Structuring the OAE around agents
Lynn Foster
Public
Seen by 72
All Agents All the Way Down....
This is another thought experiment for how to structure the Open App Ecosystem. The focus here is on economic activity, but it applies to everything people might want to include in an OAE. Feedback and discussion very welcome!
What if we did a radically decentralized version structured around agents (people, organizations)? (This is not a new idea, has been thought about and worked on by others before. But this is to take those concepts and think about using them to structure a whole OAE - the one we are working towards here.)
Every person would operate from their own Person node (which is a personal agent). Every organization/group would be an Organization node (an organizational agent). So the ecosystem centers around agents. Agents are connected to each other by various relationships or activities. Everything that happens is stored by all the agents that are involved, wherever they store their stuff.
So, let's say Jim, Tibor, and Draft decide to create something together which they want to exchange with other people. Like some fig, pig, and goat pizza.
Their Personal Agents will agree on the rules and plans for how to do it.
They will need to create some kind of joint identity which will be an Organization node (which is an organizational agent), so that other Person nodes can order pizza from it, and so that their work and other contributions can be combined into the actual pizzas. Let's call it the Open Pizza Bistro.
The Organization node could deploy Process Bots for each pizza that needs to be created, with the size and maybe some selected special ingredients or variation, like skip the pig and goat, we want vegan.
On request from Jim, Tibor, or Draft, the Process Bot will list the work requirements for this and other upcoming orders. Each of the people will commit to which work they want to do. Since they are not employees, when they actually log their work (using the Process Bot), their work will be an economic transfer of resources from them to the Open Pizza Bistro. If (for example) Bob wants to exchange some dessert for some pizza, then all of the people who had transferred some work to the process of making that pizza will be due for a share of the dessert.
Bob will need to order the pizza from some node, because it is not coming from Jim, Tibor, or Draft individually. And the organization could deploy a bot that takes orders. But sooner or later, some live people will need to do some live work, and maybe one of them will take orders, too, since bots can be a little too rigid about pizza order variations, and this dessert exchange idea might also confuse the bot, so Bob might want to connect with a human. Likewise, when it comes to ordering ingredients, and especially making agreements with ingredient suppliers, a bot might be too limited. So the Person nodes will need to act for the Organization node in many cases.
So the people who created the organization will need to make some decisions about who can do what in representing the organization. This is similar to organizations in real life. An organization can have agency (make agreements with other agents, for example) but the agreements are always executed by people who are related to the organization. So Jim, Tibor, and Draft will need to decide who has what kind of role in the Open Pizza Bistro, so the Bistro can execute its agency.
Notes
There is no monolithic system here that Jim, Tibor, and Draft need to use. Each of them is always operating from their own personal node, hosted by them or a hosting service they choose. They probably each belong to several other organizations. From a personal node, the person could potentially see everything they are involved with - their work (job, freelance, for fun, to help out a community), their offers/requests/exchanges, their accounts of different kinds, their social network, etc.
An operational system is composed of all of the resource flows or other connections among all the related nodes. Each organizational node will have in its database records of all of the economic and other events that happened in its scope. Each person node will have the same. It would be possible to look at larger pictures from some overview scope, for example, a bioregion. Or it would be possible to follow a chains of events to see where the wheat came from that went into the pizza, to see how local it was, and how much waste was created. Or to see what your friends are posting about.
So, there wouldn't be lots of different platforms that want people to join them any more, except perhaps transitionally. There would be no NRP, no Odoo, no DigLife, no Communecter. (With all due respect of course!) All the great and useful functionality from those platform apps would be broken out into pieces that could be plugged into the decentralized agent-based apps so people could see and interact with others, either as part of an organization or not. (Like the DigLife bot and diagrams and the Communecter maps and a lot of the UI could still work.)
Oliver Kalleinen Fri 15 Dec 2017 10:49AM
Hi Lynn, thanks for the posting. You mentioned that this agent based decentralised model has been researched and tested by others, do you have some references, links, names etc.? I see some overlap with sociocracy and their circle idea, but I might be wrong. The Pizza example could be slightly misleading, since we mostly do not work on pizzas (I assume...)
Lynn Foster Fri 15 Dec 2017 12:32PM
Hi @oliverkalleinen . Re. agent based decentralised apps, what I know (and others may know more):
* Secure Scuttlebutt - https://www.scuttlebutt.nz/
* Solid - https://wiki.p2pfoundation.net/Solid, https://solid.mit.edu
* PLP - Portable Linked Profiles http://hackers4peace.viewdocs.io/plp-docs/directory/ (probably out of date, but this and some other efforts were being worked on a few years ago by @elfpavlik and some others, possibly coming out of a OuiShare conference?)
I don't know if there is overlap with Sociocracy, that would be interesting. In general, this is a technical proposal, although there are social implications, like who owns and controls your data (you!), how people organize themselves into groups of their choosing, things like that. I haven't studied Sociocracy, so I'd be interested in your thoughts.
