Topic: Shahin Go-Round, Proof-of-Consistency (PoCo) and the RingChain.. (Read 800 times)

Thank you for spending time on this study, Magic Byte. Your kind reply provides great energy that we all need to continue such researches.


Actually, this is the most important part of the PoCo, where we force (holder)nodes to shape a Ring for syncing the pre-mined transactions during new block creation. because Ring is very sensitive due to faults and fails, this could quickly expose any possible forks during block creation. with PoCo, holdernodes could simply bypass the faulty node and continue - however they also attempt to reconnect to the faulty node. in other words, such disadvantage in Ring Network could become an advantage in PoCo consensus model.


well, this research project contains several sub-systems like flash-back-pinning (https://bitcointalksearch.org/topic/the-necessity-of-flash-back-pinning-in-structure-of-transactions-5089384) that could get used immediately in bitcoin core. but as you have mentioned above the whole bitcoin may not convert into such consensus model. I personally like bitcoin remains in PoW area forever. someday a team may come up with the idea of forking to have "Bitcoin PoCo" in the market.

I must start off by giving credit to the op for such a informative post and well published doc's on the subject.
It is becoming a rare thing to see people putting in the effort like this and credit should be dew for this.

One advantage of his type of network is that it can transfer data quickly, even with large number of devices connected because the data only flows in one direction, so there can't be any data collisions.
However, the real disadvantage is that if the main cable fails or any node is faulty, then the entire network will fail.

Self-Stabilizing Structured Ring Topology P2P Systems
https://ccl.northwestern.edu/2005/ShakerReevesP2P.pdf

I do like the concept behind this structure but I am unsure it would be suitable for bitcoin dew to the complexity of the structure as posted already by aliashraf.

I look forward to reading more of your work.

I wil credit some merit when I next have some.

MagicByt3

That's like saying an aeroplane is just a bunch of aluminium, silicon, plastic and some engines.

I know some good engineers that believe blockchain is just a linked list and Merkle tree is just a binary tree, mixed with hash algorithms. But we know well these all are about providing new Use Cases and we assess their importance by their outcomes.

a presentation file for PoCo Workshop is available now at:
http://www.mixoftix.net/knowledge_base/blockchain/shahin_go-round_v_1_2.ppt

/*
UPDATE: Note: page 15 of the ppt file talks about cluster mechanism and expresses the ability of masternodes to reject a cluster when a conflict with protocol happens. in fact, there are several steps before creating a cluster that reject such conflicts with protocol and giving this power to masternodes obviously could bring damage to the crypto. so this (the only) ability of masternodes to reject blocks will remove from the PoCo in next update unless a comment changes the situation.
*/

it contains the latest feedbacks from discussions here, in bitcointalk. and now we have master nodes in protocol, which have no super power, but some responsibilities. these slides could better show how PoCo will work so could expose any possible hidden flaws in workflows. related discussion are listed bellow:


- Is it possible to generate a consensus algorithm using machine learning?
https://bitcointalksearch.org/topic/is-it-possible-to-generate-a-consensus-algorithm-using-machine-learning-5079827

- Proof of Thought (PoT): The Holy Grail has arrived! Only Humans can mine
https://bitcointalksearch.org/topic/proof-of-thought-pot-the-holy-grail-has-arrived-only-humans-can-mine-4459113

- A Possible Solution To Re-Org Attacks That Isn't Inculudes Checkpoints
https://bitcointalksearch.org/topic/a-possible-solution-to-re-org-attacks-that-isnt-inculudes-checkpoints-5077722

- A new idea for node reward
https://bitcointalksearch.org/topic/a-new-idea-for-node-reward-5085744

- The necessity of "flash-back-pinning" in structure of transactions..
https://bitcointalksearch.org/topic/the-necessity-of-flash-back-pinning-in-structure-of-transactions-5089384

Forgive me, didn't realize I was overstepping the boundaries of the thread so egregiously. Apparently Bytecode Virtual Machine are what ETH and EOS have achieved, if I may mention their names in this sacrosanct space, so no whitepaper. Thanks though.

Complicated algorithm is bad algorithm. It is the rule of thumb in designing algorithms generally, but for public decentralized ecosystems like in cryptocurrency, it is a MUST.

Instead of artificially stacking up everything we could find and presenting a conglomeration of heterogeneous ideas, it would be more convenient and practical to stick with brilliant, simple  ideas of the past or invent a simpler and more brilliant one. My advice.

unfortunately this history is necessary, because it seems blockchainers have problem with detailed advantages / disadvantages of ring.. the conversation of every daily phone call that I receive about PoC begins with: "why you enforce nodes to work around a ring!?"


but these are work flows of the PoC - I can't apologize if they are complicated. ordinarily, talking/thinking about consensus is beautiful, not simple.. all these people here in bitcointalk still share their ideas/problems/misunderstandings about established PoW and everyday we learn something new from each other.. and however the PoC shows progress and I can't find a fatal flaw in it, but after all I personally do not hesitate in throwing a bad idea away.

