It's Passover again... Actually the first Passover I have here in Seattle. But I can't say it's an exciting perspective. Tonight I have the first seder at the house of a former mayor of Medina (the town where Bill Gates live, where 2-bedroom houses cost about 1.5 million dollars - yes, this place). But the rest of the week will be just taking food to work, because eastern asian food is not very Passover-friendly (too much corn starch everywhere).
Lately I have been going through this very introspective time in my life (again), where I'm having lost of thought, most of them work-related. I wished I could just forget the world for a couple of months and try to get these thoughts out of my mind before they consume me to death. The good thing is that so far I haven't lost nights of sleep because of them. Only had dreams about it.
Anyway, I'll shoot one of the ideas here just to get away from being "abstract": Imagine simulating the behavior of elements in a database to extract relations. This behavior is very simple (and, thus, hopefully scalable): send a message to an element that has semething to do with you. Then you can just analyze the patterns in these messages and extract the overall relations. Instead of creating a hugely connected graph, you can just see the log traces of a graph that doesn't actually exist and this requires much less memory and can be very powerful by changing the definition of "having somthing to do with you".
I could throw some math here right now, but this would be very cumbersome and nobody would actually be interested in it.
Now my only problem is to find time to get it done.