pull down to refresh

Does batching help? Has somebody run the numbers? Doesn't the single transaction size scale linearly with the number of the inputs and outputs? Can you show me to a transaction in the mempool with batched channel creation? (Even a test one)