Computation on the Blockchain- technical limitations?

[luckybit]

You can either go the VM route as Ethereum is doing and in that case gas is how you avoid the halting problem with your Turing completeness or you can go the decidable logic route as Tauchain is doing which means you have no halting problem because you are not Turing complete.

In my opinion the Tauchain decidable logic route is the best approach because it's not a quick hack, it's better for security, it's much more scaleable, but then it takes a lot more effort to do it the right way which is why Ethereum even with mistakes is at least here now.

Digital Asset Holdings recently bought a smart contract company called Elevence which is interesting in that it appears to be going the decidable route with smart contracts. My conclusion is blockchain computation is likely and is also easily scalable depending on the approach. Latency will not be much of an issue if the incentives are set right and there is a supply and demand.

http://www.coindesk.com/digital-asset-acquires-elevence/

[xh3]

I'm not suggesting that this be a bitshares project,  I'm just asking the Devs what they think the technical limitations of computation on a blockchain are.

So with the latency limitations, one could theoretically build clusters of blockchain based VMs inside of a single computer, and have those chains periodically register a block with some master chain.  The master chain could spit out genesis blocks for these fast, small chains when computational services are needed.  Damn I wish I knew how to build this.

[luckybit]

I think it would be cool to see a project spring up that provides a substantial amount of computation on the blockchain(like a souped-up ethereum).  to my novice computer-science aware mind, it seems like this could be done parallel blockchains that run a VM with perhaps other sidechains that could be spun up like threads, and the whole thing could have a ~1 millisecond block production time.

@abit
@theoreticalbts
@bytemaster

do you think such a project is feasible? would it be useful?
what do you think the limits of computation on the block-chain currently are?

Not feasible with current Bitshares design nor is it sensible. Computation should take place off chain. This would have to be a separate but linked project to Bitshares because it will take a massive amount of coding to get it right and Bitshares wont want to pay for it.

[xeroc]

you can have 1ms block production if you have spread the block producers across a small area .. on one computer.
Once you increase the distance between block producers, you need to add network latency .. if you scale it to the size of the planet, you end up with a fundamental limit due to speed of light and the size of the planet.

[xh3]

I think it would be cool to see a project spring up that provides a substantial amount of computation on the blockchain(like a souped-up ethereum).  to my novice computer-science aware mind, it seems like this could be done parallel blockchains that run a VM with perhaps other sidechains that could be spun up like threads, and the whole thing could have a ~1 millisecond block production time.

@abit
@theoreticalbts
@bytemaster

do you think such a project is feasible? would it be useful?
what do you think the limits of computation on the block-chain currently are?