The Semantic Web Is Sophisticated Syndication (SWISS)

Going beyond Peer-to-Peer with Many-to-Many information sharing.

Table of Contents

Synopsis

SWISS enables the disintermediation of WikiWiki, blogging, and other collaborative web application systems.

Motivation

Most current web collaboration technology depend on central administration in order to operate (there are a few notable exceptions such as SMTP, and, to an extent, USENET and IRC). While that architecture often suits the commercial interests of the typical operators of such systems, it also fundamentally limits the number of people who can use them. Not only are there intractable issues involving economics and policies (both administrative and editorial[2]), but users also surrender their data freedom and privacy rights to the service provider.

This situation is very much like the 1960s in which computer services were centrally controlled with all of the compromises of users' interests which that entails. What users desire is the independence and availability of personal computer desktop systems, particularly for essential applications and their data. Of course they also desire the freedom from maintenance and broad reach of managed web application services.

Another severe flaw in centralized architectures is that they don't scale effectively. When a system is fortunate enough to be wildly successful, whether briefly as in the "Slashdot effect" or longer term like MySpace[1], then it faces a surge in load that is overwhelming and can be extremely difficult and expensive to even partially satisfy. As Google's success illustrates, it is essential to have a system architecture that is inherently distributed in order to reach everyone in the world (see HowGoogleWorks). Google Desktop also illustrates that a comprehensive solution must support operations with and without an Internet connection.

Architecture

SWISS is a technique for sharing information (collaboration) that only requires static content servers for public services. In addition these public servers don't "own" the information they present but instead are primarily used to provide aggregated views of data and topical reference points.

The needed dynamic services are scalable ranging from single user (desktop hosted personal web application servers) up to whatever is desired. Well behaved application servers will also not own their data, but rather expose it fully (to those with appropriate permission) in transportable RDF.

The user benefits provided by the editorial and administrative policies of centralized Wiki servers becomes a smorgasbord of opt-in choices from a menu of automated rules and grassroots user affinity groups.

The practice of these policy systems will be a lot like security policies in Internet Explorer. The user gets to choose from a detailed list of decisions, which come in a few precooked prix fixe choices as well as ala carte. Then in enterprise and corporate deployment, those security policy choices and their maintenance for whole groups of users is delegated to IT administrators.

While SWISS operates using polling for syndication in the public Internet. Thus SWISS is extending the successful RSS technology into applications beyond blogs.

But users like the immediacy of the web, and as the number of users gets large polling runs into latency vs bandwidth & performance issues (although I have an idea for a nifty hack), so we also support messaging. While there are numerous messaging technologies out there, SMTP email is by far the most successful and must be supported (not surprising since it is the foundational Internet collaboration medium). Google again is showing the way in how the web and email can be combined for a top-notch user experience with GMail. For chat-style application IRC is the granddaddy, but XMPP is probably the way to go for instant messaging (although SIP is important too). The Dylan Wiki supports XMPP for change notification.

Social Web Services

Web Clipping

Bloglines has a clipping feature
http://www.bloglines.com/help/faq#saveditems
Scrapbook Firefox extension
https://addons.mozilla.org/firefox/427/. This tool would be a great base for an annotation extension for Firefox, except that involves Firefox extension hacking.
Clipmarks
http://clipmarks.com/
Clipmarks Firefox extension
https://addons.mozilla.org/firefox/1407/

Identity & Authentication

Identity deserves it's own topic.

Spam

Automatic spam checking is another big topic. In a SWISS blog or Wiki, managing comment spam is trivial; just drop the offending source from the syndication list and regenerate the page.

Spamato is an OSS (GPL) spam checking framework in Java.

This article discusses processing email messages with Mule and Spring.

What's semantic about it?

The reason semantic web technology, and particularly RDF, is important is because it provides the data format needed to link together the information needed for social/collaboration applications.

Marc Eisenstadt cites some nice resources on how this will work: http://kmi.open.ac.uk/people/marc/2006/06/29/social-semantic-web-20-peter-mika/. See also Folksonomies.

