Bribery: The Double Double Spend

[cunicula]

I'd rather wait until it becomes a problem. I disagree we need any hard forks. There are plenty of proposals that don't need that.

Like what?

Moreover, do you think big fixes like this will get easier if bitcoin grows? I expect the opposite.

[becoin]

What speed is the right speed for the Bitcoin network?

The simplest answer of course is "as high as needed". It must be a dynamic variable.

- needed just right to discourage attacker(s), no more.
- discouraging can be done only by attracting more honest miners.
- ad hoc attracting more honest miners can be done only by increasing incentives i.e. increasing the minimum txn fees.
- who will increase the minimum txn fee?
- if it is the mining community, can they abuse this power by launching false attacks just to increase their income?
- who will reject all transactions with insufficient txn fees?
- the process of evaluating network health must be autonomous.
- the process of evaluating network health must be closely linked to the process of defining "next block minimum txn fee" or something like that.

It is a question discussed last year. The question of dynamically defined minimum txn fees is a central question!

[Mike Hearn]

Like network assurance contracts.

A group of people or companies want network speeds to be higher. However, none wants to be the sucker who pays for all the others.

They broadcast an assurance contract on a separate p2p network with a pledge from themselves. The contract is a transaction with a zero value output, ie, it exists purely for fees and to incentivise mining. If others find the size of the incentive acceptable they also submit pledges in whatever amount they prefer. Once enough pledges are broadcast they are automatically combined and submitted to the main Bitcoin p2p network. Miners then race to find a block including this fee paying transaction. Once new blocks are broadcast the process can repeat. Alternatively nLockTime can be used to set up a few contracts ahead of the current chain head block.

You might say, perhaps miners would include only the incentive transaction and not any others. But with good software including other transactions, even free transactions, is so easy that miners should do it anyway for the overall health of the network (they do today, after all).

Assurance contracts are a well studied method to incentivise the creation of public goods. There are some useful economics papers on the topic if you want to read the literature.

[becoin]

A group of people or companies want network speeds to be higher. However, none wants to be the sucker who pays for all the others.

They broadcast an assurance contract on a separate p2p network with a pledge from themselves.

This is not a wise approach. What you generally suggest is a second network to support bitcoin network, a network of insurers. This would be bitcoin level 2 network. And who will insure the insurers? May be a group of people or companies on a third p2p network, a bitcoin level 3 network?

[cunicula]

Assurance contracts are a well studied method to incentivise the creation of public goods. There are some useful economics papers on the topic if you want to read the literature.

Okay, you prefer a perpetual waste of resources to a hard fork. That is ridiculous in its own right, but worse yet it is not likely to work. You should read the work of Elinor Ostrom. She tries to distinguish situations where private provision of public goods works well from situations where private provision of public goods works poorly. There are many situations where it works poorly. Bitcoin will prove to be such a case. (anonymous participants, impossible to sanction free-riders, large number of participants) all these are no-nos.

Please provide an example (comparable to bitcoin) where an assurance contract has functioned effectively.

[cunicula]

minimum txn fees is a central question!

Minimum txn fees are a hard fork (and a not particularly useful one)

[kjj]

A group of people or companies want network speeds to be higher. However, none wants to be the sucker who pays for all the others.

They broadcast an assurance contract on a separate p2p network with a pledge from themselves.

This is not a wise approach. What you generally suggest is a second network to support bitcoin network, a network of insurers. This would be bitcoin level 2 network. And who will insure the insurers? May be a group of people or companies on a third p2p network, a bitcoin level 3 network?

I don't think you understood what he is proposing.  He is saying that if the current network speed is acceptable to most people, but a few people would like it higher for their own reasons (whatever those reasons are), they have a mechanism to pay for that extra speed without changing the system.

[becoin]

I don't think you understood what he is proposing.  He is saying that if the current network speed is acceptable to most people, but a few people would like it higher for their own reasons (whatever those reasons are), they have a mechanism to pay for that extra speed without changing the system.

I don't think this makes much sense. It is just the opposite what OP is about. It is the same as - if the current network speed is acceptable to most people, but a few people would like it lower for their own reasons (whatever those reasons are), they have a mechanism to pay for that extra low speed without changing the system - and then launch an attack!

