DevelopmentalSystems

• For more on Charles Peirce, a good starting place is: ARISBE -- the Peirce gateway

• For more on Robert Rosen, two good starting places are:
  • T. Gwinn's Rosen Gateway, "Robert Rosen Complexity and Life"
  • BioTheory -- An online journal dedicated to coninuing Rosennean explorations of complexity
    • My personal favorite for a general introduction to Rosen's ideas is "The Devils Advocate written by Judith Rosen, and on the BioTheory site under the section "Judith Rosen"

• Jack Maze's work focusses on empirical investigations of developmental systems in botany.

• Stan Salthe's work focusses on the natural philosophy of developmental systems

• Kali Robson's work is an attempt to provide an introduction to the general reader of the notion of "emergence" and its role in development and evolution.

• Koichiro Matsuno's work combines empirical and theoretical investigations into what can be called "the nascent moment" when systems reach a level of complexity that their dynamics are dominated by information flow rather than energetics, and they transition from "physcial system" to "biological system".

To learn more about how to use this TWiki -- go to HarmenyTWikiGuest

Developmental Systems: Systems whose patterns of change with time, change with time. Try not to turn your head right around, grokking the concept. Special Case: Dynamical Systems. Let's nickname such systems DevSys and try and come up with a list of their properties and behaviours

For work of Jack Maze online go to: MazeResearchProgram

• Dave and I have both talked about an Idea Generator Company -- focussed on creating new technologies.
• Harmeny could act as that company -- would require re-visioning it and structureing. Could create an independant Harmeny USA. collab. with the already existing Canadian Company.
• Figure out simplest structure possible. Legally simple. Conceptually simple. Joint Venture*
• Action Item Come up with series of options for future work together (2 options max, as we are simple fellows). Deadline: Decide on options by end of August, have legally in place by end of Sept, so we can write proposals.

What We Want to Get Out of It

• Dave: Technology for distributed computing middleware platform.
  • Turn into Licensable technology and revenue.
• Mishtu: Develop the core open sourece technology, Xaya that can be adapted for this middleware platform.
  • Develop services around the open source Xaya technology.
• Action Item Dave and Mishtu's vision above, needs to be put into a simple paragraph or two, that constitutes agreement. See as an example link to : http://www.w3c.org/consortium/points
• Tyler: ????
• Ian: ???

Basic Idea -- High Level Architecture

• Use existing "web friendly" technology that is mature: Python, Postgres, R for agents, storage and analysis.
• Develop a general model of distributed information (done last winter).
• Develop a general sampling process for large distributed systems (monitoring complexity, discussed in April) to be able to track system stability, and system transition. This takes a statistical approach to large computational systems.
• Sampling process will act as a "metabolism", lower level more local signals cascade to create higher level global signals.
• There has to be a message queing technology, that inter-operates with the sampling process. E.g. what happens to messages if communication links go down???
• Customize to Energy Industry: Customize the agents, and the analytical componenent.

What is Unique

• Our concept of Agent is more robust than usual -- with the augmented analytical capabilities.
• Our integration of statistical sampling into a fault tolerant system
• By design the architecture supports lower level loss of info -- and as information moves "upstream", there is less, but higher quality information.
• Focus on "information" and what happens to information, rather than focus on the plumbing technology.

Artifacts to Create out of this venture

• DAVE: Product "Rapid Agent Development Tools" to build applications, and add extra functionality to the Open Source core.

Starting Point

• We need an R & D project that allows us to design and prototype the core.
• Probably can move fastest by tieing to energy domain -- but also propose that it is spun out as an open source product, with open source community developed around it.
• Need clear value proposition -- what is being added to the body of knowledge that is NEW as a result of this project.
  • Need an application domain in energy industry it applies to.
  • Need an analysis of what is currently missing in that field.

Dave Thoughts

• There is a core of what we are talking about building, with services, such as messaging, security, etc.
• There is the agents themselves: fairly light weight, tied to analytic core, using programming languages like python and java.
• We talked about dividing architecture up into lower level agents, simple and extremely distributed, with higher level agents with more processing power, less distributed. Above that, a presentation layer tied to web services.
• Some other issues to be sorted out: message transport (part of core service offering??). Service discovery (use an existing standard or innovate). Work flow among agents/objects/services. Work flow creates the distributed autonomous collaboration framework. These items are high level services.
• Some of the services at the core -- e.g. directory services -- what do we need to use for that piece?
• How are the high level architecture agents going to be organized. E.g. Single agent is a node. Group of nodes is a community. Group of communities is a ???? Tie security priviledges to such a grouping structure.
• In the agent world -- there are a number of FIPA standards (foundation for intelligent physical agents). Are we going to be compliant with various existing platforms, or add our own.

Next Steps

1. Mb set up an internal Wiki to collaborate on. collaboration meeting every 2 weeks. Arrange actual dates to collaborate via email -- so can adjust meetings to our schedules. Mb will send a note out, at beginning, and mid-month (1st and 15th of month emails).
2. Define what legal structure allows us all to work together.
3. Dave Create a set of use cases for the problem domain
4. Mb tie use cases to architecture
5. 1 Page high level architecture diagram and description. Dave/Mb iterative. Tyler, Ian if they wish to play with this.

Next Contact

Email Dave on August 15th

Computational Metabolism Idea For GridAgent^? / GetSmart^? distributed system monitoring technology

| DevelopmentalSystemsQuesTions^? | DevelopmentalSystemsStorieS^? | DevelopmentalSystemsDreAms^? |

DevelopmentalSystems

| DevelopmentalSystemsLinkedTo^? |

-- MishtuBanerjee - 15 Mar 2003