Ryan,
You wrtie: "Well, then I realized, having played in ensembles before, that a group may play very well with out a conductor- if the musicians are not separated by large distances. Feedback among musicians makes this possible, by careful listening while playing. But then, with each musician having adjusted faster or slower to meet at a group tempo, the actual value of the tempo may have converged to something not exactly as the composer intended."
I concur. In fact, that is the insight that led me to the result in my ICCS 2007 paper: Negative feedback informs the present; positive feedback informs the future. Negative feedback, in fact, substitutes for the "conductor." They are the same thing. In a time-dependent network model, conductors change position according to the scale of measure. Google for Braha & Bar-Yam, "From centrality to temporary fame: dynamic centrality in complex networks."
In other words, the partial order that massive coordinated future events impose on the present appears cacophonous from a future perspective because the aggregation of past symphonies and present symphonies at one instant is the same as your metaphor of many orchestras playing different scores at once.
A dissipative system in which time pays an active physical role guarantees a local one dimension succession of beats by negative feedback, while simultaneosuly creating a positive feedback loop with the future, which increases the potential for greater varieties of negative feedback potential in the partially ordered present.
You continue, "It seems there is a duality here as well. When consensus is reached, something most have been compromised. "Sync" by Strogatz provides much insight into such behaviors."
I plan to invest more study into Strogatz, Watts, et al, and small world networks. Perhaps we can travel that path together. From a future perspective, we have all the time in the world. :-)
Just one more comment: You write, "In some sort of intuitive way, 0 and infinity balance, though I'm not sure how to mathematically express that."
For physical applications, I think the central limit theorem and regression to the mean are sufficient.
Tom