So I suggest first group will pay / bribe in BTC while the second one will pay / bribe in USD or EUR?

[kjj]

I don't think you understood what he is proposing.  He is saying that if the current network speed is acceptable to most people, but a few people would like it higher for their own reasons (whatever those reasons are), they have a mechanism to pay for that extra speed without changing the system.

I don't think this makes much sense. It is just the opposite what OP is about. It is the same as - if the current network speed is acceptable to most people, but a few people would like it lower for their own reasons (whatever those reasons are), they have a mechanism to pay for that extra low speed without changing the system - and then launch an attack!

So I suggest first group will pay / bribe in BTC while the second one will pay / bribe in USD or EUR?

What mechanism would they use to lower the network speed?  It is very easy to add mining incentives, but impossible to reduce them.  There are no anti-fees that you can put in a transaction to lower the reward for mining.

The OP proposes a scheme that he thinks will break the system, but he hasn't ever done a proper accounting of the costs, risk and rewards for all of the parties at each step along the way.  His conclusion is based on shitty bookkeeping, and a desire to find a way, any way, for his prejudgment about proof-of-work to be right.

[becoin]

What mechanism would they use to lower the network speed?  It is very easy to add mining incentives, but impossible to reduce them. 

Firstly, it is not that easy to ad mining incentives if you pay in BTC. And secondly, it is not that difficult to reduce network speed if you have unlimited access to USD or EUR. You raise the network difficulty through cheap subsidized ASICS and when independent miners gave up the entire network is yours. After all ASICS manufacturers are paying dollars to produce them.

[kjj]

What mechanism would they use to lower the network speed?  It is very easy to add mining incentives, but impossible to reduce them. 

Firstly, it is not that easy to ad mining incentives if you pay in BTC. And secondly, it is not that difficult to reduce network speed if you have unlimited access to USD or EUR. You raise the network difficulty through cheap subsidized ASICS and when independent miners gave up the entire network is yours. After all ASICS manufacturers are paying dollars to produce them.

Yawn.  If you want to destroy bitcoin, it would be cheaper to round up all of the miners and shoot them.  If you want to reduce the difficulty, but still keep a functioning system, what are your options?

My apologies, I had taken the total destruction option to be an assumed, but uninteresting, path.

[becoin]

If you want to reduce the difficulty, but still keep a functioning system, what are your options?

Already explained on my previous post. You kill competition among miners through subsidized ASIC and you have what you want.

Bitcoin community is too much focuced on open source software but open source hardware is equally important for the bitcoin network. What about donating to a bitcoin specific ASIC project on OpenCores.org?

[kjj]

If you want to reduce the difficulty, but still keep a functioning system, what are your options?

Already explained on my previous post. You kill competition among miners through subsidized ASIC and you have what you want.

Bitcoin community is too much focuced on open source software but open source hardware is equally important for the bitcoin network. What about donating to a bitcoin specific ASIC project on OpenCores.org?

Not explained at all, merely stated.  Adding subsidized ASICs to the mix will increase difficulty, at least until the point that everyone stops using the system entirely, at which point the only person left mining is the subsidizer, and he can set the difficulty to whatever low value he wants, but, and this part is critical, but no one cares because no one else is using it.

As for opencores, go for it.  I'm totally in favor of more designs and cheaper chips.  Considering that we are (probably, heh) going to go from zero publicly available ASIC designs to at least three in the next 12 months, I wouldn't say that open hardware is critical here, but it is always desired and appreciated.

[becoin]

at least until the point that everyone stops using the system entirely, at which point the only person left mining is the subsidizer, and he can set the difficulty to whatever low value he wants, but, and this part is critical, but no one cares because no one else is using it.

By everyone and no one you probably mean everyone and no one among miners?!

[kjj]

at least until the point that everyone stops using the system entirely, at which point the only person left mining is the subsidizer, and he can set the difficulty to whatever low value he wants, but, and this part is critical, but no one cares because no one else is using it.

By everyone and no one you probably mean everyone and no one among miners?!

No, I meant exactly what I said, "no one else".  Would you use bitcoin if, for example, the Federal Reserve Bank was the only miner?

