3 ms·

Sure, difficulty can adjust down. But can the security of bitcoin afford the loss of that hashrate? It really depends on how much value Bitcoin is securing at

by phire 3y ago

Sure, difficulty can adjust down. But can the security of bitcoin afford the loss of that hashrate? It really depends on how much value Bitcoin is securing at the time. Because the amount paid out for mining rewards is also the rough cost for someone to execute a 51% attack. If miners can produce X hashes/sec for an easy profit at $Y, then an attacker should also be able to produce slightly more than X hashes/sec for about the same cost of $Y, or at least in the same order of magnitude. So, can the attacker make more than $Y profit/day from exploiting double-spends? The security of bitcoin becomes an economic problem. ------------------ On a related tangent, I used to track random shitcoins as a hobby. And you might be shocked at just how many 51% attacks happened. Many of the shittest shitcoins were secured by no more than a few hundred dollars per day of hashrate. Super-easy for someone to rent that off nicehash. Some of the reorgs I saw were multiple days long, and the attacker would continue attacking the same chain until their victim noticed, or ran out of money. I saw this done on dozens of shitcoins in the few months I was tracking. Based on that experience, I wouldn't trust any PoW blockchain that didn't have the majority of hardware miners for a given algorithm. So I would not trust BCH... how long before some medium sized bitcoin miner decided they could get more profit by 51%ing BCH for $175k/day? You can assume bitcoin is somewhat safe right now, but only because most of the SHA256 miners are already mining bitcoin. It would take a massive upfront investment for someone to create enough SHA256 miners to execute a 51% attack. But in your suggested scenario where the hashrate of BTC is dropping? Where are those older SHA256 miners going? The second hand market? What if someone buys them all up? And what if Bitcoin becomes actually successful? Becomes the backbone of the world economy and doing trillions per day in transactions? Someone might decide that upfront cost of creating miners is worth it, especially if they only need $657k of hashrate.

nullc 3y ago
Those are all fair arguments you can make, but none of them make a case that the current subsidy level is necessary. For a system that is actually in use, the consequences of these attacks you imagine are difficult to reason about. It's not like the system stops existing if some attacker attacks-- the attacker can reverse some of their own transactions near the chain tip. Recipients can adapt to the behavior by requiring more confirmations, restricting who and how they attack, etc. It's difficult to generalize from what's seen in those "random shitcoins" (to use your phrase) since most of them have little to no economic use, so an attack may have an oversized or undersized impact depending on pure chaos: their value is purely speculative so the narrative around the attack determines the outcome more than anything. Since you've said you tracked these things presumably you're aware that in many cases their value goes up after they're attacked. > but only because most of the SHA256 miners are already mining bitcoin That's also not clear to me, there are an awful lot of idle two generation old miners out there at any given time. > And what if Bitcoin becomes actually successful? Becomes the backbone of the world economy and doing trillions per day in transactions? The total value of activity on the chain isn't equivalent to the value available to attackers in an attack, especially to the extent that attacking undermines the value of the system as a whole. In any case, I wasn't (and am not now) trying to take a strong position here-- I would say that there are a lot of unknowns but that I think the position you were taking that the current fee level isn't adequate doesn't really pass a sniff test for me. Perhaps it isn't but if you really strongly believe that there are about 1000 cryptocurrencies that you should be figuring out how to short hard instead of posting on HN. :P
- phire 3y ago
  > but none of them make a case that the current subsidy level is necessary. I wasn't actually arguing that, yes there are a bunch of unknowns. I'm only making the argument that there is a clear formula connecting mining rewards, potential double-spend profits and the cost of attacking a PoW coin. I don't care enough to actually derive the formula and plug numbers in it to answer that question. > Recipients can adapt to the behavior by requiring more confirmations, restricting who and how they attack More confirmations don't really work as a solution. The cost to hold a double-spend open is known, so we could try requiring enough confirmations to make such a double-spend unprofitable, but the attacker can execute multiple double-spends in parallel to spread their costs. For large transactions (relative to the hashrate of that coin), the confirmations can can get quite high, which destroys the usefulness of cryptocurrency for fast and easy transactions. For one of the "random shitcoins" I was tracking, I found the centralised exchange I suspect the attacker was targeting (presumably trading them to another currency and then withdrawing). The exchange had set really high numbers of confirmations (2-5 days worth of blocks) but the because these coins were so cheap to attack, the attacker had no problem doing reorgs that large. And then the exchange delisted the coin due to "51% attacks". > The total value of activity on the chain isn't equivalent to the value available to attackers in an attack, especially to the extent that attacking undermines the value of the system as a whole. Yes, another thing to go into that formula. Centralised exchanges and other automated systems interacting with cryptocurrencies really mess things up. They provide a lot of liquidity open to double-spend attacks. In a world were cryptocurrencies were only used as the peer-to-peer cash in the way a lot of it's early proponents advocated for (and still do), where everyone can manually verify their transactions and establish trust, then your suggestions do work somewhat better. But in the highly centralised nightmare that cryptocurrencies have become... well it changes numbers going into the formula. > That's also not clear to me, there are an awful lot of idle two generation old miners out there at any given time. When hashrate is trending up, I suspect the two generation old miners are slow enough to not worry about. But the economics might change as block reward subsidies drop, or if the potential profit of double-spends goes up. > Perhaps it isn't but if you really strongly believe that there are about 1000 cryptocurrencies that you should be figuring out how to short hard instead of posting on HN. :P I mean, I considered it. Hence the research. But these were really, really, shit shitcoins, not listed on exchanges with futures markets, so the only real way to profit was to execute the 51% attacks myself, and I didn't want to go down the black hat path. Also, I moved onto other hobbies.

