http://cointelegraph.com/news/114577/chinese-mining-pools-propose-alternative-8-mb-block-size
EDIT: page 1337! gotta have something substantive on it :-)
i'm ok with it with the automatic expansion. but i'm alittle concerned with the miner voting for expansion. why just them and is voting really the way we want to go?
Yes the BIP 100 voting scheme seems a bit under developed at this point. The problem is that POW mining is the only way to prove independent existence, this is the basis of distributed consensus -- you can easily create a bunch of fake AWS "full" nodes for example -- 1000 nodes just forward requests to a single full node. And if you could solve the voting problem without using mining, you could use that solution to remove mining from consensus (i.e. from bitcoin).
One way to prove unique nodes is to require that every node use asymmetric functions to encode the data to be stored, with uniquifying "salt". This is essentially hashing but backwards.
For example, every node chooses a unique 256bit identifier (salt). Then for every 8 or 16 bits of blockchain data Di, it finds and stores Xi such that Hash(Xi + identifier + i) = Di (where "+" means "append").
This search happens continually as the blockchain is being created.
Now a client can validate the uniqueness of storage by another node by requesting random chunks of data from the node. The node responds with its ID and Xi0-i1. The client can then quickly validate this by running the hash function and comparing the result to its own blockchain.
A "full" node (one that is calculating this data and storing it as the blockchain is created) will be able to respond quickly where one that is not doing so will have to calculate the reverse hash function (essentially a random search like mining but easier because only 8 or 16 bits) on the fly.
Once you have this "proof of storage" you might be able to turn it into allowing node voting...
So someone rents hard disk space on Amazon AWS during the voting period and becomes many unique entities.
You were correct that there is no way to prove one computer, one vote. PoW doesn't work by voting. It works by random chance. The only problem PoW solves is consensus on ordering.
Ah I grow weary in this thread.
U repeated exactly what I said. And then didn't read my partial soln
Ahem, try reading again. You proposed proof-of-storage as a potential means to voting.
Prior proof-of-storage (https://bitcointalksearch.org/topic/proof-of-storage-to-make-distributed-resource-consumption-costly-310323) "wastes" storage like POW "wastes" CPU. It proves that storage exists. The proposal above proves that a unique copy of the blockchain is stored somewhere. This is much more valuable than proof that storage exists so it is perhaps better named proof-of-full-node. The reason you can't just rent hard disk space on AWS to fake this out is because you need to calculate the inverse hash for every byte of the blockchain, salted by your node id. The designer can choose how hard this is so it could take (say) a normal computer 6 months to do the entire 5 year blockchain. So the proof of storage is backed by POW which makes it hard to calculate the values on-the-fly. You could rent a bunch of AWS storage AND separately rent a supercomputer to individually calculate the hash-1 blockchain data, storing the result into AWS. But if you do that, you actually HAVE N individual copies of the blockchain so would receive N votes.
Because I am basically arguing with myself here I will now criticize the approach's limitations. It is possible that modifications to this approach could solve them:
1. "Difficulty" is chosen in advance, not dynamically. This is problematic as processing power increases.
2. It takes a long time for a new node to become a provably "full" node. Of course, nodes could become non-provably "full" nodes just by downloading the blockchain as is done today, and nodes could advertise themselves as partly provable, receiving a partial vote...
3. The provable blockchain could be constructed by many CPUs running in parallel... is this a problem? If so we could force serialization by salting Xi with Xi-1.
4. The system functions as a real-time interactive proof. There is no way store a historical proof that a node was full on date D. Bitcoin-style POW+blockchain does store historical proof of computation power.
Okay I got it now (I was very fatigued before prior reply and you didn't give any high level description). It is a historic proof that an entity was winning PoW blocks. So basically you have to set a threshold as to what level of cumulative PoW difficulty constitutes a vote, thus a centralized party decides what class of voter is included.
This is representative governance, because the users can't vote and those with more PoW can vote more times (creating numerous sockpuppets). Representative governance is even more prone to capture by the Logic of Collective Action.
Bitcoin has no choice but to head these sort of perilous directions because the fundamental design is flawed.
If Bitcoin's design would scale without intervention, we wouldn't be seeing the need to do this.
I don't think you get it. It is a proof that an entity is storing a UNIQUE copy of the blockchain. What little prior PoS work I've read seems to prove that storage exists, or that a file is being stored. Unfortunately they do not address the idea that you want to store a file in 2 places A & B and ensure that A & B are not in collusion and so the file is only stored one of the 2 locations.
Honestly I'm less interested in the BIP 100 voting scheme (and the usefulness of voting in general as a governance mechanism) as I am in healing the rift between miners and full nodes. Clearly no-one is going to implement the above proposal for bitcoin just so full nodes can vote on BIP 100! :-)
Justus: I could see your post just fine.