Quote from: inBitweTrust on February 20, 2015, 02:37:37 AM

Quote from: onemorebtc on February 20, 2015, 02:23:57 AM

that still doesnt answer my question.
may i ask why it is not enough to secure the blockchain that way?

i assume that all miners already have full blockchains and we are only talking about new nodes (old nodes already have/had the full copy and therefor a validated utxo)

new nodes would still connect to running nodes - they just wont deliever all old transactions. blockheight, headers and pow is still all the same.

i fail to see a way how a new node could be tricked or has lower security if all old transactions are obtained in a different (even centralized) way

Downloading the whole blockchain from a single torrent or source is fine if it is verified with other unpruned full nodes .
What is mainly being discussed here is either:

1) Many pruned full nodes of ~1GB in size boostrapping and creating a verified UTXO from a combination of archival full nodes which are unpruned (how many of these should be sufficient is the question)

2) Many pruned full nodes of ~1GB in size + another 1-5% of a redundantly sharded blockchain using DHT on each node verifying each other
or
3) Combination of the above

0.11 will likely include the feature to create full pruned nodes and is expected to come out in approx. 6 months so you will now see 3 types of clients in the future-

SPV clients
Full pruned nodes with verified UTXO that are around 1GB in size
Full nodes with the whole unpruned blockchain

Thus the conversation will shift from increasing the current amount of full nodes from 6400 back to 10k , to focusing on this and talking about how many Full pruned nodes we want out in the ecosystem as well to secure the network.

I would like to hear what degree of security and decentralization the people against this hardfork would like so we could both work towards a common objective.

i do understand that we are talking about putting transactions (not blocks) on dht.

it seems archival nodes are the same as a raw centralized download (at least comparable)?
thats why i asked if it is still verifiable and trustless if transactions are downloaded from a single location (just imagine the extreme: only one archive node)

i still dont see why this lowers security as you claimed (you said this would raise node count - i think this too, and this would actually increase security a little). it adds complexity but there is always a tradeoff to make.