by phire 3y ago

nullc 3y ago

Those are all fair arguments you can make, but none of them make a case that the current subsidy level is necessary. For a system that is actually in use, the consequences of these attacks you imagine are difficult to reason about. It's not like the system stops existing if some attacker attacks-- the attacker can reverse some of their own transactions near the chain tip. Recipients can adapt to the behavior by requiring more confirmations, restricting who and how they attack, etc. It's difficult to generalize from what's seen in those "random shitcoins" (to use your phrase) since most of them have little to no economic use, so an attack may have an oversized or undersized impact depending on pure chaos: their value is purely speculative so the narrative around the attack determines the outcome more than anything. Since you've said you tracked these things presumably you're aware that in many cases their value goes up after they're attacked. > but only because most of the SHA256 miners are already mining bitcoin That's also not clear to me, there are an awful lot of idle two generation old miners out there at any given time. > And what if Bitcoin becomes actually successful? Becomes the backbone of the world economy and doing trillions per day in transactions? The total value of activity on the chain isn't equivalent to the value available to attackers in an attack, especially to the extent that attacking undermines the value of the system as a whole. In any case, I wasn't (and am not now) trying to take a strong position here-- I would say that there are a lot of unknowns but that I think the position you were taking that the current fee level isn't adequate doesn't really pass a sniff test for me. Perhaps it isn't but if you really strongly believe that there are about 1000 cryptocurrencies that you should be figuring out how to short hard instead of posting on HN. :P

phire 3y ago
> but none of them make a case that the current subsidy level is necessary. I wasn't actually arguing that, yes there are a bunch of unknowns. I'm only making the argument that there is a clear formula connecting mining rewards, potential double-spend profits and the cost of attacking a PoW coin. I don't care enough to actually derive the formula and plug numbers in it to answer that question. > Recipients can adapt to the behavior by requiring more confirmations, restricting who and how they attack More confirmations don't really work as a solution. The cost to hold a double-spend open is known, so we could try requiring enough confirmations to make such a double-spend unprofitable, but the attacker can execute multiple double-spends in parallel to spread their costs. For large transactions (relative to the hashrate of that coin), the confirmations can can get quite high, which destroys the usefulness of cryptocurrency for fast and easy transactions. For one of the "random shitcoins" I was tracking, I found the centralised exchange I suspect the attacker was targeting (presumably trading them to another currency and then withdrawing). The exchange had set really high numbers of confirmations (2-5 days worth of blocks) but the because these coins were so cheap to attack, the attacker had no problem doing reorgs that large. And then the exchange delisted the coin due to "51% attacks". > The total value of activity on the chain isn't equivalent to the value available to attackers in an attack, especially to the extent that attacking undermines the value of the system as a whole. Yes, another thing to go into that formula. Centralised exchanges and other automated systems interacting with cryptocurrencies really mess things up. They provide a lot of liquidity open to double-spend attacks. In a world were cryptocurrencies were only used as the peer-to-peer cash in the way a lot of it's early proponents advocated for (and still do), where everyone can manually verify their transactions and establish trust, then your suggestions do work somewhat better. But in the highly centralised nightmare that cryptocurrencies have become... well it changes numbers going into the formula. > That's also not clear to me, there are an awful lot of idle two generation old miners out there at any given time. When hashrate is trending up, I suspect the two generation old miners are slow enough to not worry about. But the economics might change as block reward subsidies drop, or if the potential profit of double-spends goes up. > Perhaps it isn't but if you really strongly believe that there are about 1000 cryptocurrencies that you should be figuring out how to short hard instead of posting on HN. :P I mean, I considered it. Hence the research. But these were really, really, shit shitcoins, not listed on exchanges with futures markets, so the only real way to profit was to execute the 51% attacks myself, and I didn't want to go down the black hat path. Also, I moved onto other hobbies.