POS - 'Long Range Attack' fix ..

[spartacusrex]

Did try and discuss this in the technical forum, but no bites..

https://bitcointalk.org/index.php?topic=997125.msg12307016#msg12307016

Basically :

POW mainchain, with POS sidechain.

POS sidechain can store the checkpoints required to show which chain is real on the un-fakeable POW chain. Decentralised checkpointing.

This means no 'Weak Subjectivity' as Vitalik puts it.. so the first time you connect to the POS network you now have a reference block hash, checkpoint, to prove which chain is the real one.

Thoughts ?

[monsterer]

Did try and discuss this in the technical forum, but no bites..

https://bitcointalk.org/index.php?topic=997125.msg12307016#msg12307016

Basically :

POW mainchain, with POS sidechain.

POS sidechain can store the checkpoints required to show which chain is real on the un-fakeable POW chain. Decentralised checkpointing.

This means no 'Weak Subjectivity' as Vitalik puts it.. so the first time you connect to the POS network you now have a reference block hash, checkpoint, to prove which chain is the real one.

Thoughts ?

What is then the point of the POS part?

[spartacusrex]

Scalability ? Extensibilty ? Personal POS chains per company ?

You could have ONE almighty POW chain, with perhaps a smallish block size limit  ,  that is used to ground Thousands of POS side chains, each doing something funky / different.

Want to test a quantum secure chain ?

POS ethereum ?

Anything really, with the POS chains not having to worry about expensive mining hardware.

[monsterer]

Scalability ? Extensibilty ? Personal POS chains per company ?

You could have ONE almighty POW chain, with perhaps a smallish block size limit  ,  that is used to ground Thousands of POS side chains, each doing something funky / different.

Want to test a quantum secure chain ?

POS ethereum ?

Anything really, with the POS chains not having to worry about expensive mining hardware.

All those advantages can be had by throwing away the POS and just having a plain sidechain.

[spartacusrex]

Well, you still have to secure a sidechain ? A sidechain doesn't just get all the security of the main chain. Far from it.

So you are going to need to run the side chain as a federated chain, POW chain, or POS chain.

I always feel that although POW 'works', I wouldn't want to run my company governance off of that..

It's definitely a situation where POS makes much more sense. 

Especially Proof of Virtual Miner  : https://bitcointalk.org/index.php?topic=1143191.0 

You can keep your share in the company without someone 'paying' for mining equipment that makes your share negligible. The stakeholders have there say. Like a normal company.

If only we could fix the various attack vectors..

[monsterer]

Well, you still have to secure a sidechain ? A sidechain doesn't just get all the security of the main chain. Far from it.

I was thinking of a coloured coins. However, isn't the similarity important?

Maybe I'm missing the point of sidechains, but if you delegate network security to a POW chain, what do you gain by having a redundant consensus model in the side chain?

[spartacusrex]

MONSTERERERERR..! Thanks for chipping in..

As far as my understanding goes, SideChains and Coloured coins are different animals. Although maybe of the same genus.. 

Sidechains allow you to interact with the main chain, on a side chain, without having 'trust'. You can jump to a sidechain, dance a jig / interact / etc , and then return to the main chain. In a Semi-Atomic-Trustless fashion. ( Depends on how the sidechain is secured. Currently the 'Elements' Sidechain is federated, so you do have to trust the 'federators', although they can't 'screw' you in a hidden fashion. Their treachery would be public! )

In this system, you are only delegating the 'POS Decentralised Checkpointing' to the main POW chain.

This means that the POW chain is not used to determine the outcome on each individual POS chain. Just to store each POS chains latest checkpoint.

Then each POS chain can perform it's task, ( an exchange, company governance system, super fast confirmation time chain, torrent file link host, etc etc),  in the knowledge that it is not susceptible to a long range attack.

Once your POS chain is compatible with the POW main chain checkpoint mechanism, you can do ANYTHING you like on a POS sidechain. Without ever needing to change the main POW chain. No Main-Chain Hard Forks..  everererer ..

[Crestington]

MONSTERERERERR..! Thanks for chipping in..

As far as my understanding goes, SideChains and Coloured coins are different animals. Although maybe of the same genus.. 

Sidechains allow you to interact with the main chain, on a side chain, without having 'trust'. You can jump to a sidechain, dance a jig / interact / etc , and then return to the main chain. In a Semi-Atomic-Trustless fashion. ( Depends on how the sidechain is secured. Currently the 'Elements' Sidechain is federated, so you do have to trust the 'federators', although they can't 'screw' you in a hidden fashion. Their treachery would be public! )

In this system, you are only delegating the 'POS Decentralised Checkpointing' to the main POW chain.

