Bitcoin Forum

Bitcoin => Bitcoin Discussion => Topic started by: fivebells on November 04, 2011, 12:50:37 AM



Title: Alternative proof-of-work if fraud weren't an issue
Post by: fivebells on November 04, 2011, 12:50:37 AM
There's a lot of work, ingenuity and computational resources devoted to computation of the hashes which lie at the core of the bitcoin proof-of-work protection.  I do not wish to criticize that, I understand its value.  But I would be interested to hear your fantasies of what you would do with those resources if they could be turned to arbitrary computations.  Would you devote the cycles to Folding/SETI@Home, sell them to people trying to crack password hashes, try to train a machine-learning algorithm to predict stock market prices?  Go wild, I'd really like to hear people's favorite suggestions. 


Title: Re: Alternative proof-of-work if fraud weren't an issue
Post by: MoonShadow on November 04, 2011, 01:01:48 AM
I'd heat my garage.


Title: Re: Alternative proof-of-work if fraud weren't an issue
Post by: ElectricMucus on November 04, 2011, 01:04:31 AM
Decentralized Anonymous Cloud Computing, payed for with bitcoin, nothing else needed.

The press, and govt would go nuts if this ever happens ;)


Title: Re: Alternative proof-of-work if fraud weren't an issue
Post by: fivebells on November 04, 2011, 01:14:13 AM
Decentralized Anonymous Cloud Computing, payed for with bitcoin, nothing else needed.
  What if you had to choose a particular algorithm up-front, so that people could run the client without security concerns?


Title: Re: Alternative proof-of-work if fraud weren't an issue
Post by: ElectricMucus on November 04, 2011, 01:22:50 AM
Decentralized Anonymous Cloud Computing, payed for with bitcoin, nothing else needed.
  What if you had to choose a particular algorithm up-front, so that people could run the client without security concerns?
Enumeration of the instruction set with a cypher, probably waste alot of computational resources for privacy and security, but if it is practically impossible to trace a application it would be worth it.

Also see this thread:
https://bitcointalk.org/index.php?topic=34916.0


Title: Re: Alternative proof-of-work if fraud weren't an issue
Post by: fivebells on November 04, 2011, 02:34:05 AM
How about useful calculations, or which people would pay money for?


Title: Re: Alternative proof-of-work if fraud weren't an issue
Post by: ElectricMucus on November 04, 2011, 02:36:47 AM
How useful would be my suggested service?

Don't you think people would pay money to host webapps on a platform which cannot go down by design?


Title: Re: Alternative proof-of-work if fraud weren't an issue
Post by: fivebells on November 04, 2011, 10:51:27 AM
Oh, I didn't realize you were still talking about allowing arbitrary computation.  Yes, that would be extremely useful, but it would still be hard to establish as secure.

Pretty much every consumer computer in the world already has restricted-execution environments for running untrusted code from the internet.  They're called Java and Javascript, and they're a bountiful source of security (http://www.kb.cert.org/vuls/byid?searchview&query=java) vulnerabilities (http://www.kb.cert.org/vuls/byid?searchview&query=javascript), not so much because they're sloppily implemented, but because in a hostile context it's extremely difficult to establish whether a given arbitrary computation is safe to run.  Do the restricted-execution environments cited in that earlier thread provide better security guarantees than current java and javascript implementations do?


Title: Re: Alternative proof-of-work if fraud weren't an issue
Post by: ElectricMucus on November 04, 2011, 08:42:10 PM
Yes, the approach I am suggesting provides an even larger amount of security. Have you watched the video I linked in the other thread?
The plan is to take this approach described there and extend it to a decentralized cluster.

This is mainly thought to provide security as for the whole system, or the cloud if you will. On the issue of personal security there would be the advantage that code can be only written for the machine if the enumeration of the instruction set is known. Since this is kept secret it can only be derived if the attacker can corner the machine with his own machine, similar a tor attack would work.

tl;dr I want to make a cloud computing version of tor, paid for with btc.


Title: Re: Alternative proof-of-work if fraud weren't an issue
Post by: Red Emerald on November 04, 2011, 08:50:46 PM
I would really like for my miner to somehow benefit Folding@Home, but all of the ways that I have thought to mix them results in slower bitcoin hashing.


Title: Re: Alternative proof-of-work if fraud weren't an issue
Post by: fivebells on November 05, 2011, 01:53:22 AM
I researched the restricted-execution options today, and you are right, it can be relatively secure.  However, it turns out that allowing arbitrary computations is infeasible for my purposes.  Still looking for good distributed-computation problems.


Title: Re: Alternative proof-of-work if fraud weren't an issue
Post by: wareen on November 05, 2011, 10:42:01 AM
Somewhat related to this thread: I'd totally pay for somebody's hashing capacity if he/she can prove that it is used for working on Folding@home (or similar) all the time unless Bitcoin needs it to quickly fend off a potential 51% attack.

This would be a kind of insurance service for the Bitcoin network because it makes transactions safer while at the same time contributing to a worthy cause.


Title: Re: Alternative proof-of-work if fraud weren't an issue
Post by: jago25_98 on November 06, 2011, 12:43:08 AM
Yes, I'd like to see the computational proof used to reduce the tenancy of money to clump into rich and poor areas.

It sounds far fetched but I think it's possible. We can map the network and we can show how networks tend to clump in areas. Investors predict this pattern in order to predict the market. If we can have that working in Bitcoin then we can solve a great many of the problems in life.