09 May 2004

social networks networks

also another idea i have is how we visualise social networks. as i talked about before, i'm always interested in how we represent data in social networks. WARNING: incomprehensible random brainstorming session below!

it just occured to me that knowing how strongly you are connected to a person would be much more useful than the path you are connected to the person. because most of the people you are interested in being connected with are usually your immediate inner circle. so you want to know how you are connected.

one can imagine an implementation (barring all privacy and attacker issues) where every user has a list of all the networks it is a part of. for instance, i'm a part of many networks, icq, aim(ichat), msn, yahoo, liquidx.net, cam.ac.uk, unsw.edu.au, etc. think of this as the orkut type communities.

now if we use a dotted hierarchical notation, we can represent them as icq.aol.com, aim.aol.com, msn.microsoft.com, yim.yahoo.com, www.liquidx.net, www-lce.eng.cam.ac.uk, www.cse.unsw.edu.au, thedailyshow.comedycentral.com, ssh.recoil.org, itunes.apple.com, macosx.apple.com, gentoo.linux.org, python.org, etc.

so having a list of membership to networks means that now i can say i'm connected to my friends via 4, 5 or 6 networks. the more networks they are connected to me, the better they are friends with me. naively, that is what we think, but on a deeper level, this is only useful if it has any resemblence to real life. in fact, most of my friends are only connected to me online via instant messenging, but there are alot more connections outside of the online world, for instance we may share the same month for birthdays, we like cars, been to a holiday together. all those are not easily represented as a network because they are very transient. and no one has the amount of time to enter all these relevent details, and no way would that be complete.

so the difficult question is not how we represent networks, but how we create these networks in the first place. how do we document these tiny and transient networks?

maybe one thing i'm thinking of is to make a social networking site/service that just does this. doesn't need you to create a profile for yourself trying to list all the things you know and do, but just to list the events that make you close to people. so an example of this would be to come back home after dinner, and then record into this site, who you've just had dinner with, all the people there. in some sense, you are now creating a record of this network, and if you do this regularly, you can track all the people in your network without relying on them. people can collerate your network with theirs and find connection points.

it would nearly be a diary generation method, record all the people you interact with on a daily basis. document the people you see at work every day. this would be how something like a location system or automatic diary generation system would fit in and be useful.

You can reply to me about this on Twitter: