Bitcoin War: The First Real Threat to Bitcoin?

[glitch003]

https://www.privateinternetaccess.com/blog/2012/03/bitcoin-war-the-first-real-threat-to-bitcoin/

What do you guys think?  I am in no way affiliated with this article except that I just found it on the bitcoin reddit.

[glitch003]

What do you guys think?

I think there is already a massive thread about this in the mining sub forum. 

Whoops, didn't see it.  Oh well.

[theymos]

My guess is that they just don't want to deal with processing transactions for some reason. There's normally almost no cost to adding transactions to blocks, though their setup might somehow increase the costs for them. Miners will stop doing this once people start including more transaction fees, and especially once the subsidy decreases.

I'm not concerned about it. Even just one transaction-containing block every few hours would be enough for the network to operate reasonably well.

[hazek]

The thread where this is discussed for anyone who is interested: https://bitcointalk.org/index.php?topic=67634.0

[cunicula]

The most important thing about this in my opinion is that it illustrates how easy it is to monopolize the bitcoin network.

If you can get 15%, well 51% is only about 3.4 times as hard.

The question should be not if a 51% monopoly occur, but what will the monopolist decide to do once he accumulates 51%.

This could happen quite suddenly. Suppose for example that the btc price crashes again and GPU mining becomes unprofitable even with cheap electricity. Most miners will exit if they can't recoup electricity costs. Most mining is still overwhelmingly GPU mining. If this guy is a botnet, electricity is irrelevant to him. If he is a botnet, a price crash could send him directly to 51%. If investors think that 51% control leads to the destruction of bitcoin (I don't but many do), then a price crash would be a self-fulfilling prophecy.

Under a self-fulfilling prophecy, the bitcoin price has multiple equilibria. Price can be high with a diffused hashing control equilibrium and low with concentrated hashing control equilibrium. As a speculator, you can earn big bucks by investing money to switch between equilibria. All you need to do to profit is put a large chunk of money in shorts. Others will worry that heavy shorts could get the price drop and mining exit rolling, giving mystery miner more control. If you believe that most of the market also thinks 51% control is destructive, then it is individually rational to come on board and also sell. The speculator profits all the way down until the new equilibria is reached and bitcoin lies in ruin. The mystery miner and the speculator could be one and the same person. Due to this self-fulfilling prophecy effect, the actual threshold which leads to 51% control is much less than 51%, it could potentially be 20 or 30% though it is very hard to guess because it depends on perceptions about the price consequences of 51% attack.

This is not FUD, just textbook international finance. If you think those are one and the same, I'm selling tinfoil hats on bitmit.

Right now with block reward at 50, an attack is as costly as it will ever be. Around December, monopoly gets easier, and the plan for the future is to make it even easier.

Shouldn't the development team make some statement about how they plan to address this pressing issue?

[cunicula]

Now is the time for these alternative ideas to prove they are better than what we have. Bitcoin is working now, with all it's terrible flaws. It should be easy enough for a better block chain to take over. Especially when the monopoly comes.

The status quo network size does matter. Inferior technologies can persist for long periods because they have an existing user base.

If you doubt this, apply your logic to bitcoin. If the status quo network size doesn't matter for competitivenesss, it should also be easy for bitcoin to take over the banking system. good luck with that! In fact, network size is so important that it is hard to get banks to adopt even simple technologies like ACH payments.

[runeks]

Right now with block reward at 50, an attack is as costly as it will ever be. Around December, monopoly gets easier, and the plan for the future is to make it even easier.

This is not entirely true. Or at least there's another aspect to it. You're right that a lowered block subsidy will probably lower difficulty as well (if the price stays the same), but the profit for the attacker will also decrease by 50%. Right now an attacker has to buy more hardware to mount an attack, but he also makes more money because he gets 50 BTC per block instead of just 25 BTC, thus increasing the income from the attack. So both expenses and income is higher with a large subsidy.

[cunicula]

Right now with block reward at 50, an attack is as costly as it will ever be. Around December, monopoly gets easier, and the plan for the future is to make it even easier.

