A friend of mine has found this interesting essay in Nature entitled "Unity from conflict" that deals with the evolutionary mechanisms that allowed the emergence of multicellular organisms.
The problem of how multicellular organisms came about from single cells is quite intriguing. I heard from Lewis Wolpert that this is probably the most important of the seven transitions in evolution as described by Maynard Smith and Szathmáry in their book. In retrospect it is clear that such a transition is possible (since we are here) but, why did it happen?
Paul Rainey (whom I suspect might be a microbiologist) seems to be suggesting that with the right mutation rate (or right mutation bias) multi-cellularity should be possible. Organisms such as myxobacteria seem to be able to alter their mutation rate in response to stress in the environment so I guess that evolution fiddling with the right mutation rate is not unreasonable. In any case I'd rather see it from the point of view of my friend, that is, a harsh environment does enforce cooperation in a way that makes cheating very costly. In reality I would imagine that other factors such as the immune system (that in a way can be though of a police on the lookout for cheaters) or the fact that cells in a multicellular organism share the same DNA could also help explain why there is not that much cheating in our bodies.
This article is quite interesting for any one interested in cancer. At the end of the day a cancer cell is a normal cell that due to genetic or epigenetic reasons stops cooperating. Once they evolve the means to avoid the immune system and other mechanisms designed to maintain homeostasis I would imagine that the life expectancy of a tumour cell should be rather short (necrosis, running behind in the evolution game or due to a poor microenvironment) and thus crime might not pay, at least in the mid/long term (which still would leave room for a benefit in the short term that would be enough to kick-start somatic evolution).
It should be possible using a computational model to demonstrate that an aggressive microenvironment would favour cell cooperation. A mutlicellular organism in which individual cells suffer when exposed to the exterior would evolve a morphology that would minimise the interface with the outside world. it would be also quite likely that a niche of stem cells would evolve to be in charge of generating the cells in this interface that would be in need of constant repair and maintenance. That is what happens in places in which the environment is hostile to cells like the colon or the skin. If cells in the model are allowed to cheat (by means of mutations leading cells to try to avoid being part of the interface if that is their role) that would presumably affect negatively the overall fitness of the organism. However I am not sure that this would rule out other explanations for the evolution of multicellular organisms.