Retrust is an R&D project into decentralised reputation. It uses an algorithm called Evidence-Based Subjective Logic (EBSL) for subjective consensus on governing value. It's sybil-resistant and oriented around a one-way flow of governing an ecosystem's norms and rewarding value creators based on a power-weighted consensus.
Token Curated Registries are an example of where Retrust can shine:
Essentially, we are using a token to achieve consensus on a subjective criteria (a "good registry"). The problem is that in order to curtail sybils from ruining the communal ecosystem value, we are limiting the ecosystem based on economics, rather than generating value we are trading value.
In Retrust, value is generated continuously according to the subjective consensus of shareholders. And in turn, the value you have contributed determines your sharepower. It is a one-way flow of governance that insures existing sharepower governs how the ecosystem grows its value.
How do we ensure the value is kept 'sane'? How do we make sure the majority shareholder doesn't just sell their stake? This will vary from ecosystem to ecosystem, and I envision a variety of mechanisms will exist on top of sharepower - ie. staking which scales power according to time locked.
The important contribution to understand is that we are inverting the mechanisms of governing. Rather than the token's value coming from the gameplay of the TCR, the TCR is governed by who created it and continues contributing value to it.
ebsl/ - evidence-based subjective logic implementations (original and my refactored version with Numpy)retrust/ - core mechanics of the protocol (interactions, quorum and reputation engines)simulation/ - code for simulating behaviour of the different enginescontracts/ - Solidity smart contractsvisualisation/ - matplotlib-based visualisation code, used within simulation/ and datasets/Install Anaconda, which handles Python version and dependencies.
# Activate the Anaconda environment
conda activate
# Run simulation
python -m simulation.mainIn chronological order:
python -m visualisation.main) - also see in docs/algovis.mp4 (Jan 2019)