1. Since none of the sites work the same, it would be a little difficult to create a standard.

Posted at 8:04PM on Dec 18th 2005 by ragu

2. 1st, most SNS is really, really simple, so all you need is some PHP, Perl or ColdFusion hackers. There's really no need for any standards (other than data exchange protocols) to speak of. Look at HotOrNot or even any of the online dating services. Just a bunch of records that folks create for others to look up.

Now others SNS apps (LinkedIn or SocialText), need features that many enterprise collaboration apps also need:
1) security (authentication, ACL, SSL, etc.)
2) near real-time (or real-time) updates and dynamic information (multi-level connections)
3) AND scalability (it's hard to services millions of users interacting with each other)

Now if you look at J2EE or .NET spec/std., you pretty much can get these core components, but the catch is that they're complex and quite difficult to harness (just look at how much JBoss charges for consulting). But it's interesting to look at the early collaboration (social software) infrastructure providers (PlaceWare, now part of M$; Lotus Notes, ignored the WWW; Zaplets, suck in spam; and others). They in many ways were pioneers in trying to provide building blocks for social (collaboration) apps.

If you look around now, there are a few providers of application infrastructure for building SNS apps, but it's proprietary and expensive, targeting high-end enterprise customers.

Now from a open source view, goto Apache.org, collect your favorite APIs, write a few thousands lines of Java code, make connections with MySQL, and then brew your favorite SNS. So maybe it's time for someone to start an SNS Apache project?

Posted at 8:04PM on Dec 18th 2005 by John Ko

3. "1st, most SNS is really, really simple"

bullshit. or...actually, you might be right, up to a point, scalability wise.

I honestly can't think of any 'social networking' site that has any "real" load, except for maybe Friendster, because of their users. I actually agree with your point #3:

3) AND scalability

That's where the real fun is, IMHO....but I'm a performance-obsessed freak.

I'd say that I would be very surprised to see any of Friendster, LinkedIn, Spoke, Orkut, or Tribe give up any of what they have learned, especially along the lines of graph computation or database replication, due to the instant-ness requirement of having things up to date immediately. From what I can tell, they all have leaders in their fields (software developers) who are even book authors in their craft, so while I think it would be very fun for the public to discuss the technical implementations in public, I don't expect any of the people that actually work at the major SNS sites to talk much about what they're doing.

Posted at 8:04PM on Dec 18th 2005 by joe

4. Certainly, there are a class of social softwares which are simple to construct. http://del.icio.us/ is a particularly interesting one.

Building an application platform which can scale, however, is a different story entirely as joe points out. J2EE is a red herring -- scalability is a function of simplicity -- and building a complex system in a simple way is _hard_ ... tools are not a magic bullet, and if you believe you can get them for free from micros~1, you should go read Mythical Man Month again.

And, no, we [I'm an early engineer with Spoke Software] are not really jumping at the prospect of brain-dumping our hard-fought knowledge to anyone else. :)

Posted at 8:04PM on Dec 18th 2005 by Josh Sled

5. It would take a whole lot of monkeys to get J2EE / JBoss and .Net working for a SNS. As #3 points out the graph calculation real-time is a tough one especially as you add more users, the CPU cost to calculate the graph goes up exponentially. You have to traverse, calculate, and update the graph for every user and modify it every time some one new is connected. Scale is everything. Being able to serve a data set like that has got to be tough. Let me tell you, like #4 says... no one is going to give up their special sauce.

Posted at 8:04PM on Dec 18th 2005 by ragu

6. From a business perspective, it makes little sense for their to be system architecture standards at this point. companies like friendster, linkedin, myspace and ryze are on their way to creating successful businesses because they have applications built around significant networks of people.

The key is that they all have different applications and UI's built around the network. Saying Person A is connected to Person B is not all that difficult to architect. Either is writing an endorsement system or a pass along contact request system, but it is these features that differentiate linkedin from its competitors.

There are of course people that are going to continue to create clones of successful SNS sites, but the more successful ones are going to be the ones that do something different.

Also, there are different ways of establishing connections. Most of the popular sites have mutually approved connections. However, there are fan/nod based systems as well such as my Company, WhizSpark and FOAF, where the connection is one a way relationship: celebrity to fan. There are also the crazy people out there that are trying to label connections with taxonomy. Noone ever agrees on taxonomy.

All that said, it does make sense to have standards for communicating data. Some people will be comfortable with using open to the world FOAF based systems and tools should be built that allow them to plug their FOAF into SNS systems. There is also value in creating standards for communicating data in and out of SNS systems like discussion threads at Tribe.net or events at upcoming.org and coming soon to WhizSpark.

When we start to do this, we make the web what it was meant to be, a democratic system where anyone can try to improve it, like newsmap does to google news or the way amazon lets us earn a reward for pitching their books.
and where anyone can take it and make it their own, like the way Matt Haughey integrates everything from his bookmarks, blogroll, events, music, etc into his blog.

Posted at 8:04PM on Dec 18th 2005 by Pete

7. Dear Mr. Brown,

There is an open data structure called FOAF, there are two open source social networks (Slashter and PeopleAggregator) and a few efforts at answering all your questions.

:-)

Peace out.

Posted at 8:04PM on Dec 18th 2005 by Marc Canter