<<  >> (p.32)
    Author Topic: DNotes 2.0 - Staking, CRISP Interest, DNotes Pay  (Read 148894 times)
    This is a self-moderated topic. If you do not want to be moderated by the person who started this topic, create a new topic. (3 posts by 1+ user deleted.)
    TimMarsh
    Full Member
    ***
    Offline Offline

    Activity: 187
    Merit: 100

    Professional cryptocurrency writer incl DNotes.


    View Profile WWW
    July 04, 2017, 08:56:33 AM
     #621

    Proof of Stake
    I've been looking at, and thinking about, proof of stake lately, how it works, pros and cons. There is obviously a big advantage in not needing so much power per block. And I like how a new block is more likely to be created by a small node, than a massive mining company that might approach 51% of the network's processing capabilities.

    But I'm concerned about the possibility of running millions of tiny nodes to improve the possibility of gaining a double spend opportunity. Finding a balance between the big fish getting most of the chances, and it being very cheap to establish a stake in a node that has a chance at block creation seems to be the goal of most solutions.

    And so I've found myself heading down the path of maximising the benefits of randomisation. The resulting model would have a variety of random-selection algorithms. Some weighted towards a capped aggregate coin age, some weighted towards a capped total coin value, some weighted to other node characteristics. Let's imagine that there are 99 different randomising algorithms called r01 to r99. I'm going to use the term Appointed Set, to mean a set of 99 randomly selected nodes where each node was selected using its own algorithm from the collection of r01–r99.

    I imagine minting a legitimate block could then follow this process:

    1) 50 different Appointed Sets have been selected during the minting of the block which was made three blocks earlier.
    2) Then each node from the first Appointed Set selects 50 new nodes with each node based on its own r## formula incremented by 1 for each selection. eg: r23 uses algorithm r24 to select the first node in its collection of 50. Then it uses r25 for the following selection. And it continues until it has 50 random nodes selected using r24–r73 biases.
    3) All 99 nodes of the Appointed Set share their selected nodes using peer to peer. These are then arranged by r## values into the collection of 50 Appointed Sets.
    4) Then each node from that same Appointed Set mint all pending transactions and the newly shared Appointed Set Collection into their new block and compare their hashes.
    5a) If all 99 hashes from the 99 nodes in the first Appointed Set match, then the minted block is published to the blockchain.
    OR 5b) A side chain is published to the block chain telling the second Appointed Set to begin at step 2. This side chain also contains the hashes generated by the first Appointed Set so that they can be inspected to see which node(s) was corrupted.

    I'm not sure why I thought embedding the Appointed Sets three levels deep in the blockchain was valuable. It does warn potential hackers of the nodes that will be trusted, and make them targets for hacking. So maybe it would be best that Appointed Sets are generated only as 'transaction style records' for including in the upcoming block. The number of nodes in an Appointed Set are arbitrary, as are the number of Appointed Sets required in a collection. These numbers could be automatically be determined as a capped ratio to network size.

    The main thrust of this suggestion is to use randomisation in the consensus process to disrupt all possibility of 51% attacks being successful. Once this block, approved by the 99 nodes of the Appointed Set, has been published, it would still be subject to the already accepted 51% approval process currently in place to consolidate the block and move on to the next one.

    It is quite possible that my suggestion has done more to illustrate the limits of my understanding, than it has done to improve the state of POS technology. If this is the case, I look forward to comments that point out what I have not understood.



    Cryptocurrencies will level the playing field. I'm paid to write, but not paid to promote.
Page 31
Viewing Page: 32