3 ms·

> The choice of cryptography to use determines the performance and guarantees of the blockchain. For example, Dogecoin blockchain clones Bitcoin but using simpl

by FabHK 5y ago

> The choice of cryptography to use determines the performance and guarantees of the blockchain. For example, Dogecoin blockchain clones Bitcoin but using simpler cryptographic functions to increase transaction throughput; the mining in Dogecoin is based on SCRYPT which is faster and easier to run than SHA256 used in Bitcoin. This, however, results in weaker security, less robust to attacks by dishonest nodes. 1. Isn't scrypt a key-derivation-function, and thus (by design) much slower and more work intensive than SHA-256 (a cryptographic hash)?? 2. Isn't anyway difficulty adjusted depending on the available hash power (such that a target block rate is reached), so that "faster and easier to run than SHA256" does not make any sense at all? Have the authors even understood PoW?

rcxdude 5y ago
Yes, this is complete nonsense. Block time and difficulty can both be adjusted independently of your work function (in fact it's a critical property of the work function that such adjustment is possible). The difficulty is automatically adjusted to keep to the target average block time of the network as the amount of mining increases and decreses. Dogecoin simply has a lower target block time than bitcoin (which, given equal mining power, would mean less difficulty), but also has a much lower difficulty because there's much less mining power on the network due to the relatively lower mining rewards (a function of how much currency the network awards miners at the moment and the price of the coin). Less mining power does roughly correlate to an easier attack on the network, but this is mostly independent of the technical decisions made by a given blockchain (there is some impact if e.g. the work function is sufficiently different that hardware from a larger network cannot be repurposed to mount an attack on a smaller network, e.g. using scrypt means bitcoin ASICs cannot mine on the network, making an attack much less easy to carry out. Ethereum's work function is designed to be memory bandwidth instead of compute limited, making GPUs generally optimal, and monero uses a funky algorithm designed to run best on general purpose CPUs). I will also note (and this is a point I don't see made often), that there is not really a specific mechanism in most blockchains to set the total mining rewards (and thus the total amount of effort mining which is incentivised) to an optimal value. It might be entirely excessive, in which case the network is secure but even more inefficient, or it might be insufficient, there is nothing which really sets this to a particular value. This is in part because it's hard to know exactly what level of mining is necessary (what's the difficulty required to fend off all attackers? probably depends on what they could gain from an attack, which is hard to assess), and because it depends on the price of the coin, which is something the network does not really have a means to assess within itself. (This is one of the reasons why I think bitcoin as a "long term store of value" does not make sense. The deflationary design requires ever-diminishing mining rewards, but the network will still require significant mining effort to secure it, especially if the amount of value stored keeps going up. In the 'bitcoin as currency' model, the transaction fees would keep the incentive present, but with a relatively low rate of transactions you will need extremely high transaction fees to secure the network, making it potentially quite an expensive store of value to maintain.
- FabHK 5y ago
  > The deflationary design requires ever-diminishing mining rewards, but the network will still require significant mining effort to secure it, especially if the amount of value stored keeps going up. Agreed - it stands to reason that as BTCUSD rises, more miners will come in until the electricity and electronics wasted per day reaches a certain, relatively fixed proportion of the daily block rewards (900 BTC/d until 2024, currently about USD 35m per day). With that, BTC now consumes about 0.5% of world electricity. If BTC goes 10x, its electricity consumption could approach 5% of world electricity consumption - insanity.
  - rcxdude 5y ago
    My point is it also goes in the other direction (not to understate how ridiculous bitcon's energy usage is, though): the reward in BTC reduces exponentially over time (this is why there is a finite amount of bitcoin that will ever exist). Even if the value goes all the way to the moon, the rewards will eventually peter out (though it might be on the scale of many decades: the reward halves every 4 years or so). This means the cost of sustaining the 'value' stored in bitcoin will need to fall on those who own that value (lest it be stolen), and that could be very expensive compared to other forms of value storage. Bitcoin could very well be simultanously ridiculously inefficient and yet have insufficient incentive to keep mining at a secure level long-term.
    - FabHK 5y ago
      Right. Including the invisible fees (due to increased BTC supply accruing to the miners), a transaction right now costs around USD 100. As the mining reward drops, either miners make less (good, they'd stop wasting), or fees paid by the user would have to increase substantially. If users don't pay up, there'd be much less hash power devoted to the system, rendering it insecure.