[becoin]

No, I meant exactly what I said, "no one else".  Would you use bitcoin if, for example, the Federal Reserve Bank was the only miner?

Exactly, kjj. This is not only the critical part but the final part because this will be the end of bitcoin. Attackers achieve their goal - bitcoin is crashed and current monetary monopoly stays intact!

[kjj]

No, I meant exactly what I said, "no one else".  Would you use bitcoin if, for example, the Federal Reserve Bank was the only miner?

Exactly, kjj. This is not only the critical part but the final part because this will be the end of bitcoin. Attackers achieve their goal - bitcoin is crashed and current monetary monopoly stays intact!

Again, there are easier was to do that, and cheaper too.  That's why I say it is uninteresting, and why I had assumed that we were talking about something entirely different.

[MoonShadow]

Not explained at all, merely stated.  Adding subsidized ASICs to the mix will increase difficulty, at least until the point that everyone stops using the system entirely, at which point the only person left mining is the subsidizer, and he can set the difficulty to whatever low value he wants, but, and this part is critical, but no one cares because no one else is using it.

I can't see that subsidizing ASICs is going to run enough miners out of business.  Many don't, and never did, need to be profitable in any practical sense.  Many of the early full time miners were just disiplacing the electric resistive heating for their flat with as much mining heat as they could manage; simply displacing one electric heat source for another, with the potential of gaining bitcoins in the process.  Even if pool miners could be run out of business (something that I question for reasons beyond the above situation) the long term reduction of difficulty simply makes mining more attractive for those who can still justify it.

[Mike Hearn]

Okay, you prefer a perpetual waste of resources to a hard fork. That is ridiculous in its own right, but worse yet it is not likely to work. You should read the work of Elinor Ostrom. She tries to distinguish situations where private provision of public goods works well from situations where private provision of public goods works poorly.

Alright, I'll check out her work. I'm not sure any previous situation is comparable to Bitcoin, so we'll have to wait and see how well it works in practice.

There are many situations where it works poorly. Bitcoin will prove to be such a case. (anonymous participants, impossible to sanction free-riders, large number of participants) all these are no-nos.

"Impossible to sanction free riders" is pretty much the definition of a public good. The point is you don't need to sanction them. The good gets created anyway by the people who care enough about it that they want it for themselves.

Please provide an example (comparable to bitcoin) where an assurance contract has functioned effectively.

You first - provide an example comparable to Bitcoin where it didn't.

[cunicula]

You first - provide an example comparable to Bitcoin where it didn't.

This is very difficult. Assurance contracts comparable to bitcoin are not observed because the idea is obviously unworkable. In most cases, failure would be anticipated and the experiment would never be tried.

Are you familiar with the noncooperative game theory on these issues (just the basic issue not assurance contracts)?

The most directly relevant paper is "System Reliability and Free Riding" by Hal Varian (Chief Economist at Google)
http://ns2.datacontact.dc.hu/~mfelegyhazi/courses/BMEVIHIAV15/readings/05_Varian2004system-reliability-free-riding.pdf

Abstract:
System reliability often depends on the effort of many individuals, making reli-
ability a public good. It is well-known that purely voluntary provision of public
goods may result in a free rider problem: individuals may tend to shirk, resulting
in an inffecient level of the public good. How much effort each individual exerts will depend on his own benefits
and costs, the efforts exerted by the other individuals, and the technology that
relates individual effort to outcomes. In the context of system reliability, we
can distinguish three prototypical cases.

The relevant prototypical case for bitcoin is the "sum of effort case." In this case, reliability is determined by the sum of efforts of all users. The nash equilibrium is that the user who benefits the most from system contributes 100% of the effort to maintaining system reliability. Every single other user contributes zero. This is also true if there is an attacker trying to destroy the network. The attacker wins if he benefits more from destroying the network than the highest valuation user benefits from saving it. The total value of the network is irrelevant. The highest valuation user determines aggregate network security.

I will think about the economics of organizing an assurance contract. I'm pretty sure that the contract's viability will depend on the distribution of user valuations. The more equal these are, the more effective the assurance contract would be. I will think more about it. However, let me reiterate that even if an assurance contract is viable in some cases it will remain a grossly inferior substitute to a securely designed system.