The point is to decrease bloat of the block chain and block index. Many (maybe most) blocks are without user transactions right now. Yet they contribute to the block chain size. What's more important, every block in the chain requires a corresponding entry in the block index. A couple hundred bytes each. It's kept in memory for fast access. In fact, most of the daemon's ~700Mb memory consumption comes from the block index. See yourself how much space is consumed by those *.sst and *.ldb files under data/blktree. Our UTXO data base is fairly compact and may be left as it is for now. The block index can be optimised to some extent, but it's much better to make it grow slower.
do it