51% attack on Bitcoin

[philipma1957]

I was wondering what it would cost to control 51% of the mining hash rate? Is this ever a realistic threat?

It's not a realistic threat because Bitcoin network is very big in hash rate and attackers have to spent huge and expensive cost to own 51% of total hashrate or more to do attack.

Because Bitcoin network is decentralized from nodes to hashrate, if something looks suspicious, miners will change their mining pools. Communities like exchanges have bots to detect attacks and will halt their withdrawal and deposit so attackers will waste their money for attacks.

How many Bitcoin confirmations is enough?
https://jlopp.github.io/bitcoin-confirmation-risk-calculator/
https://web.archive.org/web/20181231045818/https://people.xiph.org/~greg/attack_success.html

https://howmanyconfs.com/

PoW 51% attack cost
https://www.crypto51.app/

To be honest, I don't understand these values:
Name    Symbol    Market Cap    Algorithm    Hash Rate    1h Attack Cost    NiceHash-able
Bitcoin    BTC    $517.53 B    SHA-256           378,673 PH/s    $1,646,739             0%

Means that to attack Bitcoin network for 1h, it would cost 1 million, 6 hundred, forty six, 7 hundred and thirty nine dollars?


I would think this way:
Total hash power 378673PH/s

51% of Total hash power is ~190PH/s

Each S19xp Hybrid runs at ~257TH/s
Therefore we need  190000TH/s / 257TH/s = 740 miners
Each miner costs $8500, so, we need 8500*740 = $6290000 ~6.3 million dollars
Each miner uses 5350W/h, so, we need 5350 * 740 = 396kW/h * 24h = 9504kW / day
If eectricity costs, let's say 6cents / h, we need $1368576 / day.

I'm not absolutely sure about my numbers but they seem ok to me. Anyone correct me if I'm wrong!
Now, the only question I don't know for sure how to answer is how much time would it take for this setup to take over the main chain of the other 49% of hashpower!

your math is wrong

I just made a few fixes... If it's still too off, please tell me where!

190,000 to 195,000 ph in gear

190,000,000 to 195,000,000 th in gear

I used a 100 th s19j pro

so 1,900,000 to 1,950,000 machines at 1000 each

or used your gear.

190,000,000 th /257 = 739299 units at 8500 = 6,284,046,692 which about a factor of 1000x

I think I know where I went wrong.
When I evaluated 51% of total ash power, I dropped by a factor of 1000. It should be 195000PH/s and I wrote 195PH/s.
So, it should be 195000000TH/s / 257TH/s = 760000 miners!
760000 miners * 8500 = $6.5billion
Then, 5350W/h * 760000 miners = 4.1GW/h
At 6 cent/Kwh, this would mean $246million per hour.
Per day would be $6billion, which matches your numbers. Well, apparently you're an experienced miner, so, I just wanted to do this as an exercise for mylsef.
Thanks for pointing out my mistakes!

An even thoug, my approach wouldn't work. After reading some other posts here, I think someone said it woul be needed to buy at least twice as much miners as needed to add twice the current total hash power. So, I tink I would need to double my number to match this, which makes sense.

But I had a question that I think it wasn't answered. In this scenario (introducing twice as much as current total hash power plus some more to overcome the current hash power), how much time would one need to run the attack to make it effective?

And, of course that your approach would make more sense. The one that you described as forcing the tow major pools to fork the main chain!

To be honest, I don't understand these values:
Name    Symbol    Market Cap    Algorithm    Hash Rate    1h Attack Cost    NiceHash-able
Bitcoin    BTC    $517.53 B    SHA-256           378,673 PH/s    $1,646,739             0%

Means that to attack Bitcoin network for 1h, it would cost 1 million, 6 hundred, forty six, 7 hundred and thirty nine dollars?


I would think this way:
Total hash power 378673PH/s

51% of Total hash power is ~195PH/s

Each S19xp Hybrid runs at ~257TH/s
Therefore we need  190000TH/s / 257TH/s = 760 miners
Each miner costs $8500, so, we need 8500*760 = $6460000 ~6.5 million dollars
Each miner uses 5350W/h, so, we need 5350 * 760 = 4066kW/h * 24h = 97584kW / day
If electricity costs, let's say 6cents / h, we need $5900000 / day.

