Something about computational genomics? to weed out the thousands or millions of options to model the highest probability good outcomes and the ability to circle back because there is always the occasional unseen diamond, but that the high probability hits were really good as they go through the pipeline.  So I don’t know how soon more good hits come into the pipeline or how many diamonds get weeded out and have to wait to potentially recycle as the process refines.  So your point makes a lot of sense to my simple mind if there is a spend-down of the best guesstimated combinations at the beginning of the process.