And the thing that will make this Semantic Web jazz as successful as HTML is the lovely Microformat/CSS friendly Embedded RDF.

DBin

DBin implements an ideal "native" (that is RDF) disintermediated many-to-many data sharing framework.

Getting Graphs

Differencing in the sense of Harmony will get the documents into RDF (and back in some cases), while OOHTML will enable RDF to get into documents.

Jon Udell talks about OpenFount; a tool for implementing services hosted by a web storage server (Amazon S3): http://weblog.infoworld.com/udell/2006/07/07.html#a1483

Amazon now has Simple Queue Service in addition to their Simple Storage Service. That would be an attractive alternative to SMTP (especially since it is widely blocked now) or OpenFount's OSCS for AJAX SWISS applications, except that they currently are only available on separate hosts (thus inaccessible to ordinary Javascript). Fortunately they know about the issue and I'm sure it will be resolved (with hosting options on the accounts I expect).

This article Architectures of Participation: The Next Big Thing discusses Metcalfe's law at work in "Web 2.0" social networking sites. It also cites YouTube as the "#1" Internet site because of the huge bandwidth it is using. That illustrates one of the fundamental weaknesses in any collaboration technology that depends on centralized administration (and Akamai is only a bandaid).

In this interview of Tim Bray and Radia Perlman Richard MacManus has the Sun folk espousing exactly the opposite view. They assert that "true" P2P won't work because people want services that depend on identity (true) and that requires centralized servers (false). But this isn't surprising since Tim has already demonstrated he doesn't get it by supporting Atom and then being a hypocrite by telling folks to not create spurious XML formats. Anyhow, while they can be forgiven for not knowing about SWISS because it isn't built yet, a clear refutation of the idea that central servers are needed is DBin.

Marc Eisenstadt blogs on how the Buddy list is key to future publish/aggregate world. Actually he says how he doesn't think that is quite true, but links to those who do such as Steve Boyd who also talks about The Conversational index.

http://www.redmonk.com/cote/2007/01/04/desperately-seeking-personal-directories-the-identity-20-wind-mill-or-file-under-declarative-living/

RDF & REST are a perfect pair

http://blogs.sun.com/bblfish/entry/rest_without_rdf_is_only

Related Services

These are mostly server-mediated services, but are good examples of the kinds of collaboration tools I like.

Diggo

Diggo provides some of the information sharing tools that IFCX will have (particularly the Annotea annotation features), but it is a typical centralized service:
http://www.diigo.com/

Some articles on Diggo:
http://www.solutionwatch.com/474/diigo-launches-more-than-just-bookmarking/
http://reviews.cnet.com/4531-10921_7-6624081.html?tag=nl.e404

ArtFeed

Easy RSS for artists. http://www.artfeed.org/

LOCKSS

Lots Of Copies Keep Stuff Safe is a distributed digital library project. http://www.lockss.org/

DabbleDB

DabbleDB shows what can be done manually now that would be automagic if folks used OOHTML.

CoComment

http://www.cocomment.com/

Online Copyright Syndication

This article talks about Lisensa which offers a copyright clearing service for bloggers.

VOIP & IM

While XMPP is suited for web pages, SIP is the open VOIP protocol. The two come together for the ultimate in interactivity.

SIP Communicator http://www.sip-communicator.org/ is VOIP in Java (although FLASH solutions would probably be better for most folks, it even supports cameras).

[#1] See this Baseline Magazine article "Inside MySpace" on MySpaces's ramp up, which also mentions how these scaling issues stalled Friendster and allowed MySpace to take the lead and achieve crtical mass.

[#2] Prime examples being disputes over such things as the application of Wikipedia:Wikipedia:Deletion_policy to most any Wikipedia:Criticism_of_Wikipedia. That has actually now resulted in the high visibility fork Wikipedia: Citizendium which is here http://www.citizendium.org/.