Decentralization and Trust
OneCommons takes a different approach to decentralization – at a technical level, we enable the platform to be decentralized not through a protocol but rather from the combination of:
open source code + reproducible deployment + shared persistent state
Note that the “shared persistent state” (the DataCommons) is powered by a block chain, so in a sense we are expanding the reach of its decentralization properties (you could think of it as an extremely elaborate off-chain transaction).
Our current approach places trust on the cloud providers that control the platform’s servers but one can imagine a future where the browser verifies the provenance of the code of the website it visits using hardware-based remote attestation – reducing trust to the hardware (which is practically unavoidable in any case).
But more generally our approach treats trust and decentralization as multi-faceted and subject to inescapable trade-offs, as opposed to relying solely on protocols, which is limited and dependent on a critical path of inevitably buggy software.
This pragmatic approach is illustrated above, where our current approach is still superior to placing complete trust in whichever entity happens to be running the website you are visiting.
More specifically, this hybrid approach is motivated from:
- Being realistic about the tremendous efficiencies of centralized cloud computing.
- Being realistic about the burden of re-implementing applications for an entirely new architecture.
- Recognizing the clear advantages of familiar software solutions that are managed by someone else.
- Observing that the governance model is as important as decentralization - even if a protocol is decentralized, centralized control can be exerted at many levels, such as control over critical software, protocol evolution and interoperability standards, and related institutions (e.g. marketplaces).