounted for in the assumptions I gave. I think it is quite clear that without expanded txn fees the incentives to hash decline exponetionally, while the incentives to attack increase additively. Txn fees will need to increase exponentially to offset these two effects. Without a percentage-based txn fee, there is no way this will work.

3. What you are proposing is already posible

The "solution" to the no-problem is already in place. Miners can decide which transactions they process and which transactions they dont. If they feel they are not being payed enough they can decide to not process transactions with a fee that does not met a certain percentage of the bitcoins being transfered, for example a 0.5%. So if someone with a lot of bitcoins wants to transfer them the miners can decide not to unless that person pays a fee that pleases them.

I'm still thinking about this, but I'm sceptical that there will be a functional outcome here. Right now, about 50% (something like that, exact figure please) of the money supply is being issued per annun to pay for security. If you want the ratio between the size of potential attack rewards and the cost of an attack to remain constant, you would need a 50% (again approximate) annual wealth tax to achieve this. If this doesn't scream unsustainable, then I don't know what does. How could anyone rationalize setting up a system like this? It is an incredibly inefficient use of resources.

And last, this has been discussed already. Why keep opening threads discussing the same again and again?

It seems pretty important that users and developers are aware of the security situation. Could you please help by linking to the old threads.

[cunicula]

It's a mess, but if you take some time to think about it, it becomes clear that there is something wrong.

+1

[hugolp]

And last, this has been discussed already. Why keep opening threads discussing the same again and again?

Because every time it has been actually discussed, the result was Bitcoin security NOT functioning properly after minting, and NO consensus on a choice of block size limit being in place. One cannot reduce a two-variable problem (difficulty, fees) to a one-variable problem (fees) without any explanation as to why this can be done.

Thank you for pointing out that this thread is just a rehash of another thread. Why not continue the discussion there, instead of making everybody repeat the same arguments? Propagandizing or trolling are the only two options I can imagine.

Neglect block size limit, and see what you get. All but the cheapest miners will be pushed out of the market, reducing difficulty,

No, this if false. It is true that the inefficient miners will be pushed out of the market, but that does not mean reducing difficulty in a meaningful way. What will happen is that the more efficient miners will take the place of the less efficient miners, and probably a part of the less efficient miners will learn how to become more efficient.

The failure of what you are saying is that a extremely efficient miner with very little hash power, will take very long to get a block thus taking away very little fees from the other miners. Unless the efficient miner starts expanding and adquire more hasing power it would not represent a thread to the rest of the miners. Therefore you already have more hashing power appearing.

which in turn creates competition among the cheapest miners. It's a Tragedy of the Commons, with only the non-profit miners remaining in the end, and difficulty converging to zero.

See above.

I have had discussions in lengthy threads about it, and the bottom line was fairly clear. People arguing that current Bitcoin protocol would work don't even have a model they agree on! Some want to remove limits, some don't, some think cartels will do the job, some just neglect the Tragedy of the Commons... and each argument showed significant flaws. The only model that really gets backing has difficulty not only de-coupled from possible attacker size, but also converging to very low values unless limits on transactions push fees to arbitrary numbers.

I though the Bitcoin community was dogmatic. Now the problem is too many opinions? Some people are never happy.

See the following thread for the discussion, or the linked post in which I tried to summarize models. It's a mess, but if you take some time to think about it, it becomes clear that there is something wrong.

Thread on difficulty equilibrium after minting:
http://forum.bitcoin.org/index.php?topic=6284.0

Post in which I try to show that people who don't see a problem hardly ever agree on a model:
http://forum.bitcoin.org/index.php?topic=6284.msg111735#msg111735

I dont see how you have answered to my points.

[cunicula]

3. What you are proposing is already posible

The "solution" to the no-problem is already in place. Miners can decide which transactions they process and which transactions they dont. If they feel they are not being payed enough they can decide to not process transactions with a fee that does not met a certain percentage of the bitcoins being transfered, for example a 0.5%. So if someone with a lot of bitcoins wants to transfer them the miners can decide not to unless that person pays a fee that pleases them.

Again, still thinking, but my intuition is that this would indeed occur (Edit: what would happen to difficulty in absolute terms is hard to say, it depends on USD-denominated prices. However, relative to its level prior to the end of minting difficulty will certainly go way, way down):

All but the cheapest miners will be pushed out of the market, reducing difficulty, which in turn creates competition among the cheapest miners. It's a Tragedy of the Commons, with only the non-profit miners remaining in the end, and difficulty converging to zero.

[hugolp]

I'm still thinking about this, but I'm sceptical that there will be a functional outcome here. Right now, about 50% (something like that, exact figure please) of the money supply is being issued per annun to pay for security. If you want the ratio between the size of potential attack rewards and the cost of an attack to remain constant, you would need a 50% (again approximate) annual wealth tax to achieve this. If this doesn't scream unsustainable, then I don't know what does. How could anyone rationalize setting up a system like this? It is an incredibly inefficient use of resources.

