How much would it cost right now to attain 50% of the network?

[portpho]

Let's say you just keep buying FPGAs or GPUs until you have over 50% of the network and can dominate it.

How many millions would it cost right now?

I'm curious just how rich you'd have to be to take over bitcoin.

Could you do it with $10 million as a small business owner?

Could a hedge fund CEO do it with $100 million?

Could Zuckerberg do it with $1 billion?

Could the NSA or DOD do it with $10 billion?

How about the Chinese government?

Basically, how many people in the world have the power to fuck over bitcoin if they wanted to?

[evoorhees]

I'm still a noob... but my understanding is that having 50% of the network processing power is a necessary but not sufficient condition to "take over" bitcoin.

If such a large player were to come in and do that, wouldn't all the devotees who peer at the code incessantly see quickly that it'd been tampered with and the community at large would learn of this in short order. Then people would abandon the "new" taken over network and revert to the old "correct" version and the millions or billions spent in pursuit of takeover would become a complete waste.

Do I have it correct or is noob tattooed on my face?

[JorgePasada]

I'm still a noob... but my understanding is that having 50% of the network processing power is a necessary but not sufficient condition to "take over" bitcoin.

If such a large player were to come in and do that, wouldn't all the devotees who peer at the code incessantly see quickly that it'd been tampered with and the community at large would learn of this in short order. Then people would abandon the "new" taken over network and revert to the old "correct" version and the millions or billions spent in pursuit of takeover would become a complete waste.

Do I have it correct or is noob tattooed on my face?

False, abandoning it wouldn't be the way to go. Neither would 'changing the code' in any substantive manner at least (this would be akin to printing more fiat currency)

The network would auto adjust the difficulty. It would be up to people as to weather or not they wanted to leave and start a competing currency.

All in all, it would only make the network stronger IMHO. Not only would someone have a large vested interest in it then, but with that interest they would bring legal backing to the table and hopefully establish a decent precedent for competing currencies. Of course this is putting WAY too much faith into the justice system of whichever country that happens in but . . . .that's another problem altogether.

[theboos]

To match the current nearly 2 TH/s of the network, you'd probably find it cheapest and most energy efficient to develop an ASIC. Development costs are quite wide-ranging but always high. Then you could have a large pile of ASICs submerged in refrigerated oil and overclock to the moon. Since ASICs involve relatively low marginal costs, why stop at 2 TH/s when you can have 10 TH/s for maybe an additional 1% of your fixed costs. Even so, the point that has been made many times on these forums is that even if you had a massive rig that could do 10 TH/s, all you'd be able to do is to double-spend (and on average, less than 90% of the time). You might find better use for those ASICs in, *gasp*, legitimately mining at a rate of nearly 6000 BTC per day (at current difficulty).

Alternately, if your goal is to simply bring down the network, with a reasonably sized botnet, you could easily DDOS every client on IRC, plus each of the failovers included in the source code.

[evoorhees]

I don't think he was referring to someone getting 50% of the mining power purely with the intention of mining legitimate coins. I think he was referring to the possibility of gaining > 50% of the network in order to change the code in a way that would destroy bitcoin. Having the >50% allows the intruder's "version" of the software to appear to be the "legitimate" software.

My point is that the bitcoin enthusiasts would see the switch and alert the community, so that the 50% control would amount to no more than a giant waste of resources.

[cypherf0x]

Who says the BTC generated can't be legit?  If a user/organization has the resources to generate a massive amount of BTC and keep them out of circulation then it would drive the difficulty of generating coins for everyone else through the roof.  The profitability of mining would go way down unless they have access to ASICS/FPGAs dedicated to the task.

[1bitc0inplz]

I don't think he was referring to someone getting 50% of the mining power purely with the intention of mining legitimate coins. I think he was referring to the possibility of gaining > 50% of the network in order to change the code in a way that would destroy bitcoin. Having the >50% allows the intruder's "version" of the software to appear to be the "legitimate" software.

My point is that the bitcoin enthusiasts would see the switch and alert the community, so that the 50% control would amount to no more than a giant waste of resources.

Agreed.... except the alert part, but I'll swing back to that...

Running with a 2 THash/sec network size, gaining over 50% of this would require over 1 THash/sec of power... lets say 1.25 THash/sec just to "play it safe" from the attacker's point of view.

Building a rig similar to Whitepixel http://blog.zorinaq.com/?e=42, each rig would cost $2,700 USD. Each rig has four Radeon HD 5970. The hardware comparison chart (https://en.bitcoin.it/wiki/Mining_hardware_comparison) lists these cards of capable of 807 MHash/sec each. Each rig would therefore be capable of 3,228 MHash/sec.

Target (1.25 THash/sec) = 1,310,720 MHash/sec.

1,310,720 / 3,228 = 407 machines
407 * $2,700 USD = $1,098,900 USD.

