5 ms·

I've always thought "proof of work" in cryptocurrencies was a bit of a misnomer since the the "work" is useless. Could resources be used for productive work in

by graeham 13y ago

I've always thought "proof of work" in cryptocurrencies was a bit of a misnomer since the the "work" is useless. Could resources be used for productive work instead, for example distributed computation? I'm thinking something similar to SETI@home, but with currency rewards for work done. Perhaps this would also require a central issuer.

ansible 13y ago
There is gridcoin [1], but it is just using standard mining software for litecoin, then checking to see if you're running the BIONC program, and giving you extra credit. I believe that the minute it gets popular, people will fake running the BIONC software. We need something where the proof of work is intimately tied to the actual BIONC calculations. [1] http://www.gridcoin.us/ http://www.gridcoin.us/
lloeki 13y ago
Unless I missed something[0], the work is actually useful in cryptocurrencies: it's actually bruteforcing a crypto function to find a solution for the block. Once the solution is found, the block, which contains actual transactions, is known to be valid, and the block is said to be mined. The very system that generates money is the system that validates the transactions. The miner gets the nominal fee of 25 BTC per block, plus any fees included in the transactions. [0]: http://www.righto.com/2014/02/bitcoins-hard-way-using-raw-bitcoin.html http://www.righto.com/2014/02/bitcoins-hard-way-using-raw-bi...
- aninhumer 13y ago
  It's useful in the sense that it serves a purpose, but the results themselves are essentially meaningless. What people are asking is whether it's possible to adapt some kind of meaningful computation to serve the same function.
GavinB 13y ago
Making a monetary system function is important work. From the Federal Reserve chairman to the sysadmins running wire transfer servers to the crew running the printing presses to the security guards who transport cash, there's a huge amount of work involved in running current fiat money systems. You can dislike cryptocurrencies for all sorts of other reasons, but that fact that some resources are consumed to keep it running really is going to be a feature of any sort of monetary system.
- aninhumer 13y ago
  I think you're making a rather tenuous argument here. Just because existing monetary systems involve work, and a new alternative system also involves work, doesn't imply that such systems always must.
- mushishi 13y ago
  Could you give some estimates how much energy is spent on having current fiat currencies, and how much energy would be spent if similar amount of transactions were used for cryptocurrency. I always wonder why people don't discuss this, or when do, no estimates are given.
- dllthomas 13y ago
  My worry, regarding resources, is not that any are burned, but that it's an arms race where effort to attack is O(effort to defend), which means we always need to collectively burning more resources than attackers could and we don't even get the tremendous discount afforded us by crypto in other spheres (brute forcing encryption takes effort of O(2^effort to defend), for example). If we get more efficient algorithms or processes or technology in traditional finance, the resources we spend can shrink - not so, with proof-of-work. Of course, with physical attack there is always an arms race, but bitcoin is still vulnerable to physical attack.
slaxo 13y ago
Primecoin (http://en.wikipedia.org/wiki/Primecoin http://en.wikipedia.org/wiki/Primecoin) searches for chains of primes as its POW
aninhumer 13y ago
The structure of existing cryptocurrencies does impose some restrictions on what kind of work can be used. Most notably, it needs to have reasonably predictable difficulty, and be easily verifiable. As you suggest, one thing you could do to tackle the latter problem is to have a centralised entity to verify work. This could be implemented in existing systems by making the "work" signing data with the issuer's private key, and having the real work be negotiated out-of-band. This mechanism could be extended allow multiple independent issuers.
tromp 13y ago
Cuckoo Cycle (https://github.com/tromp/cuckooh https://github.com/tromp/cuckooh) searches for cycles in a random bipartite graph (induced by Cuckoo Hashing). Instead of requiring a lot of computation, it requires a lot of waiting due to latency of random memory access.
rowanseymour 13y ago
There's http://curecoin.us/ http://curecoin.us/ which hasn't been released yet but plans to make protein folding the proof of work and so help with research into cures for various diseases. Remember though that the usual crypto currency proof-of-work serves a purpose to verify transactions.

by graeham 13y ago

ansible 13y ago

There is gridcoin [1], but it is just using standard mining software for litecoin, then checking to see if you're running the BIONC program, and giving you extra credit. I believe that the minute it gets popular, people will fake running the BIONC software. We need something where the proof of work is intimately tied to the actual BIONC calculations. [1] http://www.gridcoin.us/ http://www.gridcoin.us/

lloeki 13y ago

Unless I missed something[0], the work is actually useful in cryptocurrencies: it's actually bruteforcing a crypto function to find a solution for the block. Once the solution is found, the block, which contains actual transactions, is known to be valid, and the block is said to be mined. The very system that generates money is the system that validates the transactions. The miner gets the nominal fee of 25 BTC per block, plus any fees included in the transactions. [0]: http://www.righto.com/2014/02/bitcoins-hard-way-using-raw-bitcoin.html http://www.righto.com/2014/02/bitcoins-hard-way-using-raw-bi...

aninhumer 13y ago
It's useful in the sense that it serves a purpose, but the results themselves are essentially meaningless. What people are asking is whether it's possible to adapt some kind of meaningful computation to serve the same function.

GavinB 13y ago

Making a monetary system function is important work. From the Federal Reserve chairman to the sysadmins running wire transfer servers to the crew running the printing presses to the security guards who transport cash, there's a huge amount of work involved in running current fiat money systems. You can dislike cryptocurrencies for all sorts of other reasons, but that fact that some resources are consumed to keep it running really is going to be a feature of any sort of monetary system.

aninhumer 13y ago
I think you're making a rather tenuous argument here. Just because existing monetary systems involve work, and a new alternative system also involves work, doesn't imply that such systems always must.
mushishi 13y ago
Could you give some estimates how much energy is spent on having current fiat currencies, and how much energy would be spent if similar amount of transactions were used for cryptocurrency. I always wonder why people don't discuss this, or when do, no estimates are given.
dllthomas 13y ago
My worry, regarding resources, is not that any are burned, but that it's an arms race where effort to attack is O(effort to defend), which means we always need to collectively burning more resources than attackers could and we don't even get the tremendous discount afforded us by crypto in other spheres (brute forcing encryption takes effort of O(2^effort to defend), for example). If we get more efficient algorithms or processes or technology in traditional finance, the resources we spend can shrink - not so, with proof-of-work. Of course, with physical attack there is always an arms race, but bitcoin is still vulnerable to physical attack.

slaxo 13y ago

Primecoin (http://en.wikipedia.org/wiki/Primecoin http://en.wikipedia.org/wiki/Primecoin) searches for chains of primes as its POW

aninhumer 13y ago

The structure of existing cryptocurrencies does impose some restrictions on what kind of work can be used. Most notably, it needs to have reasonably predictable difficulty, and be easily verifiable. As you suggest, one thing you could do to tackle the latter problem is to have a centralised entity to verify work. This could be implemented in existing systems by making the "work" signing data with the issuer's private key, and having the real work be negotiated out-of-band. This mechanism could be extended allow multiple independent issuers.

tromp 13y ago

Cuckoo Cycle (https://github.com/tromp/cuckooh https://github.com/tromp/cuckooh) searches for cycles in a random bipartite graph (induced by Cuckoo Hashing). Instead of requiring a lot of computation, it requires a lot of waiting due to latency of random memory access.

rowanseymour 13y ago

There's http://curecoin.us/ http://curecoin.us/ which hasn't been released yet but plans to make protein folding the proof of work and so help with research into cures for various diseases. Remember though that the usual crypto currency proof-of-work serves a purpose to verify transactions.