Taking Down Bitcoin

[Forp]

Even if you manage to subvert 99% of the mining power to another chain, there is absolutely no way in hell you will be able to keep that a secret. Miners are going to notice.

In the logic of the attack I would not mind when they notice. Since I have 99% percent of mining power I do not care what the remaining 1% of the mining power does.

I suppose we have a misunderstanding with regard to the phrase "subvert 99% of the mining power" as it was used in an earlier posting. I understand it as "I own 99% of the mining power". It looks like you understand it as "99% of the miners are in my pool and they might also leave my pool".

I agree completely: If "my" 99% of mining power just is derived from miners in "my" pool - the miners might notice and might leave my pool. And still, even in this situation: Would they do so if they realize that their payoffs are larger in my version of the algorithm? They are getting 51 BTC (or 500 BTC) in my version. Is it tempting to move from a 500 BTC per block operation to a 50 BTC per block operation? ;-)

[Etlase2]

I suppose we have a misunderstanding with regard to the phrase "subvert 99% of the mining power" as it was used in an earlier posting. I understand it as "I own 99% of the mining power". It looks like you understand it as "99% of the miners are in my pool and they might also leave my pool".

Yeah well that's probably because "subvert" doesn't mean anything close to "own." Now you only need 99x the current mining power to pull off this attack.

I agree completely: If "my" 99% of mining power just is derived from miners in "my" pool - the miners might notice and might leave my pool. And still, even in this situation: Would they do so if they realize that their payoffs are larger in my version of the algorithm? They are getting 51 BTC (or 500 BTC) in my version. Is it tempting to move from a 500 BTC per block operation to a 50 BTC per block operation? ;-)

The payoffs are 0 when nobody accepts those BTC. This is getting silly.

[Explodicle]

Given that the network is ~11 TH, you'll need 12TH to over power it, equivalent to 480 Mini-Rigs. At $16000 each, that's $7,680,000

So as of this writing, an attack on the network would cost as little as $7,680,000

This is assuming you could get 480 MiniRigs delivered (which BFL wont/cant do), and do it before anyone else gets to power theirs up which would raise the "legit" network hashing power.

Plus electricity during the attack, plus manpower to set up all those rigs and defend against any retaliations.

[Forp]

The payoffs are 0 when nobody accepts those BTC.

Why would nobody accept those BTC?

The new block chain honors all BTCs mined in the old block chain. However, it offers more attractive conditions to every miner which adopts the new block chain. Therefore there is a clear incentive to switch to the new block chain. Even worse: If you stay in the old chain, under the assumptions of the attack made above, the old chain will lose 99% of its hash power and will be V-E-R-Y slow.

I agree, that this will not happen in reality, since (1) the assumptions on which this attack is based are unrealistic and (2) miners will not base their decisions only on maximizing their profit, but also on a social consensus on how Bitcoin should work. This is good news.

The thought experiments of  a theoretician often sound silly from a practical point of view :-).

 

[Gavin Andresen]

Miners won't switch to a 51% chain if it means they can't cash out the coins they're creating!

The attack we're talking about is "51% attacker refuses to include anybody else's transactions in their blocks."  And newly generated coins are useless to miners if they can't get transactions that spend them into the block chain....

[Etlase2]

Why would nobody accept those BTC?

Ok, "nobody" is an overstatement. However, say 1% of businesses love the new chain and decide to switch over. Now you can spend your coins mined on the fake chain somewhere. Not in very many places though, and you will have to go through all the massive adoption pains that bitcoin currently suffers from. So perhaps you'll get a few extra coins for your work (then again, you won't because you control 99x the hash power of the original chain), but they will be worth, as a whole, far, far less as not many people will want them. Not to mention everybody with existing coins will be able to double spend them all on the new network and basically bring the value down to nothing.

The new block chain honors all BTCs mined in the old block chain. However, it offers more attractive conditions to every miner which adopts the new block chain. Therefore there is a clear incentive to switch to the new block chain.