Let's see  

This is it? I'm talking about your whole idea of having consensus without proof of work or stake and you are trying to convince me by asserting that ring topology has a history of its own in networking?  We are not speaking about networking and link layer alternatives, we are speaking about consensus.
And I'm the one who is inaccurate here? Really? You are referring me to wikipedia for ring topology and IBM token ring in this context?


Unfortunately I'm kinda short of time for next couple of days, I'll come to your post later but there is one thing you need to take care of:
It is not a good practice to generate even more confusions instead of addressing the problems, as your friend, I feel an obligation to keep discussing it with you but most developers/contributors just lose the stem.  

And how would i know about "Bytecode Virtual Machine/BVM" (since you're talking about technology for altcoin while this is Bitcoin Development & Technical Discussion section) ?
Why don't you share the whitepaper/technical article of this "Bytecode Virtual Machine" since just say "it's the most efficient way to process transactions" doesn't prove anything.

Frankly i don't believe it as Bitcoin transaction is very simple and use ECDSA which have fast verification speed, unless this BVM uses cryptography signature which have faster verification time than ECDSA.
Bitcoin also use merkle tree for it's block structure which is pretty fast

Forgive the noob question, but isn't Bytecode Virtual Machine the most efficient way to process transactions? This is from my CTO's mouth...

when you rush in reply, your comment becomes inaccurate - Haj_Ali_Agha [it was Persian   ]

what you call it mysterious undiscovered environment is a formal ring topology in computer networking with its own advantages/disadvantages that suits it for PoC's decentralization approach. from wikipedia: ( https://en.wikipedia.org/wiki/Ring_network )

"Rings can be unidirectional, with all traffic traveling either clockwise or anticlockwise around the ring, or bidirectional. Because a unidirectional ring topology provides only one pathway between any two nodes, unidirectional ring networks may be disrupted by the failure of a single link. A node failure or cable break might isolate every node attached to the ring. In response, some ring networks add a "counter-rotating ring" (C-Ring) to form a redundant topology: in the event of a break, data are wrapped back onto the complementary ring before reaching the end of the cable, maintaining a path to every node along the resulting C-Ring."

so we use the unidirectional mode of ring in PoC and in the case of any mismatch inside data packets (failure), the initial isomorphic shape transforms into a non-isomorphic shape, which works as a virtual C-Ring inside a real star format of internet among nodes:

"..also known as IBM token ring networks - avoid the weakness of a ring topology altogether: they actually use a star topology at the physical layer and a media access unit (MAU) to imitate a ring at the datalink layer."

and look at the advantages part of ring topology:

~ Very orderly network where every device has access to the token and the opportunity to transmit
~ Does not require a central node to manage the connectivity between the computers
~ Point to point line configuration makes it easy to identify and isolate faults.
~ Reconfiguration for line faults of bidirectional rings can be very fast

and while as a disadvantage we have : "One malfunctioning workstation can create problems for the entire network. This can be solved by using a dual ring or a switch that closes off the break." we could use this limited amount of fault tolerance as an immediate traitor detection inside the ring.


just like Ring networks that work with MAC addresses, the PoC works with "Wallet Address" of nodes inside the ring. as we wrote inside document, nodes that hold per-mined transactions (HolderNodes) begin broadcasting their (Wallet Address + IP + Listening Port + [Wallet Addresses of Miners that they hold their jobs including their hash value of their End-of-File] ) and sync them inside a storage unit sorted by Miners Wallet Address[/u], just like classic mempool (lets call it RingPool) does in PoW. therfore, each HolderNodes will understand its next HolderNode to communicate. with this simple use of ring topology in computer networking, we could build a virtual ring and nodes must follow its rules. any misbehavior of nodes will immediately detect and runs its related procedure.

based on the EOF hash value in RingPool, all HolderNodes must fill their dedicated space for the incoming pre-mined transactions that they will receive from the ring. if not, they could directly call the absent node (the non-isomorphic phase).


no, no. look.. all nodes could access the RingPool and see the situation. when finally after several round of data circulation in virtual ring, each HolderNode will generate its RNOG (the final hash root of the block), so other nodes could see any early forks of candidate blocks. if they all do not generate a same hash value, then nodes understand there is a fork situation inside the ring and need to check the submitted process of miners' Dither Hash Value to evaluate the results and understand the winner block. refer to the whitepaper, each miner entity need to submit its Dither Hash Value to the built-in pool system of PoC to get its transaction fee - this will consider as the miners vote too.

while "the longer chain is the valid one" law of PoW is banned in PoC, under heaviest attack to the network, the attackers only could postpone the process of other transactions - for a while, nothing more. legitimate pre-mined transactions will move to the next ring and this time attackers for another delay need to generate MORE extra illegitimate transactions to postpone the process again. generating illegitimate transactions need to spend Power for mining and Network fee of PoC (that wipes out in the process).. really no incentives exist here.

anyways the protocol enforces the miners to submit their Dither Hash Value, but this would be a good idea to ask miners (not all nodes) to sign their approved block hash too. miners with more submitted pre-mined transactions in the current round have more power in voting, so when 51% of votes goes for a block hash value, the winner block will identified by the protocol.


