Re: [Social-discuss] Push vs. Pull, Events vs. Data

[Melvin Carvalho]

2010/3/9 Hellekin O. Wolf <address@hidden>

Some more thoughts regarding PSYC's approach to social networking:

- Let users handle trust, defaulting to none
- Push rather than pull
- Multicast when possible
- Allow users to share contents from their HDD, according to trust math

In other words, PSYC proposes an event-oriented multicast push
approach vs. a data-oriented unicast pull approach (Web / RDF)

I agree PSYC has a lot of desirable features, and there's much to be learnt from it.

I'm not sure the web is always pull.  Firewalls tend to make things unidirectional unless you have some kind of NAT busting technology.  Websockets (HTML5), for example, are bidirectional, as are many URI's on the web.  There's no reason why the web should not work with both push and pull.

HTTP is one specific form of URI which has the feature that it is dereferencable, and therefore, a good candicate for storing data structures, particularly data structures with global reach.   RDF has an Events Ontology as well as covering most other data formats, such as profile, trust, commerce etc, so im not sure there's a conflict between events and data.

I'd rather take the best of all worlds than saying anything is the 'one true solution'.  However, I'd struggle to see something scaling with the Web if it ignores linked data principles [1]

[1] http://www.w3.org/DesignIssues/LinkedData.html
 

x
==
hk