This lecture covers a critical component of certain digital currencies which is mining puzzles. This section covers what are the requirements of these puzzles to be good puzzles. Theoretically you could replace the word “puzzle” for any other word and it would have the same meaning. The use of the word puzzle is not within Satoshi Nakamoto’s seminal paper. He merely writes, “The proof-of-work involves scanning for a value that when hashed, such as with SHA-256, the hash begins with a number of zero bits. The average work required is exponential in the number of zero bits required and can be verified by executing a single hash”.
Questions answered in this Post:
-
-
- What is a puzzle in this case?
- Why are the important?
- What are the requirements?
- What is hash power?
Why are they important?
Andrew Miller states that “mining puzzles determine the incentive system in Bitcoin”. Thus whatever puzzle is chosen needs to ensure miner participation. In addition, if shortcuts are found, miners will ultimately choose the most efficient path and thus remove arbitrage situations that may occur. If puzzles are one of the few mechanisms that exist to maintain the protocol, it needs to be at the core and encapsulate the work. Miners are not incentivized to “do good” just to ensure the health of the system if they will not be compensated.
What are the requirements?
The first two requirements were discussed in earlier lectures and are straightforward. The puzzle needs to be easy to verify and have an adjustable difficulty setup. Bitcoin’s proof of work puzzle is easy to verify since once a valid value is found, all miners can just use that value with the hashing function and determine whether is it a small enough value (has the sufficient number of zero bits). The puzzles get solved at a known reasonable rate ensuring long term participation. Ten minutes is the current rate for Bitcoin. The puzzle is also adjustable because the value looked up is in a range which can be made smaller or larger. As long as the difficulty is set and shared with all miners, this now gives you the adjustable difficulty setup.
Another new requirements is that the probability of winning is based on hash power. Simply stated, big fish with more hardware have a higher chance of winning. Small players still have a probability to win but it may be smaller. The lecturer makes a distinction using a sequential proof of work which is marked as a bad puzzle. If the puzzle is more like who can complete N steps faster wins, then likely you’ll have a single party who has the fastest computation and always wins. Instead a good puzzle should have a weighted sample and they also bring up the term “progress-free”. Bitcoin is different in that the small and big miners are all computing and while big miners, those with more hardware, have a higher chance of winning, it’s not 100%.
-
-
I like to think of it more like a dice game where the larger miner has control of faces 1-5 and only 6 is held by the small miner. The small miner still has ~17% chance ie some non-zero chance of winning at every block. In addition, every roll of the die the percentage theoretically stays the same in a perfect world. The die is merely a metaphor but hopefully that point makes sense.
What is hash power?
Throughout the lecture, Miller used the term hash power. At some point, he substituted hash power for hardware. Now, I was still unclear on the term so I decided to take a quick trip through Google. I’ve found hash power can be used interchangeably with hash rate. Hash rate is some measurement per second that a miner does work. Examples hash rates are of the order of 16 TH/s (one trillion) hashes per second for mining rigs. This individual hash rate can be compared to the overall network which can be seen in block explorers. The probability creates gives the miner what chance they have of finding the next block as well as some expected value. Here’s a link to one chart at blockchain.com.
Wrap up
This lecture was quick. Really excited to find out more alternative mining puzzles that he alluded to.