Oleg Khovayko, Emercoin CTO, presented his algorithm for calculating the cost of a 51% attack on different blockchains. He introduced a numerically measurable hack51 coefficient by which to compare the potential resistance of cryptocurrencies to this type of attack.
The ratio of the attack price to the capitalization of a particular cryptocurrency was chosen as the ratio: hack51 = attack_cost / market_cap. The higher the ratio, the safer the cryptocurrency is in terms of the attack 51%, that is the ability to gather enough computing resources to rewrite the block chain.
The price of the attack was taken as the cost of a daily contract of cloud mining from advertisements of companies that provide such services. It then scales to a hashing capacity equal to the current hash rate of the cryptocurrency network. The weakness of this model is obvious, i.e., the inability to gather hashing power of more than half of the network to attack a particular cryptocurrency for a calculated price. Naturally, when trying to consolidate the corresponding hashing power the price of contracts will increase many times. However, it is not intended to investigate the practical sustainability of particular networks, but simply to create a criterion so that their sustainability can somehow be evaluated and compared.. It’s hard to compare how much one “very much” is different from another “very much” and which one is greater.
For the calculations, we wrote a simple PERL program that does these trivial calculations:
#!/usr/bin/perl -w
die “Usage:\n\tcontract_price contract_days contract_hashrate network_hashrate market_cap\n\n\n” unless $ARGV[4];
my ($price, $days, $contract_hashrate, $network_hashrate, $market_cap) = @ARGV;
my $price_one_day = $price / $contract_hashrate / $days;
my $cost_51 = $network_hashrate * $price_one_day;
my $ratio = $cost_51/ $market_cap;
print “One unit hash price per day:$price_one_day\n”;
print “Cost 51% attack per day: $cost_51\n”;
print “Ratio cost_51% / market_cap: $ratio\n”;
The program parameters are as follows:
- contract_price – contract price offered by the service;
- contract_days – duration of the contract in days;
- contract_hashrate – hashrate of the contract;
- network_hashrate – hashrate of the cryptocurrency network;
- market_cap – capitalization of the cryptocurrency.
.
.
The program outputs the following results:
-
One unit hash price per day: The price of one unit hash per day (in the units you entered).
-
Cost 51% attack per day: The price of an attack per day.
-
Ratio cost_51% / market_cap: The ratio of attack price to capitalization (hack51).
The program was run for several cryptocurrencies and the following results were obtained:
Cryptocurrency
Hash unit price
Attack cost 51% per day
Hack-51 ratio
BTC
0.268 $/th/s/day
$37 589,041
0.0000422
ETH
0.038 $/mh/s/day
26 676,411
0.0000640
UNO
0.268 $/th/s/day
371
0.0000463
NMC
0.268 $/th/s/day
23 627,397
1.13
EMC
0.268 $/th/s/day
12 189,589
4.57
What can be seen from this and what conclusions can be drawn? The rapid growth of prices of the most popular cryptocurrencies led to the fact that the first pair (BTC, ETH) can be attacked by about 50 millionths of the capitalization. That is, if someone wants to make a big fuss, he can do it.. It is very likely that someone could open a short on a Bitcoin ETF and then compromise the network, dropping the price of the underlying asset to near zero;
Merged Minign (UNO) doesn’t always work. Unavailability of mining in the absence of unique properties of the coin does not attract miners, which reduces complexity and, as a result, increases vulnerability. Technological coins mined by combined mining (NMC, EMC) show higher resistance against PoW attack of 51%. EMC’s fantastic resilience (an attack requires resources nearly five times the coin’s capitalization) demonstrates the usefulness of triple consensus Emercoin (PoS+PoW+MergedMining) for blockchain security. 
Now everyone can run the above program for their favorite cryptocurrency and see the hack51 ratio for it. Everyone can also try to challenge the proposed sustainability criterion and try to find a bug in the program.