<<  >> (p.3)
    Author Topic: Theoretical cryptocurrency that globally verifies so fast it can stream audio  (Read 3963 times)
    BenRayfield (OP)
    Sr. Member
    ****
    Offline

    Activity: 316
    Merit: 250



    View Profile
    January 26, 2012, 06:02:01 AM
    Last edit: January 26, 2012, 06:32:20 AM by BenRayfield
     #1

    ...compared to Bitcoin which runs on 10 minute cycles with faster estimates.

    Do transactions the same as in Bitcoin, except no blocks or scripts or other complexities.

    Run the whole network like a http://en.wikipedia.org/wiki/Distributed_hash_table

    Part of it would be similar to this, except no node owns a space of the keys. The many dimensional space would just be to order the transactions so they could be organized as leafs in a binary tree in realtime.
    Quote
    Keyspace partitioning
    Most DHTs use some variant of consistent hashing to map keys to nodes. This technique employs a function δ(k1,k2) that defines an abstract notion of the distance between the keys k1 and k2, which is unrelated to geographical distance or network latency. Each node is assigned a single key called its identifier (ID). A node with ID ix owns all the keys km for which ix is the closest ID, measured according to δ(km,ix).

    There would be no root of the merkle trees. It would be a large number of merkle forests that converge toward trees but never get there because its streaming so fast and constantly AVL-rotating the binary merkle trees to limit their depth and balance them.

    http://en.wikipedia.org/wiki/AVL_tree

    Instead of paying 50 (or 25 or 12.5...) bitcoins, plus the transaction fees, to the miner of each block, pay some fraction of the remaining transaction fee of all child transactions to the miner of each AVL branch. Each transaction would be paid to log_base_2(total number of transactions ever in the network)*rotationMultiplier*averageNumberOfTreeTypes addresses, where rotationMultiplier allows for AVL-rotations and paying miners whose block stays around for some time but is obsoleted by an AVL-rotation, and averageNumberOfTreeTypes distributes the transaction fee between the different trees its a member of, 1 tree for each dimension, and nobody knows exactly how many trees there are for security reasons.

    Some mined branches will pay off and most won't, because the decentralized AVL tree is constantly rotating to limit its depth and balance itself.

    To cause a small but big enough fraction of the transaction fees of AVL branches to be paid to the miners of those branches, each AVL branch is also a transaction and is submitted to the network the same way.

    When you first think about it, that sounds like it would cause an infinite loop of more and more transactions being created, but the loop is closed by the miners of AVL branches being motivated to mine on the things they're branching together and the address they send it to simultaneously, since that way less work is done for more mining. It can be done in many combinations and whatever turns out to be most consistent will be the most efficient way of mining.

    Since its a binary merkle tree, the number of internal nodes is always 1 less than the number of leafs, so it is practically small.

    No computer needs to store or use more than a small fraction of the transactions. All objects are transactions. Transactions and blocks and AVL binary merkle tree nodes are all the same thing.

    The major advantage of this design is transactions are searchable by all their addresses therefore its trivial to immediately check for double spending by searching the decentralized binary tree for the relevant addresses and/or previous transactions.

    Like in a distributed hashtable, transactions, as the values of the keys, would be stored based on many dimensions. Those dimensions could be substitution ciphers or other kinds of secure-hashing of the existing hashcodes and addresses of the transaction. This exponentially increases the security because no computer knows all the ways any specific transaction is stored, in many different binary trees keyed by a different substitution cipher or secure-hash of the same data. Therefore a small number of random searches of the network for transactions, across different binary trees, should reveal deception quickly. Every binary tree should have the same data. If it doesn't, the incompatible branches of those binary trees are dropped from the local computer and only their branches are used, to be AVL-rotated and branched higher for more mining rewards.

    The network organizes itself. Every AVL branch knows which range of binary numbers it supposedly contains, so you know to traverse it or not, and trust it based on a few random searches into it for things you know (1) are, and (2) are not there.

    Bittorrent can stream large data to many addresses at once. New cryptocurrency could benefit from taking some design elements from Bitcoin and Bittorrent.

Page 2
Viewing Page: 3