I really don’t know much about power plants, but I have some experience with variable torque systems:
Since generators are synchronous motors, they are like "geared" electrically to the frequency of the net. Lets suppose that the total loses of the turbine and motor when turning at 3600rpm are supplied with a torque of 1 N x m, then, if you let in a very small amount of steam, lets say 1 kg/min, that generate that torque, 1 N x m, then the turbine /generator system “electrical gear”will not "push" the "grid electrical gear", nor will it be pushed by it, and will not bring any power to the grid.
If you want to give power to the grid, then you open your steam valves, and this particular "electrical gear" pushes the big “grid gear”, bringing some power to the net. The amount given will be proportional to the steam admitted, and thus to torque as Poetix writes. It is similar to 4 folks pushing a car. While each folk could push with more or less force, all four will walk at the same speed, and the power of each one will be its speed multiplied by its force.
Here you are working against the "electrical inertia" of the whole grid, referred by Bluemax. If the case were of a single generator connected to the system, you will have a governor that follows with a PID control the electrical demands (voltage/frequency drops), and that is most times more unstable than a larger system.
It is also possible to use an asynchronous motor to generate electricity, but here the speed of the generator will be slightly higher than 3600 rpm due to slipping, the higher the slip the higher the power generated.
Sandro