ETC attack – what is the cost to mount on ETH?

The attack on ETC shows the [edit: assuming truth of emails] first third in-the-wild use of hashpower to perpetrate a 51% attack on a cryptocurrency. RIP ETC.

Given that there is at least 110K GH/s worth of hashpower that we know can solve Ethhash but isn't (because on 9 August it was here securing Ethereum and now isn't), we know there's 110 K GH/s that can be used to attack ETC. If it's worth it.

And so far, it's been worth it to burn about 10K GH/s on ETC. So as I said, RIP ETC: if there's 100K twiddling its thumbs looking for something to do, and 10K can be profitably employed, it's not difficult to dream up profitable short-term attacks.

But what does that mean for ETH?

I have been tracking cost of security. At the peak hashpower 9 August, we were paying about $1.17 per GH/s to secure the network and we had about 300K of hashpower.

To ground the numbers more precisely: 295K GH/s, block rewards of 20.5K plus gas of 2.4K total 22.9K ETH @ $363 / ETH. Someone please check that.

As of last night's numbers, we are at $0.63 per GH/s: hashpower now 187K GH/s, block rewards 17.8K, gas 0.4K total 18.2 K @ $156.4 per ETH.

So what does this mean? The first part is obvious and I said it already: there is 112K GH/S hashpower that used to be mining ETH that isn't mining ETH. It's doing something. The hardware didn't evaporate.

The second part is more insidious. There is now an available business case to attack ETH, and we can measure it.

If we assume miners are mercenary and reward oriented, and we offer to pay them $1.17 per GH/s, we know for a fact there is a pool of at least 297 GH/s that is willing to mine for that reward.

We also know at that rate it would cost us a total of about $117K to rent 200K of hashpower for 30 minutes. In this period of time, we are going to pay whoever builds the longest string of blocks (about 124 blocks) about $60K because that's what we paid per half-hour yesterday. Since our attack cost is the cost to rent the hashpower minus the reward we get as the set of miners on the longest chain, the net cost to rent enough hash-power to attack the Ethereum chain for 30 minutes is about 57K or about 380 ETH.

380 ETH is trivial.

Particularly if you could short 10,000 ETH.

The only real problem is coming up with 200K GH/s in mining power that would rather be paid $1.17 per block versus $0.63 per block. Right now we have an existence proof that there is at least 110K out there.

With this in mind, does anyone else think that reducing block rewards and forcing unprofitable miners to look for other ways of getting rewarded might be a very bad idea at this time?

Submitted January 07, 2019 at 02:52PM }
via reddit http://bit.ly/2CWeg7B

Advertisements