No, there isn't, and I'm not going to keep rehashing this.

the old chain will lose 99% of its hash power and will be V-E-R-Y slow.

This can be an attack in and of itself. Probably only likely if someone finds a vulnerability in SHA256 and keeps it to themself. In which case, bitcoin devs can fork it themselves to use a new hash algorithm.

[Sergio_Demian_Lerner]

The best way an attacker can take profit from Bitcoin is this:

1. Study the source code.
2. Find a vulnerability
3. Sell the vulnerability in the black market for 100K USD (SolidCoin?)

The fastest way Bitcoin can be taken down is this:

1. Find a vulnerability in the black market.
2. Use it.

See the related thread https://bitcointalk.org/index.php?topic=79830.0

[Forp]

The attack we're talking about is "51% attacker refuses to include anybody else's transactions in their blocks."  And newly generated coins are useless to miners if they can't get transactions that spend them into the block chain....

I agree, that THIS attack is probably useless. I understood the OP differently (he only had the 51% part in his question).

Miners won't switch to a 51% chain if it means they can't cash out the coins they're creating!

Agree again. "They can't cash out the coins they're creating!". Where do you take this assumption from? The 51% miners would, in my version of the attack, switch to a chain, which is more profitable for every participant (eg. they add 1 BTC to every address) - and they would continue to include transactions nto the chain. So there is a monetary incentive to switch to the spawned chain. It may take a while, until all wallet users switch to the new chain, of course. So it is a matter of speed of adoption.

[Forp]

No, there isn't, and I'm not going to keep rehashing this.

Very strong argument, especially from a methodological point of view. 

[btccomm]

I'll take a stab at this, although these probably aren't as technical .
a few off the top of my head:

   - collusion among the miners, or a mining cartel forming
   - collusion among the programmers who maintain the satoshi client source code
   - using automated programs to flood the block chain with transactions. thereby making sync up non functional (it takes a good long while to download the block chain the 1st time now)
          this tree seems to be growing rapidly http://blockchain.info/tree/5416502
   - hack the exchanges, and steal BTC, with no criminal prosecutions
   - use bitcoin in a way to threaten national security, feds would shut it down, (like freeze all accounts at the exchanges, w cooperation of other countries of course)
   - bitcoin clients with vulnerabilities not being updated   exampleCVE-2012-2459 (i tried to update to 062 but it failed, so i didn't bother trying again)

[btccomm]

look at the # of transactions per block since the bitcoinica incident

http://blockchain.info/charts/n-transactions-per-block

[DeathAndTaxes]

The 51% miners would, in my version of the attack, switch to a chain, which is more profitable for every participant (eg. they add 1 BTC to every address) - and they would continue to include transactions nto the chain. So there is a monetary incentive to switch to the spawned chain. It may take a while, until all wallet users switch to the new chain, of course. So it is a matter of speed of adoption.

You must be one of those something for nothing kind of thinkers huh?

It is by definition IMPOSSIBLE for the new chain to be better for everyone.  Money is just an abstraction.  If every address gains 1 BTC you haven't increased the wealth merely the accounting system and existing holders of wealth will see their wealth reduced by dillution.  There is always a winner and loser when manipulating the financial system.

I mean by your logic I could make a chain which gives every address AND ever future address 11 quadrillion Bitcoins each.  Obviously that will be the best chain.  Don't you want to be a quadrillionaire?  Who doesn't man.

Your new chain would have no value.  1 BTC or 1 quadrillion BTC it only has value if someone wants it.  Nobody will.

[gmaxwell]

look at the # of transactions per block since the bitcoinica incident
http://blockchain.info/charts/n-transactions-per-block

Has nothing to do with bitcoinica and everything to do with a crazy gambling site which operates in a way that produces a ton of transactions.