I'm not absolutely sure about my numbers but they seem ok to me. Anyone correct me if I'm wrong!
Now, the only question I don't know for sure how to answer is how much time would it take for this setup to take over the main chain of the other 49% of hashpower!

You're calculating the cost of building a farm to attack the network, and you do that also wrong as Phil said
What that calculator does is simply taking the rent cost for 1th/a and multiplying by the needed hashrate, so no initial gear cost, no energy not anything else. Much like buying a car to go on a trip versus taking rental.

After all it's pretty simple, to get that narrate you just have to rent it buy paying the miners more than they are making but you will never be able to bribe mining pools with tens of exahashes to rent you that for one hour or two.

Yes, I found my error and fixed it.
I don't know if there would be enough renting sites (actual hardware) enough to run such attack! I think Phil's approach of hijacking the 2 major pools would make more sense!

I think doing 51% or more for one month in a row would fuck the network up.

but it would really be hard as it would be with new gear.  built by bitmain sponsored by china.

they would need to use all the power from the biggest hydropower dam and 410 eh.

so maybe 22,000 mega watt dam which they have. plus a few more which they have.

the worlds biggest mine with 2 million s19 pros
so much wire pdus etc. fans ya da ya da ya da.

say one month in a row. btc would suffer greatly if that happens.

the gear maybe 4-5 billion the power .

so china could do it maybe 🤔 as they must let noone know they are gathering all that gear..

You would be spending 10 billion to wreck 500 billion which is acceptable in a war but maybe not for a btc attack.

[TravelMug]

Hey guys,

As we know, as long as a majority of CPU power is controlled by nodes that are not cooperating to attack the network, Bitcoin network remains safe.

I was wondering what it would cost to control 51% of the mining hash rate? Is this ever a realistic threat?

Appreciate your answers in advance!

Kind regards,
Papa

We are already in 2023 mate, and no one has talking about this because it's impossible to do right now. It will take a lot of resource on the entity that will try to do this. And you have to think what incentives they got for spending that huge amount of money, billions upon billions and then what? after that the nodes will be back to the chain itself.

So it doesn't make sense at all to do this kind of Rekt and definitely we are not in war per se. Maybe the entity will try to prove a point but that's it, the money and resource it will cost him/they are not worth it.

[stompix]

Fixed typos

Yes, I found my error and fixed it.
I don't know if there would be enough renting sites (actual hardware) enough to run such attack! I think Phil's approach of hijacking the 2 major pools would make more sense!

There are none, even that links that "calculates" the cost is showing only 110petahash available for rent, that's more like  0.1% rather than 51%.

Basically three possibilities to achieve that:
- you buy/build it, and the costs are in billions as per above
- you rent it, but three is nobody renting you that much
- you steal it, as in hijacking a pool or bribing some operator

But all are expensive, they require industry connections, and they all end with the same question, why put all the effort for it? A nation would just take legislative measures, a terrorist group has nothing to earn, other groups are interested in making money not spending money, so just some bored kid who has by chance manage to infect some servers of mining pools but that's quite the stretch.

We are already in 2023 mate, and no one has talking about this because it's impossible to do right now. It will take a lot of resource on the entity that will try to do this. And you have to think what incentives they got for spending that huge amount of money, billions upon billions and then what? after that the nodes will be back to the chain itself.

It's not impossible, yes as I also said the money spending doesn't make sense but it's not impossible. As for the nodes, no, the nodes will not be back to the chain because that would break the rule, they have to follow the chain with the most work, in this care the malicious one.

[darkv0rt3x]

Fixed typos

Yes, I found my error and fixed it.
I don't know if there would be enough renting sites (actual hardware) enough to run such attack! I think Phil's approach of hijacking the 2 major pools would make more sense!

There are none, even that links that "calculates" the cost is showing only 110petahash available for rent, that's more like  0.1% rather than 51%.

Basically three possibilities to achieve that:
- you buy/build it, and the costs are in billions as per above
- you rent it, but three is nobody renting you that much
- you steal it, as in hijacking a pool or bribing some operator

But all are expensive, they require industry connections, and they all end with the same question, why put all the effort for it? A nation would just take legislative measures, a terrorist group has nothing to earn, other groups are interested in making money not spending money, so just some bored kid who has by chance manage to infect some servers of mining pools but that's quite the stretch.

