-
Notifications
You must be signed in to change notification settings - Fork 58
Why do we need a trust machine
Under Construction.
Reki
You develop some services/systems/applications and always need they work "correctly".
i.e. Output the correct results.
If your application require some input data which are not administered by you...
Why do you trust these (external) data ?
Sometimes, since the original data owners are "Trusted by you", or the wrong data is not a matter.
It was a (Human) collaboration.
If your application GET some data from unfamiliar API platforms, and the wrong data will cause terrible (balabala).
Maybe you must use it, or someone monopolizes it.
How to Trust ? How to collaborate with other systems ?
An easy way:
- Annex it. => Build one platform. => Centralize all data.
But, data are valuable/expensive, it is hard/impossible to be centralized...
- Attach a signature. (non-repudiation)
- Sign an agreement. (buy the data, and deal wrong data will be punished)
- Sync. and periodically audit the data. (more cost)
The popular way:
- Let nonprofit 3rd party to help everyone. => Centralize all.
You (will) trust a data (content), since someone is responsible for it.
What is the Cost ?
Punishment, Audit, Time, Human work...(balabala).
The 3rd party will/can create the policy, require fee, increase the latency, (balabala).
What is the State ?
e.g. The Data ownership.
I own a data. => It is a single state.
The data was transferred from anyone to anyone. => A series of state.
How to get the newest state ? i.e. Who is the data current owner ?
Furthermore, How to get the newest states of all data ?
We need a way.
It can reduse the cost.
It can help the application GET the correct/newest data state easily, and keep working correctly.
We need a protocol.
It can protect the rights, interests and security of data owner.
We need a architecture.
It provides scalability and does not affect existed everything. (data-driven collaboration)
Blockchain can be a solution.
It is a trust meta-layer which is under various data layers.