I took a break after the last time discussing standardization of human behavior, but now I’m back with a new problem: if I call someone, and I hear the call waiting tone, how long should I hold? How long must I back off before I call again? Back-off time is especially problematic here, since humans are notoriously bad random number generators. Random back-off time is critical to prevent crowding of servers, but how do you prevent parents to a new baby from crashing under the load of calls? Services exist to notify the caller when the line becomes available, but that would mean that the poor parents would have to answer an endless stream of calls. I think a little protocol is in order. Here goes: if you call someone and receive a call waiting tone, enter Jewish-mother-mode (JMM) by sending him an SMS and waiting for him to call back. If the caller suspects that the person on the other end is under a heavy load, he may enter JMM voluntarily. Again, this can be implemented by the endpoint or the service providers automatically: a heavy call load can be detected (as defined by the user or explicitly by setting the user status to busy). Once it is, incoming calls will not cause the phone to ring, but instead, the caller will receive a message stating that the user is busy and asks the caller to wait patiently for the user to get back to him.
My JMM protocol
Actually, this seems like such a good protocol, I wonder why it isn’t implemented by real protocols: a get-back-to-me port, accumulating very short UDP messages which include only an address to get back to (and urgency?); the server later initiates a connection to the client when it has resources. The architecture could be split to a load balancer receiving the get-back-to-me messages, and distributing them to servers according to the current load. Take the MEGACO protocol for instance, at the stage where the Media Gateways (MGs) start looking for their Controller (MGC). MEGACO expects MGCs to be crowded at this point, and dictates a random back off. Instead, it could have the MGs send periodic “get back to me” messages and waiting for an MGC to get back to them.
Of course, all this begs the question: do humans want to be standardized? There are many examples of people modifying their behavior according to a set of rules, traffic laws for instance, but those require a punishment system to ensure they are upheld. 
Should punishment suddenly stop, how long will it take the traffic system to collapse? How long will it take the social system to collapse? According to José Saramago, not long. On the other hand, the animal kingdom is full of examples of protocols: bees signal the direction and distance to patches of flowers with a waggle dance, birds flock, and humans maintain social hierarchy according to a set of norms. It would seem unconscious protocols are better maintained than any mindful protocol.
Next time I’ll revisit this subject, I will concentrate on these form of protocols. Meanwhile, I am open to suggestions.
Comment or trackback
1. SageeB | March 25th, 2008 at 5:23 am
but what about crowding of SMS messages, all asking to “call back”? especially with old celluar phones that have a limited space for SMS messages. I think that your JMM needs a JG (Jewish Grandmother) entity, that will receive all of these messages, and decide when to bother the parents and tell them to call person X or Y.
Trackback this post | Subscribe to the comments via RSS Feed