This is not entirely true. Or at least there's another aspect to it. You're right that a lowered block subsidy will probably lower difficulty as well (if the price stays the same), but the profit for the attacker will also decrease by 50%. Right now an attacker has to buy more hardware to mount an attack, but he also makes more money because he gets 50 BTC per block instead of just 25 BTC, thus increasing the income from the attack. So both expenses and income is higher with a large subsidy.

a) If the attack incentive is monopolization of currency generation, then you are right.

b) If the attack incentive is a speculative attack as discussed in the previous post, then you are wrong. In this case attack incentives are proportional to liquidity in the market for shorts. This is likely approximately proportional to the total market cap of bitcoin in existence. That is how short can you go on Bitcoinica or through a more complex mechanism.

c) If the attack incentive is monopolization of txn fees and the imposition of an onerous fee, then again you are wrong. In this case attack incentives scale with the volume of demand for bitcion txns. Again this is likely approximately proportional to market cap.
 
It is up to the attacker to pick his attack plan, but that just makes things even worse. As the block reward goes down, attack plan (a) becomes less attractive and a combination of attack plans (b) and (c) become more attractive.

[proudhon]

From the article linked in the OP:

When computing 1 transaction per block versus 100, you can imagine the latter would be more costly than the former.

Wait, what?  How does it take any more hashing power to validate a block with 100 transaction as opposed to a block with only 1 transaction? 

[kjlimo]

From the article linked in the OP:

When computing 1 transaction per block versus 100, you can imagine the latter would be more costly than the former.

Wait, what?  How does it take any more hashing power to validate a block with 100 transaction as opposed to a block with only 1 transaction? 

I suggest reading the other thread.  I'm on page one and found this post helpful relative to your question:

Not strange.  Someone who only cares about revenue can save overhead by mining blank blocks. Pretty lame but there is little advantage of including transactions, and none for free transactions. Likely running a modified version of bitcoind who's get work consists of just the coinbase.

Personally I wish he would join p2pool.

Seems like it must take some coding or power to include all of the transactions into the hashing as opposed to just grabbing the coinbase and hashing that...  I'm not an expert which is why I posted DeathAndTaxes's post here.

[Gavin Andresen]

If you can get 15%, well 51% is only about 3.4 times as hard.

Check your math-- to get 51% you'll need to dedicate six times as much hardware to hashing as it takes to get 15%:

The formula for what fraction of existing hashing power you need add to get X% of the network is:

Code:

H = X / (1-X)

To get 15%, you need to add 17.6%.  E.g. if existing hashing power is 100, then you add 17.6 more and your fraction is (17.6/117.6) = 0.15.

To get 51%, you need to add 104%.  E.g. if existing hashing power is 100, then you add 104, then your fraction is (104/204) = 0.51.

104 is six times as much as 17.6.

[theymos]

Wait, what?  How does it take any more hashing power to validate a block with 100 transaction as opposed to a block with only 1 transaction? 

There's a tiny one-time cost to including a transaction into the block you're mining, since you need to recalculate the Merkle tree. There's absolutely no extra cost once you've added the transaction to the block; hashing 10,000 transactions isn't any slower than hashing 1 transaction once you've built the Merkle tree.

[cypherdoc]

If you can get 15%, well 51% is only about 3.4 times as hard.

Check your math-- to get 51% you'll need to dedicate six times as much hardware to hashing as it takes to get 15%:

The formula for what fraction of existing hashing power you need add to get X% of the network is:

Code:

H = X / (1-X)

To get 15%, you need to add 17.6%.  E.g. if existing hashing power is 100, then you add 17.6 more and your fraction is (17.6/117.6) = 0.15.

To get 51%, you need to add 104%.  E.g. if existing hashing power is 100, then you add 104, then your fraction is (104/204) = 0.51.

104 is six times as much as 17.6.

and this is the guy who's advocating a complete re-haul of the current system?

[cbeast]

If you can get 15%, well 51% is only about 3.4 times as hard.

Check your math-- to get 51% you'll need to dedicate six times as much hardware to hashing as it takes to get 15%:

The formula for what fraction of existing hashing power you need add to get X% of the network is:

Code:

H = X / (1-X)

To get 15%, you need to add 17.6%.  E.g. if existing hashing power is 100, then you add 17.6 more and your fraction is (17.6/117.6) = 0.15.

To get 51%, you need to add 104%.  E.g. if existing hashing power is 100, then you add 104, then your fraction is (104/204) = 0.51.

104 is six times as much as 17.6.

and this is the guy who's advocating a complete re-haul of the current system?

Adding more hashing power also increases difficulty. That is obvious.