Yes, absolutely. If it was feasible, I think someone would have attempted it for sure! So, all these said, I think we have a pretty rock solid safe system!

[dragonvslinux]

To be honest, I don't understand these values:
Name    Symbol    Market Cap    Algorithm    Hash Rate    1h Attack Cost    NiceHash-able
Bitcoin    BTC    $517.53 B    SHA-256           378,673 PH/s    $1,646,739             0%

Means that to attack Bitcoin network for 1h, it would cost 1 million, 6 hundred, forty six, 7 hundred and thirty nine dollars?


I would think this way:
Total hash power 378673PH/s

51% of Total hash power is ~195PH/s

Each S19xp Hybrid runs at ~257TH/s
Therefore we need  190000TH/s / 257TH/s = 760 miners
Each miner costs $8500, so, we need 8500*760 = $6460000 ~6.5 million dollars
Each miner uses 5350W/h, so, we need 5350 * 760 = 4066kW/h * 24h = 97584kW / day
If electricity costs, let's say 6cents / h, we need $5900000 / day.

I'm not absolutely sure about my numbers but they seem ok to me. Anyone correct me if I'm wrong!
Now, the only question I don't know for sure how to answer is how much time would it take for this setup to take over the main chain of the other 49% of hashpower!

You're calculating the cost of building a farm to attack the network, and you do that also wrong as Phil said
What that calculator does is simply taking the rent cost for 1th/a and multiplying by the needed hashrate, so no initial gear cost, no energy not anything else. Much like buying a car to go on a trip versus taking rental.

After all it's pretty simple, to get that hashrate you just have to rent it by paying the miners more than they are making but you will never be able to bribe mining pools with tens of exahashes to rent you that for one hour or two.

Exactly what I was thinking when people discuss the cost of a 51% attack on Bitcoin for an hour. However, technically they are somewhat correct, as it seems the available hash rate isn't "freely" available to hire. At least from NiceHash-able (that supplies hashing power) there is not even 1% available. As unsurprisingly there isn't 400 EH/s available to rent, as that is a shitload of hash rate.

As you said you wouldn't be able to bribe mining pools, as they have an interest in the network not getting attacked in such a manner. So trying to convince numerous mining farms to rent you hash power seems like a long-shot to me. At least if a fair rent cost per hour is around $1.3m, then I think it could be up to twice as much in order to convince mining farms to rent you power.

[teosanru]

Hey guys,

As we know, as long as a majority of CPU power is controlled by nodes that are not cooperating to attack the network, Bitcoin network remains safe.

I was wondering what it would cost to control 51% of the mining hash rate? Is this ever a realistic threat?

Appreciate your answers in advance!

Kind regards,
Papa

It's not realistic. Making a 51% attack also means that sustaining that attack for the time. Only one time 51% attack won't be enough for this. Which means it will take a lot of hardware. By a lot i mean really too much and this will come at a big cost as well and a big logistic center to set up this much computing as well. Until quantum computing kicks in it's not possible and isn't really that much of a threat to bitcoin.

[seoincorporation]

Try to do the 51% attack with money is not the right approach, because if the main mining pools organize to do the attack then the cost would be close to zero. Right now the main pool is Foundry USA with 33.8% of the network, the second one is F2Pool with 17.5% and in 3rd place we have AntPool with 16.9% of the network. So, if Fundry USA buys or collaborates with F2Pool or Antpool, they will have more than 51% of the network and they would be able to do the attack. Simple as that.

But if you think that you can spend billions in miners to get the 51% of the network, that is not possible. There are not enough miners in the market for this.

[philipma1957]

Try to do the 51% attack with money is not the right approach, because if the main mining pools organize to do the attack then the cost would be close to zero. Right now the main pool is Foundry USA with 33.8% of the network, the second one is F2Pool with 17.5% and in 3rd place we have AntPool with 16.9% of the network. So, if Fundry USA buys or collaborates with F2Pool or Antpool, they will have more than 51% of the network and they would be able to do the attack. Simple as that.