[btccomm]

look at the # of transactions per block since the bitcoinica incident
http://blockchain.info/charts/n-transactions-per-block

Has nothing to do with bitcoinica and everything to do with a crazy gambling site which operates in a way that produces a ton of transactions.

yea, i have discovered satoshidice, even placed a bet myself.  but the result is the same, flooding the block chain with transactions.  that's a lot of overhead.

it looks like the # of transactions, and the # of transactions per block is growing geometrically
but total BTC in circulation is linear
and USD transaction volume & market capitalization is flat

[DeathAndTaxes]

The network has tx rules which make any volume which would be damaging to the network cost prohibitive.

Also 20K tx is hardly "flooding".  It is 0.2 tps.  For the network to even have the volume that Paypal has we are talking something on the order of 300x to 500x higher transaction volume.  If anything SD is merely testing the network and a relatively small and weak test.

[Forp]

You must be one of those something for nothing kind of thinkers huh?

  Oops. Since I am a non native speaker of English I might not completely catch the subtle context of this remark. But, yes, I am convinced that theory and thinking is fun and pays off (at least it did so far in my life).

Money is just an abstraction. 

If every address gains 1 BTC you haven't increased the wealth merely the accounting system and existing holders of wealth will see their wealth reduced by dillution.

I mean by your logic I could make a chain which gives every address AND ever future address 11 quadrillion Bitcoins each.

Exactly ! I think we now reached the same conceptual framework (the lack of which was the reason for a bit of misunderstanding in some postings).

My issue is the following:

Let's assume we model Bitcoin swarm behavior with game theory. In this case 1 BTC is the payoff and if I increase it to 2 BTC this is a more attractive payoff. If I program a Bitcoin miner to maximize his BTC profits, he will opt for the 2 BTC chain.

Of course, you are completely right that Bitcoin miners do not work that way. They are human beings and not just game theoretical agents. They know about dilusion and inflation and therefore will not readily switch to the 2 BTC or the 11 quadrillion BTC chain. Thus, we cannot model Bitcoin swarm behavior with game theory alone. We must look at human consensus and at economic aspects. So let us do so...

However...we now realize that this provides no final answer as well. Most states, share holders, companies, federal reserve banks know about dilution and inflation, nevertheless they, as human beings, still decide to accept a MINOR form of if...a bit LARGER form...since it worked out...again a bit LARGER...and...ooops...we have a financial crash.

So, wy should Bitcoin be more stable than the dollar, when the Bitcoin system inherently has the same problem (human beings making bad decisions) as the dollar.

I must admit that I get some pleasure from thinking about this theoretical issue and trying to find an answer which Bitcoin mining cannot compensate. I am interested whether this model is correct or my line of reasoning has a flaw. I am interested which mathematical / behavioral mechanism could eventually provide for some ultimate stability in Bitcoin (which, as mentioned above, it does not have in it's current form). Compare with the dollar: It HAD a gold standard - until the president / fed abolished it. What would prevent miners to abolish the current limitation system in BTC...and opt for a 51 BTC per block chain (or, rather, 14 quadrillion in the end).

[Serenata]

DoS attacks against large pools wouldn't work IMO. Most miners mine for alternative pools and have fail-over scenarios. If LargePoolA does not respond then all the hash-rate goes towards SmallPoolB that works fine. Even if ALL pools go down, there is always P2Pool

I'm not so sure that making it illegal would be a good choice either. Take for example the banning of alcohol in the US many years ago, that effectively made alcohol more popular.

Even if exchanges go down, people would still be able to trade services and trade BTC, although in small quantities. But as more people get into it, the value of BTC will rise.

The value money has, is the value we give it. So the only way to destroy  Bitcoin IMO, would be for people to lose their interest in it.

[Explodicle]

Let's assume we model Bitcoin swarm behavior with game theory. In this case 1 BTC is the payoff and if I increase it to 2 BTC this is a more attractive payoff. If I program a Bitcoin miner to maximize his BTC profits, he will opt for the 2 BTC chain.