no, look, you know that one-time-passwords algorithms work both in time-based and counter-based modes. just like that, the PoW works in time-based mode, and PoC in counter-based mode, which means once you don't have enough time for block creation, this simply postpones the counter and buys another time frame for the delayed process. of cource there will be a timeout limit for miners to submit their Dither Hash Value (votes), otherwise the protocol ignores them for the round and do not renew the delay. HolderNodes act as a buffer in this situation.

but, I need to remind you that - refer to the whitepaper - in PoC we also have a reward fee for ALL nodes, if their wallet address matches to the hash value of a winner block. Winner nodes also need to submit and claim their reward too, but I think their votes should not consider in identifying the new block (still thinking about it). I only use miner's vote here to simulate the effect of classic PoW on miners incentives.


there is no reward / benefit in putting mined transactions in the chain and ban others! as you know, in PoC disk space should be wide and the protocol is allowed to put as much transaction as it needs inside the new block..

I am really going to kill the time, space and bandwidth - but save ENERGY for the planet 

It is why I'm calling it  a suicide attempt, you are sicking consensus in a mysterious undiscovered environment:

1- It is not possible to arrange nodes in this weird topology!  How would nodes join/back-off?

2- It is absolutely necessary to follow Butterin's proposal for this schema, nodes should sign the message one by another to prove consensus otherwise who decides about whether a block has enough votes?

3- For a message/block being circulating and getting signed in the ring there should be a time frame to avoid orphans getting into blockchain too. Again like what Butterin has suggested.

The point is, you just can't avoid both classical unsigned, asynchronous blockchain and  obsolete centralized encrypted messaging system.

To put it more simple:
Either there is an incentive for nodes to generate a block or there is not, without incentive game is over and anybody would be able to manage for a double spend attack
and with incentive there is a competition
and with competition people go selfish and try to put their own minted blocks in the chain
So there should be a protocol to regulate this race
and this protocol is based on a messaging system
and this messaging system is either asynchronous-unencrypted (both) or synchronous-encrypted(both)
there is no third chair to sit on!
The unencrypted version is what we know as blockchain technology
and the encrypted version is an obsolete idea left behind in 1980s and now as a joke is re-proposed by an ex-teenager who is just bored and wants to make himself great again!

To make it crystal clear:
Suppose I'm running a node, I'm loyal but I need money too and every single time that I receive a block from my neighbor(s)  I pick rich transactions and add them to my block and relay it to other peer(s) who are as greedy as I'm if not even more  

the built-in pool system in PoC need temporary space for accounting purposes - this has nothing to do with attacking/defending anything. I have described above the equilibrium policy of PoC. the miners and nodes with higher amount of transfer fee / reward fee have the most power in voting - which carries the power cost of mining and its related marketing cost into the equation.


when you do not need marketing knowledge and marketing process for your job, this means you have someone else in a higher level is doing it (in other words, this means you are a slave or an employee in a centralized system. Pools in PoW, and PoW in huge processing power providers - in the case of bitcoin, ASIC provider companies). with marketable mining, you could have your own job in a decentralized market.  


now I could see you didn't read that paper about The Chinese Generals Problem. pay attention, the isomorphic ring of generals is 100% vulnerable to traitor nodes, which means one traitor node is enough to break the consensus. I use this property in DETECTING traitor nodes, then immediately drop them out of isomorphic ring (replace the 1st ring with a non-isomorphic version) based on the policies of the PoC protocol. after block creation, the block confirmation process begins by voting that we need to see how percent of traitor nodes may defeat it. ** primary calculations show the fault tolerant is still 51%.

and this has nothing to do with AGP and that 99% protection in ETH.. where did you get it from?! and blocks are not reversible means there is no *the longest chain is the valid one* rule in PoC - so reaching consensus will be a time consuming process. I am rewriting this, God. the worst thing that ever could happen here is transaction censorship that an end user could resend it to another miner.


P.S.

sometimes I really surprise how we accept weaknesses of PoW as under control equilibrium, but others weaknesses as out of control sinking ship   come one, there is no 100% guaranty in any of these systems at all and PoC has its own characteristics in this area.

1- As of power consumption decrease: No such thing at all in your proposal, and it is not even possible to replace energy by space or bandwidth. Think about it! The adversary could equip himself with enough amount of such resources with neglectable cost. Security in decentralized ecosystem is about the equilibrium state of network in which attack costs considerably are higher than incentives/potential benefits for the attacker.

2- I maintain that attaching work by miners to transactions is bad Idea.


4- don't think like a PoW believer  

in PoC miner's Marketing Knowledge is inevitable for receiving job from end user. this opens new horizons of social marketing and face-to-face marketing among end users and miners. in fact, instead of counting on processing power of a miner in PoW, I try to count on the legitimacy of a miner in real world. e.g. a charity organization may host a PoC mining facility and end users that know them very good may give their raw transactions to them. in PoC miners act locally and nodes act globally.