a 51% attack costs $20,000,000 and is devastating

[giszmo]

Just a quick reminder that even though we are 10 times faster than the 500 fastest super computers on earth (in doing what we are doing), we are far from out of reach to a dedicated attacker.

If I had the job to destroy bitcoin, I would borrow $20 million to build my own little asic mining op. $20 million is about what people claim BFL raised, right? Ok, make it 40 million then, but as I said, the money will come back. I only need it temporarily.

Lets assume you can mine almost at a profit, which is the case. Ok, lets do that. Lets increase our hashing power to 51% for $40 million. This is possible as there hasn't been more money invested in current ASICs. If we haven't been profitable before, now we are profitable because we can drive all the competition out of the market. All block rewards are ours. We don't interfere with any transaction as by that we would discriminate our blocks and the devs could counter our attack. We just mine all the bitcoins and behave but let the world know that nobody will mine except for us. We sell the mined coins to pay back our debt and divert hashing power that's being freed to other block chains that surely will come up. Now we can set the rules and have inflation at will.

Am I off by some millions? Am I missing some basic protection against such an attack? Are there really more miners that would try to overpower this attacker than there are miners that would just leave once it turned out this thing was ongoing? I doubt it.



I'm afraid this can only be countered by some proof of stake. Too sad PPCoin is not an option. I kind of hope that such an attack would drive the original bitcoin to proof of stake.

<tinfoilhat>Maybe we already operate at the mercy of IMF, FED, world bank, etc., we already have full approval and bitcoin is meant to replace the dollar. Not being destroyed long ago is a very bullish sign. What would be the sense in destroying Bitcoin rather later than sooner, now that VCs are entering the field? All investors are going to be pissed about whoever destroys bitcoin, so why have more enemies than necessary?

[empoweoqwj]

U lost me at the first mention of BFL, sorry 

[oakpacific]

Your money will not come back, to 51% attack the network, you will mine 100% of the blocks, everyone will know immediately, and they either defeat you or abandon the network, then your bitcoins will soon become worthless.

Also, you forgot the $100,000-200,000/day electricity cost, and heat dissipation, and land purchase....

[giszmo]

U lost me at the first mention of BFL, sorry 

Damn
Hey, but seriously, mining consumes resources in the order of magnitude of value mined. So the initial ASICS investment can be estimated as 8 months of mining worth at $50/BTC as the current increase should not yet be priced in (they are selling at insane prices, but you wouldn't buy from them but design your own line).

We are currently at 1.25Mio Bitcoin / year, that is $50Mio/8 months. Here you have a number that doesn't require BFL or any other mining hardware producer's numbers. Users are not putting more money into mining than that and for my point of "we are not invincible", I don't care if you change that number to $500Mio, as that's still only half a day of QE and therefore well in reach to defend against substantial threats. Especially when you can get your money back by selling the mined bitcoins.

[giszmo]

Your money will not come back, to 51% attack the network, you will mine 100% of the blocks, everyone will know immediately, and they either defeat you or abandon the network, then your Bitcoins will soon become worthless.

If I claim to be the good guy who just noticed that his mining operation can be more profitable with 51%, I might lull people into believing that I'm just a business man. Maybe I sponsor the next bitcoin conference and grab a seat at the Bitcoin Foundation.

[oakpacific]

Your money will not come back, to 51% attack the network, you will mine 100% of the blocks, everyone will know immediately, and they either defeat you or abandon the network, then your Bitcoins will soon become worthless.

If I claim to be the good guy who just noticed that his mining operation can be more profitable with 51%, I might lull people into believing that I'm just a business man. Maybe I sponsor the next bitcoin conference and grab a seat at the Bitcoin Foundation.

No, long before you cross into the 51% domain, people will start noticing, when you are at 40%, you already would have more than 60% of chances to mine 6 blocks in a row, you have no excuse. And people will abandon the network not because they don't believe you ,but because it makes no sense economically to mine anymore, when you get all the blocks. And if you fancy being the sole miner in a network and still confident to be able to somehow lure people into use it, why not just fork?  It's only one day's work after all.

[jaywaka2713]

What people don't understand is that you literally need to be an electrical scientist to power such a farm. No home or building can draw the required power, so you may as well wire it into the nearest power station.

[CurbsideProphet]

This has been discussed ad nauseam, surely someone around for so long has seen this brought up before?

[giszmo]

This has been discussed ad nauseam, surely someone around for so long has seen this brought up before?

