ORAO: Brick by Brick

Earlier today we released An Introduction to ORAO, aimed at people new to crypto and looking for a more general introduction to our project. To go with that we decided to create a more in depth piece that walks our readers through our platform in greater detail.

We assume at this point you know what an oracle platform is, and why oracles are neat. So let’s get straight into the meat. ORAO’s architecture can be broadly divided into four core functions: Predefined Data Protocols, our Machine Learning powered Reputation system, the Auction system, and Cross Chain Interoperability through Polkadot. Let’s dig in.

Predefined Data Protocols

Transaction size on blockchains like Ethereum come at a significant premium, so both data providers and data buyers have strong economic incentives to keep data products as slim and shallow as possible. If you ask for the price of Bitcoin as a trader, you generally want to know some fairly rich information. What’s the price of Bitcoin on Binance? And on Coinbase? Huobi? If I sell 1000 BTC in one go, how much slippage would that produce? What’s the trading volume like, hour to hour? As a trader you can get all this information quite easily through APIs from all the exchanges you care about. Try putting all this information on the blockchain while paying gas fees for every byte of data though and things get… prohibitively expensive. So when a smart contract asks oracles for price feeds, they typically get very shallow information.

And much of the time, that’s fine. Market forces are what they are, and oracles and smart contracts are designed accordingly. However, with the explosive growth of DeFi, the dearth of more complex data products is quickly becoming a limiting factor. ORAO’s solution to this is to create a library of many different Predefined Data Protocols. With ORAO the network is able to accommodate any number of variations on data protocols. If there is market demand for a specific package of data, that package can be turned into a data product.

The same design that allows for a rich variety of different financial data products also opens the door to non-financial data. Our data protocols can accommodate anything from esports tournament outcomes to seasonal precipitation levels in your area of choice, or even ads-b and other flight data picked up from the data provider’s personal location. Lack of oracle support is becoming a serious bottleneck on the growth of DeFi — and ORAO is here to fix it.

Machine Learning powered Reputation Scores

So the problem with pure reputation systems is that they can only account for what data providers have done in the past. If you tell the truth 100 times and then lie the 101st time, most reputation systems will give a big seal of approval for you to stick on your lie. Now, your lie will get exposed afterwards, of course, but this means that there is room for an attacker to play it straight for a long time, making a modest profit as a normal data provider, while at the same time looking for an opportunity to deceive a buyer in a way the seller can exploit for their own ends. If a big bet is riding on who won the latest Starcraft tournament, a betting smart contract can be fooled by fabricated data, resulting in payouts to people who should have lost, and money lost for people who should have won.

Earlier oracle networks have some protection against this in the form of a stake that data providers have to put up. Get caught lying and lose your stake. That’s certainly a disincentive, but it only means that the seller has to look for an opportunity to target a richer victim.

ORAO has developed a new approach, with the aid of machine learning. Using TensorFlow, the ORAO network maintains neural nets that calculate trust scores based not only on past accuracy but also on the data on its own merit. Incoming data products are judged based on not only what the data provider has provided in the past, but what other providers are saying as well.

These neural nets are initially trained by our developers. This allows new data protocols to be deployed quickly and immediately benefit from a robust reputation system, but after that data produced by oracles is used.

Auctions

Data is not always so easily divided into strictly good or bad. Our neural nets grade data based on intent, of course, but it also leaves room for factors like freshness. If two data providers always tell the truth but one of them consistently has two seconds older data than the first, a reputation system should account for that. At the same time we can hardly call the slow oracle here bad and be done with it.

For this reason, ORAO has an auction system, where data providers can carve out their own place in the market based on just how good their data is. Some buyers will always want the fastest, most accurate data they can get, and are willing to pay a premium for it. For others the requirements are less stringent. When you need to know who won the White Sox game yesterday, two seconds of extra ping isn’t really a factor, so you’ll usually go with whoever is supplying data more cheaply. If you’re about two drop 1000 wrapped BTC on Uniswap, well now we may want our information to be as up to date as possible, even if it costs a few extra DAI.

So ORAO sorts data providers into tiers based on their reputation scores, creating a market where buyers can choose providers according to their own priorities.

Cross-Chain Interoperability through Polkadot

Have you, uh… Have you seen transaction costs on Ethereum lately? It’s… not good.

Don’t get us wrong, it’s extremely cool that DeFi has grown to the point where the market is so hot people are willing to pay a hundred bucks to trade on dexes. But it’s not ideal.

When we started planning out ORAO, it was apparent to us that there likely will not be a one smart contract blockchain to rule them all. The future is interconnected, and the networks that make it will be those that support many chains, and interactions between chains. And Polkadot is a fantastic base layer for networks that want maximum interoperability. Thus, ORAO is built from the ground up for Polkadot. This means we will be able to support Ethereum, and EOS, and Kusama, and Tron, and any other chain where we see demand. With Polkadot we will have the best possible cross-chain interoperability, in speed, scaling and security all in one.

Chains like Ethereum will improve their scaling, of course. Layer two solutions, sharding and Ethereum 2.0 will massively increase the throughput of the Ethereum network, and other chains will follow. But with the exponential growth we have seen in crypto and especially in DeFi, it seems incredibly obvious that scaling will be an ongoing issue for many years to come. With the wealth of variety in data protocols ORAO can support, we also expect data buyers and providers to take advantage of our interoperability and utilize chains where transaction costs are lower at the moment to process more complicated data products. The future is going to be interesting indeed.

Hopefully you enjoyed this deeper dive as a companion piece to our first introduction. If you would like to learn more about ORAO, you can check out our website, dig into our whitepaper, or join our Telegram channel to speak directly to members of the team and our community.