[runeks]

Wait, what?  How does it take any more hashing power to validate a block with 100 transaction as opposed to a block with only 1 transaction? 

There's a tiny one-time cost to including a transaction into the block you're mining, since you need to recalculate the Merkle tree. There's absolutely no extra cost once you've added the transaction to the block; hashing 10,000 transactions isn't any slower than hashing 1 transaction once you've built the Merkle tree.

That's not really true. What I mean is, when you do have the transactions, it's true, adding them to the block you're mining is nothing. But finding out which transactions to include by maintaining a local copy of the block chain requires quite a lot more bandwidth than not doing so. Ie., say a botnet operator was mining, the initial block chain download of 1GB isn't exactly free (nor quick). It took my computer 24 hours the last time I did it.

To put it in perspective, using the most recent block as an example (block 171800 right now), keeping a local copy of the block chain and figuring out which transactions to put in the blocks you're mining earns you an extra profit of 0.022%. It seems obvious to me that the extra effort might not be worth this in some cases.

[Steve]

Under a self-fulfilling prophecy, the bitcoin price has multiple equilibria. Price can be high with a diffused hashing control equilibrium and low with concentrated hashing control equilibrium. As a speculator, you can earn big bucks by investing money to switch between equilibria. All you need to do to profit is put a large chunk of money in shorts. Others will worry that heavy shorts could get the price drop and mining exit rolling, giving mystery miner more control. If you believe that most of the market also thinks 51% control is destructive, then it is individually rational to come on board and also sell. The speculator profits all the way down until the new equilibria is reached and bitcoin lies in ruin. The mystery miner and the speculator could be one and the same person. Due to this self-fulfilling prophecy effect, the actual threshold which leads to 51% control is much less than 51%, it could potentially be 20 or 30% though it is very hard to guess because it depends on perceptions about the price consequences of 51% attack.

This is not FUD, just textbook international finance. If you think those are one and the same, I'm selling tinfoil hats on bitmit.

This is all quite plausible except for one sticking point…how would one go about acquiring the bitcoins necessary to take such a large and profitable short position?  Seems to me that it would drive the price dramatically higher, which would bring more miners in.  Shorting activity might then drive some miners back out, but I'd suspect some of those miners you attracted when acquiring bitcoins would remain in anticipation that the price would eventually recover.  Naked shorting in any large quantity or on a long time frame in bitcoin is practically impossible.

[cbeast]

Under a self-fulfilling prophecy, the bitcoin price has multiple equilibria. Price can be high with a diffused hashing control equilibrium and low with concentrated hashing control equilibrium. As a speculator, you can earn big bucks by investing money to switch between equilibria. All you need to do to profit is put a large chunk of money in shorts. Others will worry that heavy shorts could get the price drop and mining exit rolling, giving mystery miner more control. If you believe that most of the market also thinks 51% control is destructive, then it is individually rational to come on board and also sell. The speculator profits all the way down until the new equilibria is reached and bitcoin lies in ruin. The mystery miner and the speculator could be one and the same person. Due to this self-fulfilling prophecy effect, the actual threshold which leads to 51% control is much less than 51%, it could potentially be 20 or 30% though it is very hard to guess because it depends on perceptions about the price consequences of 51% attack.

This is not FUD, just textbook international finance. If you think those are one and the same, I'm selling tinfoil hats on bitmit.

This is all quite plausible except for one sticking point…how would one go about acquiring the bitcoins necessary to take such a large and profitable short position?  Seems to me that it would drive the price dramatically higher, which would bring more miners in.  Shorting activity might then drive some miners back out, but I'd suspect some of those miners you attracted when acquiring bitcoins would remain in anticipation that the price would eventually recover.  Naked shorting in any large quantity or on a long time frame in bitcoin is practically impossible.

How do you naked short sell currency? It's called counterfeiting, and that is impossible with Bitcoin.

[cunicula]

If you can get 15%, well 51% is only about 3.4 times as hard.

Check your math-- to get 51% you'll need to dedicate six times as much hardware to hashing as it takes to get 15%:

The formula for what fraction of existing hashing power you need add to get X% of the network is:

Code:

H = X / (1-X)

To get 15%, you need to add 17.6%.  E.g. if existing hashing power is 100, then you add 17.6 more and your fraction is (17.6/117.6) = 0.15.

To get 51%, you need to add 104%.  E.g. if existing hashing power is 100, then you add 104, then your fraction is (104/204) = 0.51.

