<aside> đź”— See https://github.com/filecoin-project/FIPs/discussions/562
</aside>
A large majority of users onboard data onto the Filecoin network via an aggregator. Today the work done by aggregators is unverifiable and unprovable. The user relies on the aggregator to perform the work correctly and at the same time, it is impossible to prove to a third party that a given piece of data was included in a deal.
The proposal is split into three parts each of them addressing a part of the requirements.
Today a sector carries a set of deals, with information about where which deal is situated stored out-of-band within Filecoin’s consensus layer. The Data Segment Format aims to allow for this data to be stored in-band, within the deal or sector data itself. The mechanism is independent of whether, in the future, Filecoin enables fully off-chain deals, where the deal maker and storage provider come to an agreement without involving Filecoin’s consensus, or the on-chain deal-making continues; it can be applied either on the sector or a deal level.
The primary goal is to inform the Storage Provider what data is stored where within the deal or sector in a verifiable way, without having to utilize the limited chain space.
A data segment is a sequence of data which is guaranteed the following properties: it is possible to verify its inclusion within the container and it can be trivially found in the container by the Storage Provider to facilitate data retrieval.
It is important to highlight that on-chain deals, as realised today, fulfil the definition of a data segment but use the chain to facilitate the execution of the second property.