Look at the hinge picture. See how the two hinge pins are parallel? Now imagine the blades turning 180 degrees. The hinge pins will now be at the "opposite" angle to before, despite that the blades are symmetrical so identical at 180 degrees to 0 degrees.
So at 0 degrees, increasing torque will, say, increase pitch of the "right-hand" blade while decreasing pitch of the "left-hand" one. But at 180 degrees it will be the opposite. This has the presumably beneficial effect of allowing the craft to climb by simply increasing rotor speed steadily (it will "wobble" a bit but in a spiral fashion which will let it climb without too much inefficiency). Put another way, this means that increasing torque at 0 degrees but decreasing it at 180 degrees allows an asymmetrical pitch to be maintained at a rotational speed which can be seen as constant (over the long term).
In other words, your intuition falls down because the mechanism is not as symmetrical as your brain wants it to be at first glance. Symmetry is an intuitive and attractive for mechanical systems, but it is actually limiting in many cases, and this is a great example.
In an old-school helcopter you have, say, one main rotor with a vertical shaft, and one tail rotor with a horizontal shaft. Each of them has complex control mechanisms implemented in hardware: the main rotor has swash plates, and so does the tail rotor (or it has variable speed). The pilot controls the system in a direct and even crude way: a side stick adjusts blade pitch, sometimes via a simple cable linkage. Throttle is kept quite stable much of the time.
In this newfangled bird, we use some combination of GPS sensors, an input which is presumably a touchscreen (with all the associated latency, plus wireless), and a speed controller which has to adjust multiple times per revolution of the rotors. For some information on why this is tricky, start here: http://en.wikipedia.org/wiki/Nyquist_frequency
The information is the position of the rotor, pulled from the motor sensors. The control is the pulsing of the motor's torque at the right position to increase lift in a particular area of the rotor's arc. Informational control simply means you are pulling information out of the system to better drive the system itself.
So at 0 degrees, increasing torque will, say, increase pitch of the "right-hand" blade while decreasing pitch of the "left-hand" one. But at 180 degrees it will be the opposite. This has the presumably beneficial effect of allowing the craft to climb by simply increasing rotor speed steadily (it will "wobble" a bit but in a spiral fashion which will let it climb without too much inefficiency). Put another way, this means that increasing torque at 0 degrees but decreasing it at 180 degrees allows an asymmetrical pitch to be maintained at a rotational speed which can be seen as constant (over the long term).
In other words, your intuition falls down because the mechanism is not as symmetrical as your brain wants it to be at first glance. Symmetry is an intuitive and attractive for mechanical systems, but it is actually limiting in many cases, and this is a great example.