But if you think that you can spend billions in miners to get the 51% of the network, that is not possible. There are not enough miners in the market for this.

Almost totally agree pools would be the ones to do it.  And you would need top four or five to agree to do it.  Plus they need to do it for quite a while.  weeks not days.

If it started people would leave the pools. My measly 3 ph would go to a small pool and we would need quite a few small pools to pick up the slack.

 I suppose the developers could stop the chain. Not so sure how they can but I am thinking they can

[BVeyron]

Hey guys,

As we know, as long as a majority of CPU power is controlled by nodes that are not cooperating to attack the network, Bitcoin network remains safe.

I was wondering what it would cost to control 51% of the mining hash rate? Is this ever a realistic threat?

Appreciate your answers in advance!

Kind regards,
Papa

Unfortunately it can happen. IDK the price of such hustle, but the sum is surely affordable for the worlds main tycoons and banks, especially if their actions will be organised. That's actually the main threat to decentralisation.

[death69]

You're diving into the deep waters of the 51% attack concept, aren't you? Well, strap in; this is going to be a thrilling ride.

Now, assuming you'd want to control 51% of the Bitcoin network, the first thing you'd need is an insane amount of cash. How much are we talking? Let's just say it would be easier to buy a small country. Joking aside, you're looking at a multi-billion dollar investment.

But let's remember, Bitcoin isn't just a tech structure; it's a community. If there were sudden, inexplicable shifts in the hash rate, the community would take notice. The integrity of the network relies on collective trust. So, even if you could afford the staggering price tag, attempting a 51% attack would be like trying to rob a bank with everyone watching.

To answer your question – a realistic threat? Highly unlikely, given the astronomical cost and the inherent checks and balances of the Bitcoin community.

[BitDane]

In theory, a 51% attack is possible, but in practice it is impossible for many reasons, but the most important of them, as everyone pointed out, are economic reasons, because there is no economic benefit from this attack. It means that you are spending millions of dollars to steal some dollars. This is illogical.

I think way back, 51% attack could have happened but the pool operator did not take advantage of the scenario.  Ghash.io briefly exceeded 51% of the hash rate, and that forced the community to discuss a solution to the threat[1]. So, it can be possible in practice if the operator of the huge mining pool agreed to merge their hashes and attack the Bitcoin network.

You're diving into the deep waters of the 51% attack concept, aren't you? Well, strap in; this is going to be a thrilling ride.

Now, assuming you'd want to control 51% of the Bitcoin network, the first thing you'd need is an insane amount of cash. How much are we talking? Let's just say it would be easier to buy a small country. Joking aside, you're looking at a multi-billion dollar investment.

Or the consensus of different mining operators to join hashes to create a 51% attack on the network.


[1] https://en.wikipedia.org/wiki/GHash.io

[Cricktor]

I don't understand the fuss about 51%-attacks in the light of past years hashrate and its growth or even basically from 2010 on. Despite the vast amount of money needed one should also think about what is possible with such an attack. It has been calculated that its very very expensive. What do you get? Well, you can double spend your own transactions. You can temporarily censor others transactions by rewriting very few past blocks from the tip of the blockchain. If market becomes aware of your malicious actions very likely the BTC price will plummet. I hope you have ca$h left to buy at low prices. 

That's it. You can't steel coins controlled by private keys that you don't own yourself, not even if you have 99.9999% of total hashrate.
You can only buy other coins at lower prices from panic sellers when you bring the market to plummet. But that's technically not stealing, it's trading under manipulated market conditions. Don't take my last sentences too seriously, my point is, if you don't have others private keys, you can't move/steal others coins, regardless of your hashpower in possession.

Did I miss something?

[stompix]

Exactly what I was thinking when people discuss the cost of a 51% attack on Bitcoin for an hour. However, technically they are somewhat correct, as it seems the available hash rate isn't "freely" available to hire. At least from NiceHash-able (that supplies hashing power) there is not even 1% available. As unsurprisingly there isn't 400 EH/s available to rent, as that is a shitload of hash rate.

As you said you wouldn't be able to bribe mining pools, as they have an interest in the network not getting attacked in such a manner. So trying to convince numerous mining farms to rent you hash power seems like a long-shot to me. At least if a fair rent cost per hour is around $1.3m, then I think it could be up to twice as much in order to convince mining farms to rent you power.

