Protos

Why do institutions form?

• Cultural and social norms as a kind of institution
• Social cogs and wheels can produce persistent social structures and social forces which in turn may evolve institution like objects
• Agents can learn to coordinate and work with each other as in reinforcement learning
• Conscious deliberation about agents may give rise to birth of institutions

Why do institutions lead to inequality?

• ???

At what point do we call it an "institution?"

• Put another way (Milton): What criteria do we use to determine that a collection of entities is a group that has a collective purpose and indulges in actions that are deliberately intentional and focused on addressing the group's overarching goals?

  • Surely, these criteria have to be context-specific. Are there "universal" principles for intuiting/deducing/inferring what these criteria are (or ought to be) in a particular context?

• The stages of institution development:

  1. (Case A) Sam, Joe, and Bob are living near each other and just happen to interact with each other a lot.
  2. (Case A½ - Rajesh) Sam, Joe, and Bob learn to help each other because they discover it is optimal in the long run. (Reinforcement learning; game theory: cooperation games)
  3. (Case B) Sam, Joe, and Bob say to each other, "hey guys, we've known each other for a while, and built trust, and it's time to take more mutual advantage of this relationship. Let's agree to pool resources and allow each other to draw on the pool, etc."
  4. (Case C - Rajesh) Sam/Joe/Bob actively bargain with each other to agree on what is the optimal resource allocation agreement for their group. This is also possible in cooperative game theory. The steps of the bargaining is very algorithmic and the computational complexity of achieving fair bargains is reduced when trust is involved. (Think regular USDollars vs Bitcoins).

  Stephen's (possibly naive) claim: no amount of Case A happening just "becomes" Case B until the agents make a deliberate decision to go to Case B. Corollary: if true, the simulation must explicitly model groups of agents that have made said choice.