You have a herd of cattle moving in different directions. The cows in the herd are more or less always moving, at different direction and in different velocities.When a cow bumps another cow it affects its direction and perhaps its speed so that they would not keep brushing up against each other. If we take a snapshot of this herd of cattle, we can try to predict the eventual direction of the herd by looking at the edges - for example, if all cattle at the bottom of the herd and at the sides are moving up in a forceful manner, we can imagine the cattle in the middle will change their direction upwards as well (because that's the path of least resistance).
We can also think of a different example. We can imagine a group of academic researchers such that each researcher has some degree of influence on others. Each researcher also has his own topic, which could range from biotechnology to environmental science to algebra. When we observe this system over time, we can imagine that researchers (assuming they're not particularly independent) will change their "research direction" over time to fit the direction of the science herd, and we can probably point out the researchers that will have the most "pull" in setting this direction.
My question is, given this sort of setting in the abstract and in a specific moment in time, how do we identify which actors will have the most influence on the eventual direction of the group (Human beings seem to be able to do this quite intuitively!)? Has this problem been studied?