The world is currently blended consisting of the physical realm and the digital realm. People operate sometimes in the first one, sometimes in the second one, sometimes in between.

For example, take a nuclear factory where the verification of agreements between nuclear powers also include radiation measurement techniques on a facility. The measurement should be done in a digital realm. But it will be transmitted to a controlling party in a digital realm. How will it convert from a physical metric into a digitally typed number? Someone will type in manually? A trusted device will transmit automatically? A device provided by a controlling party will transmit automatically?

This question popped up in 2012 in the paper “A New Approach to Nuclear Warhead Verification Using a Zero-Knowledge Protocol” presented by Alex Glaser, Boaz Barak, and Rob Goldston. They suggested using zero-knowledge proofs for checking the radiation level on a facility. However, without oracles the mapping of physical data into digital data it’s not reliable.

This data mapping from physical to digital (and vice versa) penetrates our everyday life everywhere. Sometimes with higher stakes (as in example described above or in case of transmitting measurements from vital health hardware), sometimes with lower stakes – state of meat on its way from the farm to the restaurant or stock prices.

If we are dreaming about having a provable and verifiable world one day – we need to have a wide variety of oracles with different levels of reliability.

Oracle as a service – centralized oracle that reports fair information because it’s paid tons of money for doing so. We assume that it cherishes its business on a long-term timeline and doesn’t have any reasonable incentive to cheat.

It mostly fetches clearly structured open digital data such as stock prices, commodity prices (e.g. gold), sports or weather data. It trusts its sources (i.e. no one checks physically the weather or sits at the football game to control if the score is fair).

Centralized oracle cannot help Princeton researchers to get measurements of the radiation level at a nuclear facility.

Oracle reports honestly because if it cheats it is slashed. One of approaches to this model is utilizing restaking in a decentralized manner (e.g. eOracle).

This model is still limited but already more flexible because there is some responsibility model imposed. I mean that slashing is a more clear incentive model than the common sense of “presumably they’d like to run their business for a while”.

While i see this model as way more powerful, flexible and accountable than the centralized vanilla oracle, I see two main constraints:

1. To be robust, it should be well decentralized. If we want to use it for a nuclear facility – we have to send quite a few controllers to the nuclear facility and ensure that all of them are independent of each other.

2. Money is not always an issue. There are ‘parties’ in the world having the money printer at their disposal. That is to say, for some use cases where the reward for cheating is high – parties will be happy to pay for it very high money price.

On a separate note, most things in the world are very cheap. Check how much the spies (who put their life or freedom at stake) are paid. It’s often several thousand dollars per operation. One more reason not to rely on economic security too seriously.

We need oracles with incentive models more robust than economic security. We need to be able to make stakes very high. For example, one should be able to stake the beloved cat, citizenship, business license, secrets (e.g. cheating on the partner), etc.

Only really high stakes will allow us to have a reliable mechanism for mapping physical data into digital and either using it in on-chain real-world applications or in generating zero-knowledge proofs.

As a thought experiment, assume we have a controller and we know their secret. The controller values his secret more than the sum of money that the nuclear facility is ready to pay him for forging measurements. Then we can send one controller and get fair results with high probability.

One can say well… this is a thought experiment: how will you know a secret or stake a cat? And they will be right. But this is the direction to think in. We currently have almost all passports NFC-chipped and stored in digital databases. Staking citizenship seems to be not too imaginary (if the issuing country for some reason agrees playing this game).

We are moving towards the digital realm more and more i.e. with every day there are more and more things to stake.

When it comes to incentive models in the real world – how people, companies, and countries ask and convince each other to do what needs to be done – there are three things that matter: money, status, and sex. They can be granted or if someone has them at their disposal – they can be removed. We ‘just’ need to map this incentive model into the onchain restaking framework.

Of course it’s not just. It’s a very sophisticated problem that I am not even sure can be really solved. But this is the main bottleneck to the wide blockchain adoption and the only one where I am not sure it’s solvable.

No real world data – no real world blockchain and zero-knowledge applications. If there is a reliable oracle mapping wide variety of real-world data into digital data in a robust way – every single blockchain application will be ready to pay for it. Because no oracle – no application. Look at how much money flows into the Chainlink pocket.

Introducing such an oracle will allow using blockchain and zero-knowledge proofs as a generalistic robust commitment mechanism where commitments will be met with high probability. The users of such mechanisms are countries, governments, and enterprises. It’s not to say that aforementioned parties always want all commitments to be met. Sometimes space to deviate is a part of the strategy and without it it is not worth chipping in. But sometimes parties are strongly interested that no one deviates. And they will be ready to generously pay for it.

The optimal game-theoretical strategy is to always deviate unless you are sure that all other parties play fair. The formula for being sure is extremely high stakes + cryptographical guarantees (commitments).

While I am not sure if we can ever have an oracle under money, status, and sex staking model, having a decentralized oracle with economic security is already a big deal compared to centralized vanilla oracles. Huge respect to people building those.

~

Have a thought you want to share – drop me an email lisaakselrod@gmail.com.