Thanks a lot for the reply, let me think it over and come up with a proper response. You and Conrad may be right about the proposal being irrelevant to DDoS, so I'll try to come up with the more solid argument supporting my point of view, otherwise, I will refute the proposal myself). At the moment I feel like I need to specify game theoretical model and consider all costs and incentives in more details. Sincerely yours, Ilya Eriklintsev.

On 22 June 2017 at 16:47, John Hardy <john@seebitcoin.com> wrote:

Hi Ilya,


This proposal wouldn't work because bad actors can perform PoW just as cheaply as any other participant.


The transaction fee already acts as a mechanism to prevent spam. It is not a problem to have a lot of low value transactions in the mempool as thresholds can easily be set for them to be disregarded/expire - a 300MB  maxmempool size by default eliminates any real 'DDOS' risk. Spam only really becomes an issue when it enters the blockchain. If a spammer is willing to pay the tx fee to spam, they'd be willing to pay the PoW.


The only actors who can spam the blockchain at zero cost are the miners themselves (beyond the opportunity cost of including genuine fee paying transactions). They can even do it without their transactions ever hitting the mempool or including a fee, though this behaviour would be easy to spot.


If miners are colluding to spam the mempool or blocks in order to increase the average transaction cost overall there is little that can be done as the network relies on 51% of hashpower being honest. A miner creating spam transactions that enter the mempool has the risk that another miner would include it in a block and they would incur an economic cost if this happened.


I had an idea for a dynamic blocksize that required miners to pay a percentage of the transaction fees to the next mined block. Here is a link: https://lists.linuxfoundation.org/pipermail/bitcoin-discuss/2017-January/000123.html


If it was established that miners spamming blocks with transactions was an issue, this could be used as a disincentive as it means the cost for doing so becomes non-zero.


Regards,

John Hardy


From: bitcoin-dev-bounces@lists.linuxfoundation.org <bitcoin-dev-bounces@lists.linuxfoundation.org> on behalf of Ilya Eriklintsev via bitcoin-dev <bitcoin-dev@lists.linuxfoundation.org>
Sent: Wednesday, June 21, 2017 9:55 AM
To: bitcoin-dev@lists.linuxfoundation.org
Subject: [bitcoin-dev] BIP Idea : DDoS resistance via decentrilized proof-of-work
 
Hello everyone,

recently I have got an idea that in my opinion will improve bitcoin network, making it better store-of-value for growing cyberspace and cryptoeconomy. Sorry for longread below and thank you for your time.

Decentralized proof-of-work and DDoS resistance for Bitcoin

Abstract

By introducing some new block validation rules it is possible to make Bitcoin network more secure, decentralized and DDoS resistant. The idea is to modify simple proof-of-work puzzle in such a way that user transactions could be hardened with the same proof-of-work algorithm thus incentivising all the miners to include that particular transaction. Such mechanism will effectively give a handicap to every miner who includes "mined" transaction into next block, increasing probability of him getting block reward.

Problems and motivation

This document will address the issue of a continuous DDoS attack targeting the Bitcoin network, e.g. full nodes mempools constantly being overflowed with transactions carrying small value reduce system primary ability to transfer value (and hence making it perfect store of value). Valid transactions are cheap to create (in the sense of computational effort required) and no adequate mechanism exist to make transaction total value increase probably of its confirmation by the network.

Currently, miners decide which transactions to include in blocks because it's them who are securing Bitcoin network providing proof-of-work certificates stored inside every block header. Miners have to store the whole blockchain at all times, so one of the costs is storage which grows linearly with the transaction size (blockchain size as well). Another cost is network bandwidth which depends directly on the size of data to be communicated over.

The only incentive a person who wants to transfer his bitcoins is allowed to use is setting of transaction fee, that is going directly to the miner. This solution probably was intended to utilize free market (as implied by Satoshi introducing sequence numbers) to determine appropriate fees, but that is obviously not the case, in the current bitcoin network operating in full block capacity mode. This fee market deviates significantly from a free market premise (also attempts being made to make it closer, e.g. in BIP125 where Replace-By-Fee signaling is supposed to help in replacing "stuck" transactions with noncompetitive fee).

Currently, bitcoin network is susceptible to the DDoS attack of a kind. Adversary creating and translating into the network a lot of transactions carrying small value (e.g. only miners fee), will be able to impair the ability to transfer value for everyone in the world, should he has enough money to pay the fees. Miners would continue to work providing security for the network and new blocks will consist of transaction transferring negligible value. It's a major drawback because the cost of such attack doesn't grow asymmetrically with the cost of BTC asset.

Proposed solution

So how do we incentivize all miners to include our transaction carrying a lot of value in the next block? The only thing a miner supposed to do to get a reward is to produce Hashcash proof-of-work, thus providing cryptographic security guarantees for the whole Bitcoin blockchain. What if including our transaction in a block would reduce effort requirements for the miner produce valid block?

We could do so by extending the concept of proof-of-work, in such a way that we do not sacrifice security at all. Here are both descriptions proof-of-work as-is and to-be:

Standart proof-of-work: hash(previous block hash + current block target + current block metadata + current block transactions) < target

Decentralized proof-of-work: hash(previous block hash + current block target + current block metadata + current block transactions) - sum( FFFF - hash( previous block hash + raw_tx ) ) < target

It is possible to imagine completely mining agnostic proof-of-work, for example, the following PoW would do:

Distributed (mining-agnostic) proof-of-work: sum( FFFF - hash( previous block hash + current block target + current block metadata + signed_tx ) ) < target

Described protocol change could be implemented as user activated soft-fork (described in BIP148), introducing new blocks with the modified proof-of-work concept.

Economic reasoning


An adversary whose goal is to prevent the network from transferring value will have to compete with good users hash rate using same equipment good miners will use. And it's far more complicated than competing with others using the money to pay transaction fees.

In order to investigate probable consequences of protocol upgrade and stability of implied economical equilibrium, we need an adequate game theoretical model. Such model and numerical simulation results should be obtained and studied before any protocol change could be considered by the community.

To me it seems like a win-win solution for everyone owning BTC:

Miners benefit: as the result DDoS attack will be stopped, Bitcoin becomes perfect store-of-value finally. Political decentralization of hash rate will be incentivized as mining equipment becomes relevant to every user.
Users benefit: miners will have direct incentives to include transactions deemed important by their sender and supported by some amount of proof-of-work.

Sincerely yours, Ilya Eriklintsev.