Of course, you are completely right that Bitcoin miners do not work that way. They are human beings and not just game theoretical agents. They know about dilusion and inflation and therefore will not readily switch to the 2 BTC or the 11 quadrillion BTC chain. Thus, we cannot model Bitcoin swarm behavior with game theory alone. We must look at human consensus and at economic aspects. So let us do so...

However...we now realize that this provides no final answer as well. Most states, share holders, companies, federal reserve banks know about dilution and inflation, nevertheless they, as human beings, still decide to accept a MINOR form of if...a bit LARGER form...since it worked out...again a bit LARGER...and...ooops...we have a financial crash.

So, wy should Bitcoin be more stable than the dollar, when the Bitcoin system inherently has the same problem (human beings making bad decisions) as the dollar.

I must admit that I get some pleasure from thinking about this theoretical issue and trying to find an answer which Bitcoin mining cannot compensate. I am interested whether this model is correct or my line of reasoning has a flaw. I am interested which mathematical / behavioral mechanism could eventually provide for some ultimate stability in Bitcoin (which, as mentioned above, it does not have in it's current form). Compare with the dollar: It HAD a gold standard - until the president / fed abolished it. What would prevent miners to abolish the current limitation system in BTC...and opt for a 51 BTC per block chain (or, rather, 14 quadrillion in the end).

Miners could "merge mine" both chains at once at no additional cost, assuming their only difference is block subsidy. Some alternative cryptocurrencies already have higher relative block subsidies.

During the P2SH dispute a lot of people got in their heads that miners determine the rules, but it's really more a matter of consensus between users. If some miners started trying to mess with the subsidy, all the clients would reject it and use the original blockchain automatically.

[DeathAndTaxes]

Money is just an abstraction. 

Exactly ! I think we now reached the same conceptual framework (the lack of which was the reason for a bit of misunderstanding in some postings).  My issue is the following:

Let's assume we model Bitcoin swarm behavior with game theory. In this case 1 BTC is the payoff and if I increase it to 2 BTC this is a more attractive payoff. If I program a Bitcoin miner to maximize his BTC profits, he will opt for the 2 BTC chain.

No it wouldn't.  Once again money is an abstraction.  It is possible the chain which pays out less BTC has more buying power.  If BTC 1.0 pays 50 BTC per block and BTC pays 2000 BTC that alone provides no measure of value (buying power).  If the original chain allows me to buy 1 Whooper combo meal w/ 1 BTC and the second chain take 200,000 BTC to buy the same combo meal it would be flawed thinking to switch chains simply because the "number is bigger".

They know about dilusion and inflation and therefore will not readily switch to the 2 BTC or the 11 quadrillion BTC chain. Thus, we cannot model Bitcoin swarm behavior with game theory alone. We must look at human consensus and at economic aspects. So let us do so...

No need your "game theory" answer is already invalid.   The goal of any venture is to maximize wealth not money.  Money is merely an accounting system for wealth.    If you want lots of money I can get you some Zimbabwe dollars.  You can be a trillionaire.  They don't buy anything but they have lots of zeros.  Likewise an alt-chain which simply increases the nominal number of BTC isn't going to increase the purchasing power of miners or users.

You can't create value or wealth via accounting games.  Money is merely arbitrary.  Satoshi could have chosen a block reward of 1 BTC or 1000 BTC.  All it would do is change the nominal amount of the block reward.  The value would remain the same.  With a block reward of 1 there would be 1/50th as many BTC in circulation and the exchange rate would be ~$250 per BTC.

[Skybuck]

1. Wait and be patient/have patience

2. Wait until miners traded all their bitcoins for real dollars.

3. Wait until 50 goes to 25 goes to 12.5 goes to etc until it reaches such a low level that it's not worth it anymore.

4. Wait until miners decide that the transaction fee is not enough for them to continue operation and start attacking the network in hopes of higher payouts by crime.

Thus I do believe that in the future miners will turn against the bitcoin system. So it will auto self-destruct All that it requires is some patience lol.