Re. the pizza example, actually some of us are very interested in the OAE as it relates to creating software that can be used to coordinate economic production and distribution in a post-capitalist situation. So for us it is a relevant example :slight_smile: - as would be 3d printing, open hardware design, food growing / processing / distribution, etc. Of course, the OAE will also want to support administrative, social, governance related things.
Bob Haugen Fri 15 Dec 2017 4:39PM
In a conversation in Scuttlebutt about this topic, I referenced @gregorycassel 's work on agreement-based organizations:
I'm assuming that the people meet and using some agreement-forming protocols to make some agreements on all of those rules. The consensus required would be only among the people who are involved in the agreement.
Who can join, and how they join, would be decided in those agreements.
Thereafter, as more people join, the rules might get changed, and in that case the logic scope of consensus is the same.
For economic exchanges, the required scope of consensus is the agents involved in the exchange.
This is a lot more like #holochain than blockchain.
For agreement-forming protocols, the basic one is Conversation for Action. Greg Cassel is working on a vocabulary and more refinement of CfA for organizational governance
And in another Loomio OAE thread, I just saw that Greg posted a couple of new links to his work:
a two-part essay "Developing P2P Networking and Nonlinear Dialogue" broadly outlining my key design priorities:
In other words, in conjunction with the purely economic interactions, we will also need the organizational interactions.
Greg Cassel Fri 15 Dec 2017 4:48PM
I'm glad you noticed that! Unfortunately I've been horrendously tardy in focusing on Conversation for Action, but it's an increasingly clear (to me) component of the documents listed in my new essay (Part Two) which you linked above.
CfA is a big part of both the Structured Conversation section of this Community Markup Language doc, and in Resource Development and Support System.
I'm really tardy at focusing on some of these key modules, because I feel like I'm rewiring my own nervous system constantly for a properly networked approach which only uses hierarchical organization when that's truly rational. I'll try to get better at connections and mapping.
Greg Cassel Mon 5 Feb 2018 6:24PM
Note: here's my first try at a diagram which broadly illustrates relationships between models referred to above.
Greg Cassel Fri 15 Dec 2017 4:40PM
This human-centered/ person-centered approach is basically sound IMO, and similar to what I suggest in Peer-to-Peer Digital Networking/ Sources and Destinations with its emphasis on Agents, Devices, Applications and Addresses.
In my models, User Identities, Accounts and Bots may perform specific potential functions for Agents.
Unfortunately I haven't been able to focus on that section of text recently, due to other creative demands. I'll be reformatting that text as modular entries on Github, closely interlinked with Modular Organization Terminology.
Lynn Foster Fri 15 Dec 2017 6:50PM
By the way, I'm aware that there are bunches of architectural questions to be looked at on this topic. And also aware that I am not quite technical enough to be able to speak completely coherently about them.
But one of them is does this version of the OAE live on the web, or use one of the technologies that doesn't use the web protocol layer (like ssb or holochain), or are there different oae's for these different experimental architectures and hopefully some day they will connect, or..... ?
And how does it relate to mobile devices?
I think right now (could be very wrong) that we want to be on the web for most people, and also should be very willing to experiment with other underlying technologies. But... another piece of this puzzle is that for many (most?) of the things people will want to use an OAE for, the web isn't really necessary, something like ssb, where you are interacting with people in your network is enough. But... for some things (anything that wants to be published beyond your network) - open documents, offers/requests, open hardware designs, etc. - the information needs to be widely broadcast, not just within your network.
Other architectural thoughts?
Greg Cassel Fri 15 Dec 2017 7:41PM
for many (most?) of the things people will want to use an OAE for, the web isn't really necessary, something like ssb, where you are interacting with people in your network is enough. But... for some things (anything that wants to be published beyond your network) - open documents, offers/requests, open hardware designs, etc. - the information needs to be widely broadcast, not just within your network.
I agree that many important uses will only require small networks. I strongly believe that some uses should be broadcast publicly worldwide, and I don't think that private and public uses need to (or should be) deeply separated in terms of technical architecture-- although communities certainly need to develop strong standards regarding the public broadcast of private media (or adaptations of it.)
I admire ssb and holochain. I don't know yet whether they're technically capable of scaling indefinitely for a global network of networks on the scale of the World Wide Web. (That probably depends greatly on formal details of p2p network development, such as hierarchical levels of organization and signal routing and storage protocols.) I also don't know yet if the WWW & Internet protocols-- and their stewards-- are *in*capable of adapting to suit the types of agent-to-agent networking we'd prefer to develop.
Bob Haugen Fri 15 Dec 2017 7:56PM
I think the Web is capable of supporting these types of networks, but the early adopters we need are moving past the Web because it has been captured by surveillance capitalism.
I still think we need the Web and it is recoverable, but some of the people who understand what we are trying to do and can help do it are in the crazier fringe groups.
Bob Haugen · Mon 18 Dec 2017 4:49PM
It looks like pretty much the same idea, though...or at least part of the same idea.