Meet RepuCoin: The Best Pupils Develop Their Academic Reputation Over Time
Oh, I love being in research. One minute you are scratching your head about a solution to a problem, and then you read a paper, and it moves your thoughts on another step in your journey in the discovery of new things.
Every teacher will know that a good and trusted pupil will develop their academic reputation over time. A student who works hard and always produces correct answers will be trusted in the future to do the same. A pupil who has a poor attendance record and who produces poor outputs, will often not be as trusted to perform a given task. A weak pupil who produces amazing work will often come as a surprise. We do the same with academic research, and where researchers publish papers over time, and these build up their esteem within their field.
And so on the Internet, reputation increasingly means something. A person who just appears on Twitter with 10 followers is likely to be much less trusted than someone who has 10K followers, and who tweets each day. If a person also has a blog, and a YouTube channel, and has published in reputable publications, each of which has some form of continuous output, their identity will often be highly trustworthy, and their opinions are more likely to be taken seriously than someone who has just set up an account.
And so a great deal of research time is going into finding ways to get rid of proof-of-work methods in cryptocurrencies. For this, we must build a consensus on the current state of our transactions on a network. In order to avoid an attack, we make sure it will just be too costly to add invalid transactions. One of the most successful ways to overcome this is to provide a proof-of-work in order to prove that a node can perform the correct answer and can be trusted. This is the core of Bitcoin, and where nodes — the miners — must race to produce the required hash for a block.
An attack on the network is then where an adversary takes over at least 50% of the network’s computing power and can force the network to agree to incorrect transactions. Recent attacks on cryptocurrencies have shown that this is possible, especially where an adversary invests in large-scale GPU clusters to ‘take-over’ the infrastructure for a short time. An adversary who has more than 50% control of the network voting for a consensus will then be able to sway the other nodes that their viewpoint of the consensus is correct.
But a new paper from Yu et al addresses the take-over problem, and limits the voting power of nodes within the infrastructure over time [here]:
Within RepuCoin, each miner has a reputation on the network. This is based on its correctness of its previous work and its interactions in creating the blocks. For this, we have consensus groups which exist within a group of miners (X). These are selected based on their reputation scores, and are responsible for running the consensus protocol. The voting on a consensus is then created using a weighting of the reputation score of the miners. A quorum is then based on having enough votes, and also the cumulative reputation of the consensus group.
Time-based reputation is often key to overcoming short-time attacks. Thus the reputation scores vary across the miners based on their time within the network:
Thus we must have a social community and where voting must take place on key decisions. This is the case for most of the permission-less blockchain methods (those which are public), and where a community must vote on key changes to the operation of the infrastructure. No one entity can then take over the control of the operation of the infrastructure, as they would have had to existed over a given time period, and also produced good work. This will become expensive for the adversary, as they must prove computing power into the infrastructure, and also produce good results.
We are prototyping a new Internet, and one which is secure, trustworthy and resilient. We are not quite there yet, but it is evolving. We thus have a 1980s viewpoint of our digital work — centralised, insecure and untrustworthy- or our new distributed, security and resilient world.