Well, first my bad at saying 0.1 based on a website that monitors it, if we look at the actual marketplace, yes they do have around 1%, or 3 exas.

Now, there is a trick to this, don't know why I mentioned it first since it was the most important and adding more to Phil's hint at hijacking being the easy way out,  if you rent hashrate from nicehash, it doesn't mean automatically you're free to do anything you want with it, Nicehash for example,has  a policy with limited mining pools to which you can point that, so you will need to either get Nicehash to approve your malicious pool or bribe the pool to which you point this.

So no f#$% way would a big player like mara or core or riot allow you to rent this and point to your pool at Iwannahackbitoin.cheap 
They will rent it to you at a 10% surcharge but only if you mine at either foundry or maybe one or two trustable alternatives.

Don't take my last sentences too seriously, my point is, if you don't have others private keys, you can't move/steal others coins, regardless of your hashpower in possession.

Did I miss something?

Yeah, imagine that 10 blocks with like the last block mined that had 7,897.78BTC worth of transactions so around 200 mils, gets invalidated so 2 billion worth of payments deposits, and withdraws suddenly are nulled or are spent differently, multiply this by a bigger timeframe of around 24h and see who will use bitcoin for anything after such an event or how much trust would anyone put in it.

Some are too focused on the aspect of investment where you just buy and hodl and forget that for it to actually have value you need to usage, and this would be sent in a coma if someone would really manage to get such an attack going for a longer period of time, not 1h no 4 h but something in the days value. Highly unlikely but it won't be a walk in the park if it does happen.

even if you could afford the staggering price tag, attempting a 51% attack would be like trying to rob a bank with everyone watching.

No, it's like robbing a bank with a tank and the only guard watching is a guy in the 90s in a wheelchair and armed with a squirt gun.
Since you mentioned this, si a pool does a 51% , you, like the other guys watching, what are you going to do to stop them?
Genuinely curious!

[Cricktor]

Yeah, imagine that 10 blocks with like the last block mined that had 7,897.78BTC worth of transactions so around 200 mils, gets invalidated so 2 billion worth of payments deposits, and withdraws suddenly are nulled or are spent differently, multiply this by a bigger timeframe of around 24h and see who will use bitcoin for anything after such an event or how much trust would anyone put in it.

Some are too focused on the aspect of investment where you just buy and hodl and forget that for it to actually have value you need to usage, and this would be sent in a coma if someone would really manage to get such an attack going for a longer period of time, not 1h no 4 h but something in the days value. Highly unlikely but it won't be a walk in the park if it does happen.

To my Bitcoin protokol knowledge, correct me if I'm wrong, if you rewrite a block or more around the tip of the blockchain and you decide to make some transactions disappear in your own mined blocks, those "erased" transactions don't disappear into thin air. Other nodes that recognise your faster growing chain fork (which has to have more chain work than the existing blocks which you try to erase) and accept it as the valid chain of max. hash work, will put the transactions from the now abandoned blocks back into their mempool. That applies to any transaction that is missing in your new blocks compared to the abandoned blocks.

You can only dismiss transactions completely and make the network forget them if you have nearly 100% of the hash power and can thus keep the undesired transactions as long as out of blocks until they get removed from mempools due to exceeding the expiery period (standard 336h, but individually configurable per node). Well, that's even more unlikely.

No question that such activities would upset the Bitcoin market to hell and plummet it. So there's highly likely no benefit for such a malicious actor, except his sole purpose is to kill Bitcoin. Quite an expensive kill, I say.

[Kryptowerk]

Some other cool, game-theoretic facts about a 51% attack on the Bitcoin network, besides the already mentioned (astronomical) initial costs:
- As soon as the network is attacked, the Bitcoin price would probably drop by a huge factor, so even double spending your own transactions will most likely not make up for your initial costs AT ALL
- You still don't and never will controll any private keys that you didn't already control before the attack
- You can't change previous blocks. If you really control MUCH MORE than 50%, maybe you can afford to even rewrite a few of the latest blocks, but the digital-pyramid that the Bitcoin blockchain is, will have maybe handful of its hundrets-of-thousands of stones changed, the rest will still stand unchanged.