So, with just slightly more than $1 million dollars you could acquire just slightly more than half of the current network.

Now, as far as the community noticing... why would they? Bitcoin is anonymous after all. And, as long as all 407 of those boxes weren't brought online at the same time, their presence could be masked as "normal network growth".

[1bitc0inplz]

I don't think he was referring to someone getting 50% of the mining power purely with the intention of mining legitimate coins. I think he was referring to the possibility of gaining > 50% of the network in order to change the code in a way that would destroy bitcoin. Having the >50% allows the intruder's "version" of the software to appear to be the "legitimate" software.

My point is that the bitcoin enthusiasts would see the switch and alert the community, so that the 50% control would amount to no more than a giant waste of resources.

Agreed.... except the alert part, but I'll swing back to that...

Running with a 2 THash/sec network size, gaining over 50% of this would require over 1 THash/sec of power... lets say 1.25 THash/sec just to "play it safe" from the attacker's point of view.

Building a rig similar to Whitepixel http://blog.zorinaq.com/?e=42, each rig would cost $2,700 USD. Each rig has four Radeon HD 5970. The hardware comparison chart (https://en.bitcoin.it/wiki/Mining_hardware_comparison) lists these cards of capable of 807 MHash/sec each. Each rig would therefore be capable of 3,228 MHash/sec.

Target (1.25 THash/sec) = 1,310,720 MHash/sec.

1,310,720 / 3,228 = 407 machines
407 * $2,700 USD = $1,098,900 USD.

So, with just slightly more than $1 million dollars you could acquire just slightly more than half of the current network.

Now, as far as the community noticing... why would they? Bitcoin is anonymous after all. And, as long as all 407 of those boxes weren't brought online at the same time, their presence could be masked as "normal network growth".

Ah, crud, I just realized my math was wrong...

Adding 1 THash/sec to a 2THash/sec network only gains you 1/3 of the new network size. You'd need over 2THash/sec, so that when you added it to the existing 2Thash/sec network you had over 1/2 of the total network.

Just double all my numbers... so it looks like it would cost over $2 million USD (and raising each day) to gain 1/2 of the network.

[xenon481]

~$2million is relatively nothing when you consider that ~$400,000 trades hands on mtgox every day.

[portpho]

That makes sense, thanks 1bitc0inplz, very interesting

I guess there are a few reasons someone with a few million dollars to burn might want to do this.

1. They have a vested interest in conventional currency. Think Mises.org conspiracy theories about international bankers who would see bitcoin as a threat to their economic domination.

2. A government decides bitcoin is illegal (esp because people aren't paying taxes!) and chooses this form of technological force to shut it down since conventional force is not sufficient to enforce the law in this case.

3. A malicious and bored millionaire just wants to do it for fun, to see if he or she can.

4. Someone thinks they could make a profit by sucking up all the bitcoins. Considering that people of limited resources are currently investing thousands of dollars in bitcoin speculation, maybe some angel investor with a billion dollars thinks this is a fun way to diversify his portfolio for a few millions.

5. A bored Saudi prince decides this is more fun than buying another private jet.

It would be interesting to see if someone could compromise the currency while still maintaining public confidence in it. That could be very profitable.

It seems to me that a currency which can be compromised by only a few million is on a precarious basis. How can the public really have confidence in a currency that may, unknowingly, be completely compromised by powerful established interests? I mean who is to say that the NSA has not already drawn up a project to cripple bitcoin? ~$2million to the NSA is utterly nothing. And, if we're being honest, they already have the hardware hooked into the net to do it. So it would likely be more of an opportunity cost issue rather than buying up the hardware.

[bcpokey]

I don't think he was referring to someone getting 50% of the mining power purely with the intention of mining legitimate coins. I think he was referring to the possibility of gaining > 50% of the network in order to change the code in a way that would destroy bitcoin. Having the >50% allows the intruder's "version" of the software to appear to be the "legitimate" software.

My point is that the bitcoin enthusiasts would see the switch and alert the community, so that the 50% control would amount to no more than a giant waste of resources.

Agreed.... except the alert part, but I'll swing back to that...

Running with a 2 THash/sec network size, gaining over 50% of this would require over 1 THash/sec of power... lets say 1.25 THash/sec just to "play it safe" from the attacker's point of view.

Building a rig similar to Whitepixel http://blog.zorinaq.com/?e=42, each rig would cost $2,700 USD. Each rig has four Radeon HD 5970. The hardware comparison chart (https://en.bitcoin.it/wiki/Mining_hardware_comparison) lists these cards of capable of 807 MHash/sec each. Each rig would therefore be capable of 3,228 MHash/sec.

Target (1.25 THash/sec) = 1,310,720 MHash/sec.

1,310,720 / 3,228 = 407 machines
407 * $2,700 USD = $1,098,900 USD.

So, with just slightly more than $1 million dollars you could acquire just slightly more than half of the current network.

