On 13th July, the Founder of IOHK, Charles Hoskinson spoke about the scalability of Cardano during the Symphony of Blockchains presentation. The London meetup focused on the importance of design and had the designers of IOHK and the founder of Kuva as its keynote speakers.
After the presentation, Charles was questioned on the scalability of Cardano to which he replied saying that there has been “tremendous progress” with Cardano’s scalability.
Hoskinson said that the research project for Oroborous, the new proof-of-stake algorithm which according to him is ‘the heart of Cardano’ will assume the scalability of Cardano. He said that the research program for Oroborus started in mid-2015 with a paper called G-Cal 15. The main aim was to create a foundation which would talk meaningfully about blockchain.
He said that they first looked into what a secure ledger was in the most foundational sense and then studied proof-of-work to understand its security characteristic. They also researched and understood if alternative consensus could achieve the desired level of security.
“So, the first question was proof-of-stake. For example, can it achieve the same level of security as proof-of-work? That’s a little bit of controversial topic as some people acknowledge. So wrote a paper in 2016 called Oroborus and it wasn’t particularly a practical paper but what the paper did is, the paper proved that you could achieve under certain unrealistic assumptions, the same level of security as proof-of-work with proof-of-stake.”
He further said that the Oroborus Genesis was a follow up for the Oroborus paper where the synchronous has been moved to semi-synchronous and static corruption to adaptive corruption in order to go from checkpoint to bootstrapping on the Genesis block.
They have also been working on resolving all the practicality problems. It is, at present, equivalent to the security of Bitcoin but wouldn’t work on real-life cryptocurrencies.
Following this, the next step after practicality is scalability which is sharding the protocol. This would allow it to run in parallel and is more about accepting the right trade-off. Charles said that there’s a lot of ideas on how to go about doing this.
He further adds:
“So what happens when you go from a single threat to a parallel in these systems is, first you lose a little bit of Byzantine resistance, usually go from 50% to about 33% to a quarter. That’s the first trade off that you accept, the other thing is, if everybody follows, you have great performance but if people start not following the protocol the usual performance degrades pretty rapidly with all of these types of protocol”
If users do not follow the protocol, then the solution would be to detune the protocol. At optimal configuration, the transactions can go from 200-300 to tens and thousands.
The next step after this would be to decide if you want to everything on-chain or off-chain. If it’s going it be on side-chain, then there would be much more performance but there is a trade-off, trade-off in terms of who’s in control, availability and privacy. For this, there’s a new version of Oroborous called Oroborus Hydra because hydras have many heads and similarly can maintain blockchains and many shards concurrently.
He says that they believe that this would give them a reasonably good performance which would be enough to maintain a production system at a scale of millions of users. Later on, they could add more to it, like overlay protocols and then shift various performance out of the chain like using multi-party computation or payment channels in order to build a good ecosystem.