104 is six times as much as 17.6.

My math is sloppy, true. However, Gavin's calcuation is also misleading. He assumes that none of the other miners drop out if difficulty increases. By contrast, I'm implicitly assuming that the current difficulty level is such that a large share of the other miners are on the verge of becoming unprofitable AND that miners drop out when they become unprofitable.

According to Gavin's formula if a is the starting hashing power of the attacker, b is the starting hashing power of the rest of the network, a would like to increase his share by y-fold, then he actually needs to increase his hashing power by x-fold according to the following formula.
 
x=by/(a+b-ay)
For y=3.4, a=15, and b=85 we get x=5.89 like Gavin says.

By contrast, I'm assuming that a large proportion of the miners in b are GPU miners at the threshold of becoming unprofitable. [For example, I'm in a high electricity cost location. I mined from June to August 2011, but no longer] I need at least 41% of the other miners to fit this criteria. If these miners drop out in response to a difficulty increase, then whenever a adds a unit of hashing power, a unit of hashing power will drop out of b. This response prevents total hashing power and difficulty from rising. If this is the case, we get the following formula.

x=y

So to increase our share by 3.4 fold, we just need to buy 3.4 times as much mining equipment.

The real answer lies somewhere in between 3.4 and 5.89. I think 3.4 is much closer to the real answer than 5.89, but you would need to do some empirical data analysis (rather than theoretical) to estimate it.

[cunicula]

This is all quite plausible except for one sticking point…how would one go about acquiring the bitcoins necessary to take such a large and profitable short position?  Seems to me that it would drive the price dramatically higher, which would bring more miners in.  Shorting activity might then drive some miners back out, but I'd suspect some of those miners you attracted when acquiring bitcoins would remain in anticipation that the price would eventually recover.  Naked shorting in any large quantity or on a long time frame in bitcoin is practically impossible.

Your right, that is the main sticking point.

You are misunderstanding shorting, however. To short, you borrow bitcoin from someone else and then sell it for dollars. Selling the BTC for dollars places downward pressure on price, not upward. If the price has fallen when you payback your loan, then you earn a profit (hopefully more than the interest on the loan).

The current place to borrow Bitcion is Bitcoinica. I'm sure there's not enough liquidity there now to make this work (e.g. the starfish would come up too soon). However, if the financial market develops further...

[runeks]

Under a self-fulfilling prophecy, the bitcoin price has multiple equilibria. Price can be high with a diffused hashing control equilibrium and low with concentrated hashing control equilibrium. As a speculator, you can earn big bucks by investing money to switch between equilibria. All you need to do to profit is put a large chunk of money in shorts. Others will worry that heavy shorts could get the price drop and mining exit rolling, giving mystery miner more control. If you believe that most of the market also thinks 51% control is destructive, then it is individually rational to come on board and also sell. The speculator profits all the way down until the new equilibria is reached and bitcoin lies in ruin. The mystery miner and the speculator could be one and the same person. Due to this self-fulfilling prophecy effect, the actual threshold which leads to 51% control is much less than 51%, it could potentially be 20 or 30% though it is very hard to guess because it depends on perceptions about the price consequences of 51% attack.

This is not FUD, just textbook international finance. If you think those are one and the same, I'm selling tinfoil hats on bitmit.

This is all quite plausible except for one sticking point…how would one go about acquiring the bitcoins necessary to take such a large and profitable short position?  Seems to me that it would drive the price dramatically higher, which would bring more miners in.  Shorting activity might then drive some miners back out, but I'd suspect some of those miners you attracted when acquiring bitcoins would remain in anticipation that the price would eventually recover.  Naked shorting in any large quantity or on a long time frame in bitcoin is practically impossible.

How do you naked short sell currency? It's called counterfeiting, and that is impossible with Bitcoin.

Since most people quote the bitcoin price as the price on Mt. Gox, naked short selling of bitcoins is entirely possible. It only requires that Mt. Gox sells bitcoins that it doesn't have. This can continue until users withdraw more bitcoins from their Mt. Gox account than Mt. Gox actually has. It's the same as fractional reserve banking. If users accept a deposit in a Mt. Gox account (or a Mt. Gox redeemable code) as bitcoins, Mt. Gox can, in effect, increase the supply of bitcoins, thus leading to a drop in the purchasing power of each bitcoin.