Using ideas loosely derived from machine learning theory, I will present a highly speculative program to predict developmental pathways, using some general constraints imposed by the embryo plus the fact they evolved. Segmentation is one situation where interesting insights were obtained. I will be in touch about people I would like to see later in the month, some combo of biology and CS. This topic is less developed than what I had orig given you, but if your audience can tolerate the speculative, maybe the above will be amusing, the drug resistant bacteria story will be out soon in PNAS. Also the above has more CS/theory content.