by FabHK 5y ago

rcxdude 5y ago

Yes, this is complete nonsense. Block time and difficulty can both be adjusted independently of your work function (in fact it's a critical property of the work function that such adjustment is possible). The difficulty is automatically adjusted to keep to the target average block time of the network as the amount of mining increases and decreses. Dogecoin simply has a lower target block time than bitcoin (which, given equal mining power, would mean less difficulty), but also has a much lower difficulty because there's much less mining power on the network due to the relatively lower mining rewards (a function of how much currency the network awards miners at the moment and the price of the coin). Less mining power does roughly correlate to an easier attack on the network, but this is mostly independent of the technical decisions made by a given blockchain (there is some impact if e.g. the work function is sufficiently different that hardware from a larger network cannot be repurposed to mount an attack on a smaller network, e.g. using scrypt means bitcoin ASICs cannot mine on the network, making an attack much less easy to carry out. Ethereum's work function is designed to be memory bandwidth instead of compute limited, making GPUs generally optimal, and monero uses a funky algorithm designed to run best on general purpose CPUs). I will also note (and this is a point I don't see made often), that there is not really a specific mechanism in most blockchains to set the total mining rewards (and thus the total amount of effort mining which is incentivised) to an optimal value. It might be entirely excessive, in which case the network is secure but even more inefficient, or it might be insufficient, there is nothing which really sets this to a particular value. This is in part because it's hard to know exactly what level of mining is necessary (what's the difficulty required to fend off all attackers? probably depends on what they could gain from an attack, which is hard to assess), and because it depends on the price of the coin, which is something the network does not really have a means to assess within itself. (This is one of the reasons why I think bitcoin as a "long term store of value" does not make sense. The deflationary design requires ever-diminishing mining rewards, but the network will still require significant mining effort to secure it, especially if the amount of value stored keeps going up. In the 'bitcoin as currency' model, the transaction fees would keep the incentive present, but with a relatively low rate of transactions you will need extremely high transaction fees to secure the network, making it potentially quite an expensive store of value to maintain.

FabHK 5y ago
> The deflationary design requires ever-diminishing mining rewards, but the network will still require significant mining effort to secure it, especially if the amount of value stored keeps going up. Agreed - it stands to reason that as BTCUSD rises, more miners will come in until the electricity and electronics wasted per day reaches a certain, relatively fixed proportion of the daily block rewards (900 BTC/d until 2024, currently about USD 35m per day). With that, BTC now consumes about 0.5% of world electricity. If BTC goes 10x, its electricity consumption could approach 5% of world electricity consumption - insanity.
- rcxdude 5y ago
  My point is it also goes in the other direction (not to understate how ridiculous bitcon's energy usage is, though): the reward in BTC reduces exponentially over time (this is why there is a finite amount of bitcoin that will ever exist). Even if the value goes all the way to the moon, the rewards will eventually peter out (though it might be on the scale of many decades: the reward halves every 4 years or so). This means the cost of sustaining the 'value' stored in bitcoin will need to fall on those who own that value (lest it be stolen), and that could be very expensive compared to other forms of value storage. Bitcoin could very well be simultanously ridiculously inefficient and yet have insufficient incentive to keep mining at a secure level long-term.
  - FabHK 5y ago
    Right. Including the invisible fees (due to increased BTC supply accruing to the miners), a transaction right now costs around USD 100. As the mining reward drops, either miners make less (good, they'd stop wasting), or fees paid by the user would have to increase substantially. If users don't pay up, there'd be much less hash power devoted to the system, rendering it insecure.