Modular logistics: brainstorming thread

[cwgoes]

Basic concept:

Unify transport logistics providers under a common interface. Instead of contracting with individual transport providers (airlines, Uber, trains, etc.), users contract with insurance providers, and buy a contract to transport them (and/or goods) from their origin to their destination, with a payout function depending on whether they arrive at the destination on time or not (could vary, e.g. declining slightly if a bit late, then free if more than X hours late - this is similar e.g. to some European transit regulations). These insurance providers then purchase individual legs of the trip from different actual transit providers (such as drivers, flight operators, train operators, etc.) in ways compatible with the user’s preferences (user can restrict which transit providers are allowed). These legs can be resold and repurchased during the trip if later, more optimal configurations become possible (or perhaps if the journey is delayed and the insurance provider would rather pay for a more expensive, faster leg than lose the contract payout).

Topics to consider:

• Market structures of different payout contracts, legs, transit hubs (bus/train stations, airports, etc.)
• Market structures of insurance providers and transit providers, potential centralization concerns
• User experience of such an application
• Solving, mapping, and constraints over three dimensional space and time
• How to express user preferences in the appropriate sorts of ways

This is probably most similar to existing travel aggregation applications such as:

• Some public transit apps in Europe where you can at least book source to destination on trains
• Multi-hop routes on airlines (but the market is still quite fragmented)
• Ctrip, maybe (I’m not too familiar)

But none of them are quite as unified as what is being proposed here, and don’t include this insurance structure or potential dynamic reconfiguration (which should improve overall system efficiency).

cc @apriori for your context - does this make sense?

[apriori]

Modular/Logistics

Lending Libraries

For lending libraries, a well generalizable model - books exist and people would like to request books in certain places. Books also need to stored, they take up physical space and have unique identifiers with ispn. If I am in Berlin, and you are coming through New York I would like to have this intent, and you are coming to Berlin at some point.

You publish an intent of your planned trajectory and carrying capacity. I will spend this much time there and have space for 10 books in my luggage. Even non-interactively or interactively - be nice to pick up books for these people and deliver them. Capacity, travel itinerary and what you can carry.

Ride share/transport general

• Carry capacity - uber driver have 3 seats in my car and plan to have sthis
• Ride-share from belrlin to hamburg 2 seats left, and you can chip in gas money.

I want to outsource the book to the public library in my neighborhood, after can be mapped to ride-share.

If you have a festival needs xyz need so much would cable, time for helpers, counter intents, i have wood transport food. 3 meals a day for 100 people a group can sey we can provide these given these resource inputs

Where the lending library breaks down is good that can be used up. With food, you have state transitions from food to heat. Fuel → propulsion. Maybe the festival org or ride-share is the more universal model. All of these models should be edge cases of each other. Relatively universalized model.

Nice ontology for physical logistics which is sufficiently standardized - focus on explaining some specific examples (ride-sharing) people know and is obvious because it interacts with trust models. Uber is acting as a trust mediator for the drivers. How will this actually look. There will still be rating providers like Uber insurance decoupled from operation.

• Everyone can be a riding provider or larger ones that aggregate data as one day. When you come to Berlin, you probably want the Berlin ratings agency as a proxy, maybe vetted by the city council or something.
• Good to go into the - nice benefit you get here of sort of aligning your usage of applications with your own personal trust network. Either way friends then pay Uber driver.
• Or just because its higher value trust information - uber drivers are vetted no gaurantee they will behave in a particualr way if this is a riding provider clued in on at risk population this would be much different

Community Lending Library (Shed)

Community lending library - clearly benefits from c/d and automated infrastructure. Community 3D printing library where scheduling is handled by Anoma application and there is some coordination of pickup and distribution; e.g., drill hammer all friends need a drill hammer

For the lending out, you can re-use the trust networks. I only lend my more expensive tools to friends who have given more cheap tools intact. Specialized cryptography things and don’t want to give it to someone who knows how

Scale-invariant, so if you want to organize a festival with 10k people, need more than 1 shed. CCC events different groups that have their stuff, but some coordination to deduplicate resource usage. There could be some efficiency gains if the discovery process could be approved for this. These are the communities that would be keen to get on board for some testing. Are they useful? lending library need to scan stuff. If you want to inventory all of your stuff for a festival requires qr codes and a certain degree of logistics. CCC has standardized containers which are packable and countable, labeling every single item is somewhat infeasible. If we get the lending library for books, every item is self-labeled already, and it is a nice thing to test it out. We test it as Anoma and scale from there to something like the tool shed or whatever.

• Do we need an ontology?
  • I need a cab; global
  • I need Berlin transport; local

Locality in the sense that people can have whatever ontologies they want, but only need them to cooperate. We can easily support translations. Say you have a different word for car, and you accept the accuracy of the translation, you can use your ontology and the system can translate for you. I want x to get from a to b when you use FedEx is not specific.

• Anoma- everyone becomes a part-time FedEx driver when they take an item, and they get Anoma distribution network. trust logistics are complex, you need a web- of-trust.

Web-of-Trust

• Web-of-trust as affordance - mechanics of consensus are not necessary.
  • Network of trust
  • I want to find a driver that is within 3 hops
  • directed and is a graph but not acyclic
  • it has edge weights - dimensions and levels; some are more general and not fungible.

Ratings

When using the ride-sharing app I would like to include Ds rating in measurement. A question of trust with a new dimension of trust. Do I trust this person’s driving ability, this much, and this persons’ judgement?

• I trust D 100% - integrates opinion
• Its not even about trusting D’s judgement and my judgement about something divergent.
• Our ratings of cakes will be divergent, and Chris will estimate this divergence.
• How much do I weight Ds opinion about the baking skills?

In the ride-sharing example, a ratings provider can do even more fancy and useful things like autocorrelation as in if you and I take a bunch of car trips and cakes we can use the system for how ratings align, which is a good future predictor

We just want to formulate a basic structure for what exists and put measurement on the client side, so you can learn from your own past and other people’s past and make adjustments for what other people’s rating systems would be.

What are the failure modes? If you are an Uber driver and ratings drop below some threshold, you don’t get more riders. If you are a centralized company, you get an appeals process like this.

How do people get the information? People get info from ratings agency, people share info for them. As long as this type of behavior is localized and not such a thing as there is only one driver. This is the failure case to guard against.

Example

What does it mean if someone rates this guy 0 because he is physically violent?

• Implement the machinery to implement these processes - appeals process
• Some of the impetus of appeals comes from central ratings agencies.

Let’s find out whether our architecture will make it easier or not to have monopoly. We can at least reduce the friction for users using the local trust data.

Soul Bound Tokens - possibly for ratings.