> <@ruben_kelevra:matrix.org> Well, currently the blocks which get received need to match the checksums which is saved in the CID (and the metadata below the CID).
> 
> Accepting encrypted content would mean you would accept a huge amount of chucks before you can check the integrity and you cannot determine which chunk is faulty.
> 
> This would open the door for DoS attacks.
> 
> Also matching encrypted content to not encrypted content with the same CID makes encryption useless. Everyone who got the unencrypted content can determine which content you have saved encrypted on your node.

ok, imagine if Alice want some large file, which stored by Bob, Clara and Diana, but all of they are want to keep original content confidential. In case, when all communicate p2p, using e2e encryption, possible two varians - common secret, which used by all participants and shared previously, or when each pair of users use different secret, generated by for example using ECDH. In second case, each from Bob, Clara and Diana, should to send to Alice blocks, encrypted using different secrets, so if each block will be marked by hash from open data, Alice can identify every block, for example for drop duplicates without spending resources on decryption, or Bob, Clara and Diana can make decisions who which block must send to Alice, using common information calculated from open data, without necessarily of share that content. Yes, I'm understand, that this can sounds a bit strange, but I'll explain that in my proposition for RFP0