Quote from: freezal on February 03, 2017, 06:47:37 PM

Quote from: SatoNatomato on February 03, 2017, 05:33:24 PM

Quote from: freezal on February 03, 2017, 04:53:44 PM

Hi tonych,

Shower thought (almost certainly a stupid one as always happens when you don't exert deep thinking, as I haven't finished reading the white paper, and have no time right now to research feasibility):

Would it be possible to optionally (because it changes somewhat the privacy/anonymity model) save assets (including blackbytes) in the public DAG as a compressed and encrypted payload (not much differently of what is saved locally today) paying due commissions?
Maybe we could now use a different seed for each asset to generate private keys for them and some hash of these to encrypt assets payload after each transaction.
Wallets would scan the DAG trying to decrypt asset payloads to get balance and history (some optimizations may apply).
Somehow old saved payloads in the DAG could be pruned as assets carry their own history.

Pro:
- Massive improvement in usability, no need to back-up local assets after each transaction (simplicity is beauty).
Cons:
- Privacy model changes (thus to use it optionally) as now assets history only remains private as long as no one discover a way to decrypt the payload which is now publicly available.
- Increased storage requisites for the DAG (but it'd be optional and has costs)

There is nothing stopping you from tar/zipping up, encrypting and storing your own blackbytes as data anywhere public, even on Byteball. But thats just reckless in my opinion, aside from increased costs.

There is really no benefit to storing the whole thing in Byteball, if you actually are so certain of your encryption and passwords not getting compromised store it elsewhere were costs are lower. Compared to say storing the data in ipfs and instead paying lower fee by storing the hash to the ipfs data in Byteball. Well actually there is something to explore here.

Like this, tar compress the blackbytes data, encrypt with AES passphrase, the sha256sum of the blackbytes.data.tar.bz2.enc is published to ipfs (you are seeding it now). The sha256sum+salt of your passhprase (or just bcrypt or whatever that thing is called) is published as ipns name pointing to the sha256sum of encrypted data. Now you can look up your blackbytes encrypted data if you know the passphrase and lookup in ipns what the hash to the encrypted data is. This ipns name is what you store in Byteball if you really want.

IPNS, because if you want to update your data, when sending/receiving blackbytes, you publish that, and repoint ipns to the latest blackbytes-encrypted data which you now also seed.

In this way you offload everything to ipfs/ipns and really you dont need to keep any new hashes in Byteball. If you want to help seed/share others encrypted data, and others to help you, well, thats easy if you want to just publish hashes and begin sharing things you dont know what it is, and if you want economical share only other people who also share/constribute, thats more tricky.

I wouldnt do this though.

Thank you for your reply.
I think I could understand your suggestions and they are interesting and have their own merits, but the point of my post was to have all this integrated into Byteball.
The goal was to make possible and automated that a mechanism like the one tonych offered today to recover all bytes balances and transactions of any wallet by just typing the seed would also work for assets (blackbytes included).
The idea was to use a generated random seed (not a human password or passphrase) not only to generate keys for assets but, via extra hashes of those keys, to generate a random encryption key for each transaction in order to break traceability of saved payloads.

Ah I see. That makes sense.

Quote from: drays on February 03, 2017, 08:33:05 PM

Quote from: tonych on February 03, 2017, 06:19:50 PM

The keys are smaller in size than the original payloads but still have to be stored privately and backed up, so the backup problem is not solved.

The best solution of the backup problem is imho multisig.

Sorry if this was addressed previously, but if the backup is a problem (and I believe it is actually), why won't the wallet create a backup automatically after each transaction, and save it to separate designated place (on the same device as a first step). What I am missing here?

There was an idea by someone to just zip up the data folder, or even just the parts containing blackbytes, and use that as backup.

Maybe that is low-priority feature because multi-device/multi-sig is easier way to backup?