Kill the blockchain, save your hard disk!

[wolterh]

I have come up with an idea to replace the blockchain, but I need you to find flaws in it or join efforts with me! Read on https://docs.google.com/file/d/0B-Sl27NCyKZOY1JuVlR5R3NnNTA/edit

[fusecavator]

Assuming I read that right, this is entirely dependent on the ability to set up that situation where a user is not able to edit their data, nor is a history kept. I don't see any way this could be done.

The next reasonable choice would be to have the certificates encrypted by the certificate creators
and given a copy to the wallet owners. This way they will have documents that vouch for their
wallets' states, and they will be unable to edit the certificates without corrupting them (if it is taken
for granted they will not be able to crack the encryption). A strong enough method of encryption
must be selected so this can be taken for granted. The certificates should include information on
how to contact the node which encrypted (and thus can decrypt) the information in the certificate.

Regardless of whether they can edit the certificates, nothing prevents them them from backing up and restoring them, or even just running many clients and using themselves as witnesses to make a full wallet, in which case they could spend infinitely, as no ledger is kept other than their own.

As people would be able to run multiple clients and use those as their witnesses, to validate the wallet you'd have to have a majority of the network validate each transaction, and therefore also know the size of the network. This could also cause issues if too many nodes are down, as not enough validates would be able to be done.

[wolterh]

Thank you for your comment. So I take it you state the following: a) it is not possible to make sure a user edits a certificate for his benefit; b) even while a certificate is not edited, doublespending could be achieved by using old certificates; and c) a person might run a great number of nodes in order to help themselves validate cheatful transactions.

So I attempt to refute.

a) Whether you are talking about the certified user or the certifier user, it is true that both will have rights to edit the file as per their filesystem. However, if a certified user edits his certificate, he will only be able to corrupt it, since it will be encrypted by the certifier user and hence not readable by the certified user. This is, given a strong enough method of encryption is used, and of course: that the users are not associated in any way (see c).

b) Certificates will have, encrypted within the wallet data, a wallet transaction counter. This will help nodes tell, from a number of certificates required for the transaction, if there is a certificate which is older and therefore not valid. (Every new transaction the counter is updated.) This needs not to be punished, since a node that takes a nap will perhaps miss a transaction and in consequence ignore that a certificate it gave is old.

c) It is true: two nodes in fact may correspond to the same user, or correspond to associated users trying to cheat the system. This is why the system must rely on two specific backup conditions: 1) an honest majority of users and 2) the validation of a transaction by specific nodes. While the first point may be quite obvious and understood, since every cryptocoin until now seems to rely on an honest majority, the second may not be as transparent, and here I explain it: if you own a node, you want the system to be healthy, and so you can hand pick several other nodes that must validate a wallet before the wallet can be involved in a transaction with any of your wallets. This way, even if a majority backs up a corrupt transaction, your hand-picked nodes will be able to deny the transaction. The probability that a corrupt node will know which nodes you trust and hand-picked will be extremely low in a big network.