The effluent of Eflactem

Andrew Chen has an interesting post here on what happens in a network if Metcalfe's Law is reversed - first, his diagram:



(I have a small beef - a veal? with his cost side, as he is ignoring transaction costs, but thats for another post). As he notes, this happy hockey-stick growth party goes swimmingly until:

....one or two people announce the intention to leave. If those folks are fun and entertaining, there's an immediate realization that the quality of the experience is about to go down. And yet more people announce their intention to leave, and so on, until you are left with the party hosts and a big mess

In other words, what happens when those big attractive nodes with 5,000 friends start to leave...and take those friends, and blog like mad that they are off etc etc - yes, you get the reverse Metcalfe's Law (Eflactem's Law, as Andrew calls it). Or, to misquote Tom Lehrer, a social network is like a sewer - what you get out of it depends on what you put into it - and if whats being put into it is a giant sucking sound of disappearing people - well, its still your sewer and you are in the effluent.

Andrew notes that Social Networks need to re-learn something that many previous networks have learned to their cost - churn counts. After a while, as growth slows (see our mathematics on that impact here) its as critical to think about retention of existing customers (well connected first, then les autres).

This is basic game theory - if the network's behaviour is continually disappointing the user, the user will go away (if there is another game to go to or a no-game option). "Defective" behaviour is far easier to get away with if there are no repeat / continual interactions, but when many interactions are repeated, then frequent defective behaviour will lead to increasing defections. It is for this reasonn I think Facebook got into trouble with its Application Spam - one poke is fun but 50 later is a defective experience

In fact there must be a basic law in here- in any networked system, as the number of repeat interactions increase, so the importance of designing the system to maximise "positive" interactions increases with it.