Yep, I have. Still it pisses me off that people state over and over again that we are 10 times more powerful than the 500 supercomputers that cost billions to construct, thus it would cost tens of billions to destroy bitcoin. No. It costs some lousy 20 million to destroy bitcoin. All should challenge these loud-mouths that brag about bitcoins invincibility.

[threeip]

Your money will not come back, to 51% attack the network, you will mine 100% of the blocks, everyone will know immediately, and they either defeat you or abandon the network, then your bitcoins will soon become worthless.

Pretty sure Gavin and the superfriends would say 'f#$% this' and the above scenario would happen.

Isn't this pretty much what happened with the '2 chainz' blocksize crisis?

And people will abandon the network not because they don't believe you ,but because it makes no sense economically to mine anymore, when you get all the blocks.

Also there's more interesting things you can do with $20m 

[oakpacific]

Your money will not come back, to 51% attack the network, you will mine 100% of the blocks, everyone will know immediately, and they either defeat you or abandon the network, then your bitcoins will soon become worthless.

Pretty sure Gavin and the superfriends would say 'f#$% this' and the above scenario would happen.

Isn't this pretty much what happened with the '2 chainz' blocksize crisis?

Everyone has the power to disagree, and you can't beat a 51% attacker unless you have the resource, whatever your decision is.

And yes, do I need to bring up again that after 4 years no usable alternative Bitcoin software is in widespread use out there?

I don't think folks like Gavin like to do something like this, but imagine what will happen to them if they sit there and do nothing, if there is any oligarchy it's made by us, not them.

[SGExodus]

You will need more than 100% of the current hash power, i.e. 120TH/s to attempt a 51% attack on the current network.

For illustration, we can use Avalon bulk order chips at 0.08 BTC per 300Mh/s for some quick calculation.

Roughly 400,000 Avalon chips will get us 120TH/s of hash power

That's about 32,000 BTC in price.

Assuming 16 chips per board, and each board + assembly cost 1BTC,  we will need a total of 25,000 boards, or 25000 BTC

Assuming 1 power supply unit to support 8 boards each, and each cost 1 BTC, we will need 3125 PSU, or 3125 BTC

Summing up, you can spend 60125 btc to get about 120TH/s of hash power, way less than $20 milllion

And we only need 60125 / 3600 or 16.7 days to generate ROI, or rounding to 20 days if including electricity and other setup expenses

Stop reading, go ahead and raise your fund to build this setup

[threeip]

And yes, do I need to bring up again that after 4 years no usable alternative Bitcoin software is in widespread use out there?

I use my own transaction signing scripts and offline wallets. Some people use blockchain.info wallets. I guess the main client works pretty well for most people, plus Armory for extras, or Electrum client, or android client 

[oakpacific]

And yes, do I need to bring up again that after 4 years no usable alternative Bitcoin software is in widespread use out there?

I use my own transaction signing scripts and offline wallets. Some people use blockchain.info wallets. I guess the main client works pretty well for most people, plus Armory for extras, or Electrum client, or android client 

But still none of them can work without bitcoind as a backend? That's what I meant.

[threeip]

Sorry, i get it. Its late here 

[01BTC10]

And then the dev change hashing algorithm...

[Melbustus]

OP's main point that a 51% attack is doable by a determined attacker is correct. It may not make sense to do so economically, but the attacker may not be economically motivated.

Bitcoin is the world's most powerful distributed computing project by an order of magnitude or two, but hashpower still needs to rise several orders of magnitude for bitcoin to be comfortably 51% attack resistant.

That said, hard fork is the solution if it ever did happen. It'd be quite disruptive, but it wouldn't be fatal.

[oakpacific]

OP's main point that a 51% attack is doable by a determined attacker is correct. It may not make sense to do so economically, but the attacker may not be economically motivated.

Bitcoin is the world's most powerful distributed computing project by an order of magnitude or two, but hashpower still needs to rise several orders of magnitude for bitcoin to be comfortably 51% attack resistant.

That said, hard fork is the solution if it ever did happen. It'd be quite disruptive, but it wouldn't be fatal.

Actually, with checkpointing, you only need to take extra precaution whenever you deal with large amount of bitcoins, make sure to wait long enough(much more than 6 confirmations but no more than 200).

[Eastwind]

Summing up, you can spend 60125 btc to get about 120TH/s of hash power, way less than $20 milllion

And we only need 60125 / 3600 or 16.7 days to generate ROI, or rounding to 20 days if including electricity and other setup expenses

Stop reading, go ahead and raise your fund to build this setup

The problem is, it will take me about two months to build 120TH, at that time, the next work power is already 240TH/s. I will have to double my investment.

[bitpop]

You don't even know what a 51% attack is