Yes, you do have to lock the clapper box, the tendency is if cutting on the bottom the tool wants to rise due to it's distance from the pivot point of the clapper and in cutting on top the tool wants to rise and bind up in the cut. Look at the clapper box on nearly any shaper; there are setscrews usually on both sides of the clapper box, with matching divots in the clapper to lock it down, they are there for a purpose. I just read a chapter in Lewis E. King's book Machine Shop Operation booklet #3 Shaper Operation, published by Macmillan, that sometimes keyway cutting is done on the return stroke rather than the normal cutting stroke, I wonder that it may be better to pull the cut than pushing the cut, that is what keyseaters do.
Sadly, my shaper seems to not have said set-screws or any similar locking mechanism.
I guess my question was mostly further clarification to this: "due to it's distance from the pivot point of the clapper". What determines the critical distance here? Why is it not a problem on the normal tool holder?
My theory above was that it simply had enough torque to lift the clapper when pushing, however with a sufficient amount of weight placed further away from the tool (such as a super long bar, with the cutter set further back), you could get the same effect as a normal tool holder.
Or is it the ratio of 'up/down' to 'distance'? If so, do those 'drop boring bar' type holders (where it holds the boring bar in an anderson style holder) have the same problem? Could one build a sufficiently long one of those to not have that problem?