In reply to geoffwhite:
We should probably all extend cams more often and further than we do in vertical (rather than horizontal) cracks.
We are taught to place cams with the stem pointing in the direction of the force than will be applied during a fall. In most cases this means nearly straight down. What we should be taught is to place cam such that by the time we fall off they are still pointing down.
As you move up above the cam, your waist is likely to be out from the wall and the previous runners will be causing drag, this means the cam will rotate, often to be horizontal.
Cams hold when a force is applied perpendicular to the cam, i.e. along the stem. If the stem is pointing straight out this can’t happen and the cam has little holding power. What will happen is that the cam will rotate. Cams don’t rotate at the mid point of the axle, but about their outer cams, i.e. the inner cams are pushed up. So during the period of rotation, you are being held by the outer lobes only. These are often the lobes where the rock is wider, curved, weaker, and hence the placement might fail.
It is worth studying cams when cleaning the pitch, are they still pointing downwards? They should be, but often are not. Ask yourself, would you be happy to lob onto it? Or would you prefer to rotate it downwards, then jump.
This effect of being pulled upwards to the horizontal is greater the greater the drag and tension in the rope. So when you fall off, almost all cams get dragged up. (Watch the draws as you pull the rope at the climbing wall – they all go horizontal, and they are perfectly aligned with little drag and you gently pulling the rope with little tension.) So, in a fall, if the top piece holds for a fraction of second, then blows, the cam below is very likely to now be horizontal, possibly even pointing upwards. Unless you extend, it is quite possible that that slightly higher, not very good, back up piece you placed mid-way through the crux might be your downfall by increasing the possibility of the low bomber piece failing
I think the problem is that we were taught to extend to reduce drag, but it might not have been explained why drag is dangerous, rather than just a pain. Drag lifts cams and wires out or into non-ideal orientations.