Agent swarm governance #31
Closed
badibam
started this conversation in
Ideas & Thoughts
Replies: 2 comments
-
This is a very logical idea and sounds fairly practical to implement. |
Beta Was this translation helpful? Give feedback.
0 replies
-
This qualifies as changing the architecture and scope of the project. You're welcome to try this elsewhere, but it does not abide by ACE framework or HAAS |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Back when I studied social psychology, I came across some information that I found interesting in terms of how groups are governed.
In this post I'm going to speak from memory and in a very global way.
Studies have shown that in a group :
1- Pyramidal, top-down modes of governance are the most effective for executing decisions.
2 - Bottom-up modes of governance are more effective for decision-making. The idea being that people are better able to improve their own work, than their superiors who are not directly connected to it.
So, why not have two different modes of discussion between agents, depending on whether decisions are made or executed?
An example of non-pyramidal governance is holocracy.
Paragraph from https://www.investopedia.com/terms/h/holacracy.asp :
What Is Holacracy?
Holacracy is a system of corporate governance whereby members of a team or business form distinct, autonomous, yet symbiotic, teams to accomplish tasks and company goals. The concept of a corporate hierarchy is discarded in favor of a fluid organizational structure where employees have the ability to make key decisions within their own area of authority.
In my memory, it was presented like this:
Each team draws a representative, who acts as spokesperson.
The representatives from each team come together to form a circle.
If there are several circles of representatives, it starts again: each circle of representatives appoints a representative, and so on.
Obviously, this system would have to be adapted for agent swarm.
What got me thinking about this while watching today's video is the savings in tokens that it enables, by sending information back in a synthesized form.
And I thought, if we're starting from scratch (unlike a human enterprise, where habits can be tenacious), we might as well apply the methods that work best in a radical way!
Beta Was this translation helpful? Give feedback.
All reactions