Decentralized networks

For a long time the idea has been floating around to have independent and decentralized networks as an alternative to the existing centralized (= one user database, one management) hospitality networks.

The idea is interesting, but has some difficult aspects, that can be discussed on this page.

Why?
Is decentralized always better?

Always? Probably not. But that should not stop anyone from exploring!

The great benefit of decentralized networks is that they allow to experiment with different technology, different styles of moderation, different safety and trust features, different search features, different looks, different peer groups.

Local computers with normal browsers
You can find them at home, at friends' places, at school, in Internet Cafés. They can be used for surfing the web, but we can't expect any special software on these machines.

(Small) social networks with user profiles
This is where you would upload your pictures, write about your hobbies, etc. You could even use different sites for that: Have your photos on flickr, but your profile information somewhere else.

One of these networks can store your friend links, another your messages..

Meeting places (groups, forums)
If user profiles are distributed on different networks, we still want groups and forums that connect people from different networks. The most simple example is: A and B each have a blog, that are technically independent of each other. A leaves a comment on blog B, with a link pointing to blog A.

The problem is that B can't be sure if A is really A, or if it C who claims to be A to spread all kinds of libel and nonsense. So, the advanced version would provide a possibility for A to prove that it is truly A (the person with blog A), and not someone else.

Dedicated systems for friend links, messages, different kinds of interaction
Honestly, I have no idea how this would look like :)

Services for Moderation and Conflict Handling
Similar to the Couchsurfing MDST, there could be services that can filter out inappropriate references, or decide which profiles are relevant to you or not. Sounds quite abstract and unclear, needs to be fleshed out. Somewhere else I used the term "reference realms".

Aggregation Services
This can be bigger players (google, whatever) that scan user profiles on different sites and offer this information for searching or to deliver pages with information mashed up from different sources.

We can want this or not, we can fight these aggregation services or use them. In any case, they are part of the picture, and can't be ignored.

How it can work
A lot can be said about decentralized networks, but in the context of hospitality exchange the most important is user profiles, (location-based) member search, messages between members, and trust / friend connections. All of this has to work across different networks, otherwise it won't be really useful.

The biggest challenges here are privacy and proof of identity: How can one restrict the visibility of information that is shared across different networks? And how can I prove that I am the owner of profile A on network X, when interacting with members form network Y?

User accounts / profiles
The idea would mean that there is not one big network where you create a user account, but instead you would choose one or more networks where you want to create an account and have the information you define stored in the database.

Problems:
 * Multi-accounts: Duplicate search results, and redundant information to update in different places.
 * Agony of choice: How to choose my favorite network?

Solutions:
 * Tools for import, export and profile migration
 * Auto-update a profile on site X with information from site Y.

Searching member profiles
To make member search useful, it has to be cross-network. This means, a search request either has to request information from all connected networks, or you need some kind of search engine that crawls member profiles on different networks and caches the information.

Problems:
 * Privacy: Information that is shared between an arbitrary number of networks is practically public information.

Solutions:

Messages from one user to another
You go to someone's profile on network X, click "send message", type the message, submit. You leave a link to your own profile on network Y.

Problems:
 * You need to type in the link to your own profile manually, which is inconvenient, and you can make typos.
 * You can type in a fake link to someone else's profile.

Solutions:
 * OpenID? Does that allow to identify you as a member of a different network?

Trust/Friend links between members
Just like the references or friend links in couchsurfing, you could create a trust link from profile A in network X to profile B in network Y.

This can only work if there is a way to prove that your friend link does in fact come from profile A in network X, and not from somewhere else. This is the same problem as with messages.

Profile data visibility constraints
In traditional networks it is possible to show some profile information "only to my friends", or have some other constraints. How can I control that if the friends' profiles are on separate networks?

Trust links between networks
Just like there can be a trust network of people, there can also be a trust network of networks..

Examples

 * Crash at Mine (?)
 * Noserub
 * Q: How does Noserub solve the above "problems" ?
 * OpenSocial? Shindig?