Think about the way we play Tetris. When the block is moving, we transform and rotate it freely. Cubes that make blocks are clearly connected and it’s intuitive that their representation in code should be as well. On the other hand, when we try to complete a slice (in 2D, a row) and we succeed, the cubes are removed and the block that was their origin doesn’t matter at this point. In fact, it shouldn’t matter — some boxes from a block may be removed and others not.
Tracing an origin of a box would require constant splitting and merging geometries and trust me, that would be a crazy mess. In original 2D Tetris, sometimes the color of a square was the indicator of the origin block. In 3D, however, we need a quick way to show the z-axis and color is perfect for this.