Too long? See the video explaining the stuff here Loom | Free Screen & Video Recording Software | Loom
Project statement
Through my last round iteration (implementation here) I solved web3 on-boarding and squad formation, however, retention remains elusive. People join DAOs and think their raison dâĂȘtre (sole objective) becomes the DAOâs initial collective goal. They blindly chase it, ignoring the possibility that it might not be potential best match for them or for DAOâs people and ultimately drop-off. This project aims to devise an algorithm to solve that via shuffling people (and DAOs) over different projects when they want to, conditioned on their intrinsic motivation.
This way DAOs become modular and can be âPlugged-inâ to other DAOs.
The how
The base theory will be Active Inference which has demonstrated self-assembling and autopoiesis
in natural and machine intelligence.
-
Core squad for a project is assembled via Mutual Information (MI) analysis over project objective (o) and person skillset (s).
; where H denotes Entropy -
Now, after a while, one member finds theyâre not really aligned and want a âshuffleâ, this would call Voodoo algorithm to find an alternate project by searching in direction of the memberâs gradient vector (mathematically signifying change).
-
While working on the initial objective, some members (or all) might have developed cohesion, therefore they can âlock-in squadâ and then search in direction of the collective gradient vector. This would be a means to an informed web3 pivot.
Deliverables
Develop math for How(2&3) by
- Forming Vodoo Gradient Vector (VGV).
- Defining how the vector translates along other projectâs latent space.
- Encapsulating a squad in prior mentioned two for âlock-inâ.
- Finding Collaboration-Index at every geometric translation instance and hence the pivot.
Squad
Amit Singh (metamyth)
R&D, Paperplane Technology
Researcher, Active Inference Lab
Twitter: https://twitter.com/not_amyth?t=BSpbbssQAIkrRAgg1ZTUIQ&s=09