Now, as far as the community noticing... why would they? Bitcoin is anonymous after all. And, as long as all 407 of those boxes weren't brought online at the same time, their presence could be masked as "normal network growth".

Ah, crud, I just realized my math was wrong...

Adding 1 THash/sec to a 2THash/sec network only gains you 1/3 of the new network size. You'd need over 2THash/sec, so that when you added it to the existing 2Thash/sec network you had over 1/2 of the total network.

Just double all my numbers... so it looks like it would cost over $2 million USD (and raising each day) to gain 1/2 of the network.

Your math also doesn't account for secondary costs. Aside from the 2 million USD of hardware (assuming you can find 1,628 5970s in short order, no small task), 407 machines running 1,628 GPUs would likely eat roughly 610 KiloWatts of power, requiring 5,087 Amps (or if you were on an industrial power grid 2,543 Amps). Either renting out expensive space at a data center or building a custom housing/generation complex would be required. Additionally 610KW = > 2Million BTU/hr, requiring some form of cooling to the tune of being able to remove 2 Million BTU. Custom networking and so on.

It obviously wouldnt be tens of millions more but there are lots of additional costs to running huge farms of machines.

[jibjabz]

So can someone tell me how this doesn't mean that someone w/ $2-5M can take over the entire BTC economy?

[1bitc0inplz]

Your math also doesn't account for secondary costs. Aside from the 2 million USD of hardware (assuming you can find 1,628 5970s in short order, no small task), 407 machines running 1,628 GPUs would likely eat roughly 610 KiloWatts of power, requiring 5,087 Amps (or if you were on an industrial power grid 2,543 Amps). Either renting out expensive space at a data center or building a custom housing/generation complex would be required. Additionally 610KW = > 2Million BTU/hr, requiring some form of cooling to the tune of being able to remove 2 Million BTU. Custom networking and so on.

It obviously wouldnt be tens of millions more but there are lots of additional costs to running huge farms of machines.

No, you are right... but I don't think the electric bill is a "big concern" for someone willing to spend over $2 million dollars (just in hardware) to "hack" a digital currency. I was mainly looking at this from a perspective of "what if the a government wanted to control Bitcoin" perspective....

[evoorhees]

Maybe someone could take it over for that amount, if done today. However, there exists a fortunate dynamic in that such an attack only becomes more likely to the extent that the bitcoin economy grows. And, to the extent the bitcoin economy grows, it becomes increasingly harder to take it over.

This is a wonderful natural defense mechanism of the system... the more attention it draws, the more dangerous people may want to attack it, but then the harder it is to attack. By the time the "bored Saudi prince" discovers this whole thing, perhaps it'll be too expensive for him to intervene. Here's hoping, anyway =)

[portpho]

So can someone tell me how this doesn't mean that someone w/ $2-5M can take over the entire BTC economy?

They can...

[BitterTea]

Assuming that $2-5 million would get you >50% of the network's computing power, you would be able to reverse your own transactions. There are other potential attacks that would require more or fewer resources and more or less luck, but double spending is the main threat of a computing power takeover.

[jasonk]

Let's say you just keep buying FPGAs or GPUs until you have over 50% of the network and can dominate it.

How many millions would it cost right now?

I'm curious just how rich you'd have to be to take over bitcoin.

Could you do it with $10 million as a small business owner?

Could a hedge fund CEO do it with $100 million?

Could Zuckerberg do it with $1 billion?

Could the NSA or DOD do it with $10 billion?

How about the Chinese government?

Basically, how many people in the world have the power to fuck over bitcoin if they wanted to?

The network currently is about 2,000 Giga hashes at max.

It costs about $1 per Million hashes of computer power (for decent quality parts). 1000M hashes = 1 Giga hash.

You would need basically 1000, $1000 dollar computers.  Total cost would be about 1 Million dollars would do the trick JUST for computer hardware.

Now if you didn't already have a facility to keep the computers, I imagine that would be in 100's of thousands or even a million or two for a data center capable of running that many computers.

Or one could potentially be leased.  I would say the minimum upfront cost to control over 50% of the network would be around 1.5M dollars.

[mjsbuddha]

deepbit has nearly 50% and it doesnt cost him a dime. it makes him money

[portpho]

why would deepbit not steal bitcoins if he has 50%?

since it would be impossible to catch him doing it... i think people would have to assume that he is

[BitterTea]

why would deepbit not steal bitcoins if he has 50%?

since it would be impossible to catch him doing it... i think people would have to assume that he is

No, it's not impossible to catch. In order to "steal" coins, he would spend them on some item or service and wait to receive it. Once he did, he would direct deepbit to build not from the current block, but the one prior to the transaction. It would have to generate all of the blocks that have passed since and surpass the length of the current chain, at which point the transaction would be erased, giving him back his coins.

It would be fairly obvious this was occurring, and all of his miners would jump ship, because it is not in their interest to take part.