search

dionyziz / popow

Research and implementation for non-interactive blockchain proofs of proofs of work
49 stars 6 forks source link

Compare our Definition with "DMMS" from the sidechains paper #12

Open amiller opened 7 years ago

amiller commented 7 years ago

Our definition is stronger, right? Can we point out the flaw precisely? What even exactly is the existing definition?

I think we basically should provide an alternate form of the definition, which focuses on the "expected work" property, and show that's too weak. https://lists.linuxfoundation.org/pipermail/bitcoin-dev/2014-March/004727.html

https://blockstream.com/sidechains.pdf

More related work: https://bitslog.wordpress.com/2014/04/25/smartspv-a-better-simplified-payment-verification-for-smartphones/

amiller commented 7 years ago

Here is another paper where DMMS is mentioned and given a partially formal definition:

https://download.wpsoftware.net/bitcoin/pos.pdf

solegga commented 7 years ago

the object that is relevant here is this "compressed DMMS" or compact SPV from the sidechains paper . This paper is pretty clear: it state it has limited scope in terms of security analysis. It identifies the variance problem that these POPOW like proofs can have and states explicitly "A detailed analysis of this problem and its possible solutions is out of scope for this document." The other papers do not even attempt to touch the security aspects of compact SPVs. so given this, I am not sure if it's worth to actually write another definition of POPOWs just do say it's a flawed concept in the way it was envisioned. I think we just state that we formalize for the first time what NIPOPOW is. it was left open in previous works and we, here, solve it. Also just to be clear: DMMS as defined above is not NiPOPOW. it is an attempt to formalize what POW might be in the context of blockchain protocols.

solegga commented 7 years ago

something we might point out is that "cumulative/ compressed DMMS" does not preserve its security properties in a straightforward way as one might expect at first sight. But it's not like the side-chains paper does not anticipate this... so i'd rather put this in the intro without going into many technical details about it.