This means that the POW chain is not used to determine the outcome on each individual POS chain. Just to store each POS chains latest checkpoint.

Then each POS chain can perform it's task, ( an exchange, company governance system, super fast confirmation time chain, torrent file link host, etc etc),  in the knowledge that it is not susceptible to a long range attack.

Once your POS chain is compatible with the POW main chain checkpoint mechanism, you can do ANYTHING you like on a POS sidechain. Without ever needing to change the main POW chain. No Main-Chain Hard Forks..  everererer ..

Than that means that the cost of the Checkpoints would be the minimum fee of the POW chain to sign the transaction of the latest checkpoint. What about size of the POW chain and connecting to it without storing the full chain? SPV node system? How would that effect memory usage?

[spartacusrex]

Than that means that the cost of the Checkpoints would be the minimum fee of the POW chain to sign the transaction of the latest checkpoint. What about size of the POW chain and connecting to it without storing the full chain? SPV node system? How would that effect memory usage?

Hmm.. not sure.. let's think.

As far as the checkpointing. Assuming you would want to store a checkpoint every 24hrs for you POS chain, and that there were 10,000 different POS chains running of it, that would be, well, tiny. Easily fits into 1mb blocks..   

Other txns could be cross-chain transfers, going from one side chain to another. This would need to go through the main chain, but it would not be a daily occurrence for most users. I can't really say how much of this would happen without knowing what chains were running.

The main chain would really just be a clearing house. All the 'Action' would be happening on the various sidechains.

So it might be quite an expensive min fee on the main chain. But that's OK, as the fees on the side chains could be tiny. (I think this is the current position being pushed by blockstream..)

As far as running a node, I think it would be no different from now. Just that you could run the main chain node, or a side chain node, or both in some merged-mine hybrid.

Main Chain SPV clients would be fine, but as I said, I think most of the daily user action would be occurring on a side chain, so they would not be as prevalent as on Bitcoin.
 

[monsterer]

In this system, you are only delegating the 'POS Decentralised Checkpointing' to the main POW chain.

This means that the POW chain is not used to determine the outcome on each individual POS chain. Just to store each POS chains latest checkpoint.

Understood. I guess the primary negative aspect of a side chain is that you then need to install both blockchains in order to use a side chain?

[monsterer]

My personal preference in fixing nothing at stake, is to make something be at stake:

Have block producers compete for transaction fees (and block reward, potentially) by burning some of their stake. The guy who burns the most wins the transaction fees (and block reward).

[spartacusrex]

Understood. I guess the primary negative aspect of a side chain is that you then need to install both blockchains in order to use a side chain?

Yep - You do need to run both currently, not ALL the sidechains, but the main chain and sidechains could use UTXO compression and other tricks that are floating around to minimise blockchain size. (Something I think should become the default for all coins IMHO - I think Cryptonite was the first TINY coin chain ?)

I think POS has many MANY potential uses, which POW simply is not suited for. (And VICE VERSA - but maybe MORE uses for POS - just not as 'Currency'.. ;-p).

So if we need ONE POW chain, a simple decentralised timestamp server really, to fix a little thing like long range attacks, but then get all the benefits of POS, for thousands of separate chains, that sounds acceptable to me.

Now.. Just need to fix those pesky short range missiles..

[monsterer]

So if we need ONE POW chain, a simple decentralised timestamp server really, to fix a little thing like long range attacks, but then get all the benefits of POS, for thousands of separate chains, that sounds acceptable to me.

Now.. Just need to fix those pesky short range missiles..

At the risk of invoking all the POS vs POW demons, even if you had POS sidechains with POW checkpointing, you only prevent reorgs greater in length than the checkpoint interval. Rather than a series of bandaids on a broken system, you need a new system IMO.

[Crestington]

So if we need ONE POW chain, a simple decentralised timestamp server really, to fix a little thing like long range attacks, but then get all the benefits of POS, for thousands of separate chains, that sounds acceptable to me.

Now.. Just need to fix those pesky short range missiles..

At the risk of invoking all the POS vs POW demons, even if you had POS sidechains with POW checkpointing, you only prevent reorgs greater in length than the checkpoint interval. Rather than a series of bandaids on a broken system, you need a new system IMO.

Yes, then it really is no different than if you were to have something like POW/POS and having it merge mined with a larger POW chain like Bitcoin/Litecoin but the problem is to store the POW chain you are mining on to use as central checkpointing.

What about checkpointing on another POS chain?