This lecture describes puzzles that may be socially beneficial. Socially beneficial is a pretty loose definition. Questions answered in this Post:
- Can the wasted work from Bitcoin be recycled?
- Protein folding and Alien detection
- What is Primecoin?
- Recovering or repurposing wasted hardware: Permacoin
- Storage Based Puzzle
Recycling Mining Energy
Per the lecture, Bitcoin consumed 150 MW – 900 MW power approximately in mid-2014. A paper from June, 2019 mentioned that the annual carbon emissions is 22.0 to 22.9 MtCO2 in Nov 2018. They also mentioned the annual electricity consumption of Bitcoin to be 45.8 TWh ( terawatt-hour ). The benefits are obvious in that it may reduce energy costs and reduce the negative environment impact. Prof. Miller mentions some natural choices regarding protein folding and search for aliens. The first one is finding a low energy configuration. The second is find an anomalous region of a signal. They have similar characteristics to the current Bitcoin puzzle in that you’re trying to solve a problem that has a large problem space. There is a website called Fold.it that contains crowd-sourced gamified tasks that allow people to participate in scientific research. Proteins are composed of long chains of amino acids and they have a specific stable configuration. The specific shape means that some amino acids are near the center while others are far apart and this shape is the lower energy configuration they can keep. The hypothesis for why the game exists is that humans’ pattern-recognition and puzzle-solving abilities are more efficient than existing computer programs at pattern-folding tasks. Not sure if this is still an open question or if things like deep mind and deep learning change or disprove the hypothesis. The second option is searching for anomalies in space which can help detect extra terrestial life. Both the protein and detecting proteins are classified as crowdsource distributed computing problems. There is an article from Valentine’s day 2018 on how cryptocurrency mining is actually hampering the serach for ET life. SETI (Search for Extraterrestial Intelligence) uses GPU chips for their research. Radio-astronomers use them because they are processing large amounts of data and looking at many frequency channels to find the anomalous signal types. This issue is not unique to them given that video gamers have also mentioned they’ve now had a higher cost of GPUs. From my basic understanding of mining, when there was a market downterm, some miners turned off their mining rigs because it was less profitable. As the price of cryptocurrencies go up, that makes the mining more valuable. Prof. Miller brings up the notion that there is a centralized administrator for these problem sets and define the exploration space for participants. Bitcoin doesn’t have this and thus instances of the problem need to be auto-generated. There was not a clear way to generate these problem systematically to miners and thus while the problems are good, it’s not feasible to do it in a decentralized fashion. So what else is there?
Primecoin
Prof. Miller brings up Primecoin [[http://primecoin.io/]] which addresses these prevoius problems of needing a centralized resource to choose the problems. Primecoin aptly named involves finding large prime numbers. It’s consensus work is having nodes search for chains of prime numbers, specifically prime chains composed of Cunningham chains and bi-twin chains. A Cunningham chain is a chain of numbers where each number has the form 2^i*a + 1. Each is a large (probable) prime such that p is divisible by H (prev || mrkl_root || nonce). Probable prime classificaiton allows for efficient prime testing algorithms to be run as determining primality for very large numbers can be expensive. To date, most of the largest Cunningham chains have come from the Primecoin miners. He briefly mentions that it could be useful but then dismisses saying that the chains found are overkill. Thus, I’m unaware what it’s usecase is beyond helping science and looking at the distribution of primes. The Bitcoinwiki mentions that there may be a connection between the Riemann zeta function and prime distribution and relevant to other modern sciences. Alas, that is beyond what I was able to understand. The paper [http://primecoin.io/bin/primecoin-paper.pdf] similar to Bitcoin was published by a pseudonym Sunny King. It’s short in that it’s only 6 pages.
Permacoin: using storage-based puzzle
He mentions that upwards of 100 million dollars are spent on customized hardware. The hardware is so specialized that the investment is useless for other application. What if that wasn’t the case? Permacoin is mining with storage. You get massively distributed, replicated storage system. Then we get an example. There is a large file F that we’re storing and this F is chosen globally at the beginning by a trusted deal and then each user stores a random subset of the file. Thus this is where he introduces a new type of puzzle, storage-based puzzle.
What are the steps for the puzzle?
A Merkle tree is used where each leaf is a segment of the file F. Then miners will generate a keypair wich determines a random subset of file segments. Then for each mining attempt, a miner wil select a random nonce and then generate a hash h1 which is H(prev || mrkl_root|| PK || nonce). h1 select K segments from the subset. Then a second hash h2 is generated which is H(prev || mrkl_root || PK || nonce || F). Then they get the block if h2 is less than a certain target value. Thus the participants need to keep storing parts of the file. Permacoin adds a benefit for UTXO storage.
Wrap Up
Useful proof of work could be great. Its benefit must be pure public good. Other puzzles have been explored but none have truly captured mainstream adoption so far.
