I recall reading an explanation of it somewhere. Its much harder to describe than to show in person. It breaks the repetition of the typical baseball drill to the dog and strengthens his understanding of casts. It would probably be better referred to as a "Moving L" (or "V" if you were working on angle backs).
Instead of the pitcher's mound (dog / intersection of the "L") always being in the same place, it's going to relocate before each cast. I'll try to explain as simply as possible :?
Example: Two bumpers. Set dog on the pitcher's mound. You're at home. Toss a bumper to 1st and 2nd base. Send the dog on a back. When he returns, take that bumper and throw it to third. Now (this becomes hard to explain) move behind the plate, down an extended third base line. Dog didn't move from where you received the bumper, so now your old plate is the dog's new mound. As you move away from him and down this line, the original 3rd base becomes a 2nd, and the old 1st stays a first, even though the dog has relocated.
This gives you a great opportunity to reinforce the dog's learning of backs and overs, as the dog is often cast to a bumper that he might have seen from another position. In a different scenario, this over bumper might have been a back just before, depending on what direction you go after receiving the bumper.
It may sound confusing , but think of this: if you want the next cast to be a left, then leave the dog where you received the previous bumper and move so that the remaining bumper is now a left cast. Where you toss the bumper you just received doesn't matter too much, because you're going to move again after you receive the left, and can position yourself so its a left, right, back or wherever you want. The last bumper tossed keeps 'em guessing, since they know there are two bumpers out there in different locations. This also forces them to wait for your cast and helps prevent auto-casting.
Hope this explains it well enough.