But again you are forgetting that the decline in Bitcoin generation does not go away suddenly. Its progressive over time and wont end until 2040. Miners have plenty of time and also the incentives to adapt.

What you are proposing is a fixed level of fees depending on a set of conditions that you think are the best. But how is this fixed system that you are proposing not more dangerous and less adaptable than a system where the miners, the people with the experience and the information, decide at every moment the best strategy? Imagine the percentage or rules you propose are too much or too little? That would hurt Bitcoin big time. Its better to leave the miners decide what they need to charge to be in business.

EDIT:

Again, still thinking, but my intuition is that this would indeed occur:

Already answered to him. See above. Competition does not work the way he is portraying it. According to him/her competition is a tragedy of the commons. If that were true, the human race would have disappeared long ago.

[Vandroiy]

The failure of what you are saying is that a extremely efficient miner with very little hash power, will take very long to get a block thus taking away very little fees from the other miners. Unless the efficient miner starts expanding and adquire more hasing power it would not represent a threat to the rest of the miners. Therefore you already have more hashing power appearing.

This makes no sense. TX fee price is determined by the most efficient miners -- all prices will cascade down if one appears that offers fees smaller than any other. Unless you assume all miners equally efficient, the more efficient ones can charge lower TX fees and thus drive the less efficient out CONTINUOUSLY. The process will repeat every time difficulty adapts. All miners reaching equal efficiency in a competitive environment doesn't seem likely either.

I need not go into your other points. An argument with one point broken is broken. I am tired of this, I again see no model coupling attacker size to network size and thus difficulty. The "argument", apart from being false, would apply to any network size. That does not seem reasonable at all.

Please, read a little more carefully before claiming things I can show false in one paragraph and absurd in one statement.

[cunicula]

I'm still thinking about this, but I'm sceptical that there will be a functional outcome here. Right now, about 50% (something like that, exact figure please) of the money supply is being issued per annun to pay for security. If you want the ratio between the size of potential attack rewards and the cost of an attack to remain constant, you would need a 50% (again approximate) annual wealth tax to achieve this. If this doesn't scream unsustainable, then I don't know what does. How could anyone rationalize setting up a system like this? It is an incredibly inefficient use of resources.

But again you are forgetting that the decline in Bitcoin generation does not go away suddenly. Its progressive over time and wont end until 2040. Miners have plenty of time and also the incentives to adapt.

What you are proposing is a fixed level of fees depending on a set of conditions that you think are the best. But how is this fixed system that you are proposing not more dangerous and less adaptable than a system where the miners, the people with the experience and the information, deciding at every moment the best strategy? Imagine the percentage or rules you propose are too much or too little? That would hurt Bitcoin big time. Its better to leave the miners decide what they need to charge to be in business.

Security is a canonical example of a public good. You need to have a serious analysis of the incentives to demand and supply security services before you entrust provision to the market. The normal case is that security is greatly under-supplied because no one in the user base has an incentive to purchase it. If you were to give one person a legal monopoly of mining, (i.e. centralize transaction processing), the incentive situation would improve considerably. Supply-side competition in this case is harmful, it leads to a competitive "race to the bottom" in which security is severely under-supplied.

The only possible (costless) way out of this quagmire is if security isn't really necessary after all. For example, if you can't make mischief with say 75% of hashing power, then the future collapse in security is not a problem. Views on this issue are conflicting and my professional expertise is in economic analysis not the workings of the algorithm.

[hugolp]

The failure of what you are saying is that a extremely efficient miner with very little hash power, will take very long to get a block thus taking away very little fees from the other miners. Unless the efficient miner starts expanding and adquire more hasing power it would not represent a threat to the rest of the miners. Therefore you already have more hashing power appearing.

This makes no sense. TX fee price is determined by the most efficient miner. Unless you assume all miners equally efficient, the more efficient ones can charge lower TX fees and thus drive the less efficient out. All miners reaching equal efficiency in a competitive environment doesn't seem likely either.

I need not go into your other points. An argument with one point broken is broken. I am tired of this, I again see no model coupling attacker size to network size and thus difficulty. The "argument", apart from being false, would apply to any network size. That does not seem reasonable at all.

Please, read a little more carefully before claiming things I can show false in two statements and absurd in one.

Efficient miner is different than a miner with a lot of hasing power. I can start mining with a very efficient system (consuming less energy for Hash than my competition) but still have very little processing power, and I would need time and money to get to their position, which in turns gives them the opportunity to react. Its called the market process.

Therefore being efficient does NOT imply getting a lot of fees. It will tend to go there with time, but a) the efficient miner needs to add more hashing power, thus avoiding the problem of reducing the hashing power, b) the others will react and try to become efficient as well.

You are basically saying that competition is a tragedy of the commons, which is basically nonsense.