By metric i just meant some scale like iq by which we could measure "how much" of a trait someone has. Suppose you had a few million genomes sequenced and statistics on each person such as iq, looks etc, could you not use statistical analysis to work out which genes/groups of genes correspond to each trait and use the results to work out the likely traits of an embryo? I didn't really understand your reply which probably means I didn't ask my question very well. If you produced children using methods A,B and C how much better would these children be (roughly) in terms of IQ score, 40 yard dash time, height, etc or is there just no way to tell?
This question stemmed from a podcast where they discussed using these methods briefly and it probably does a better job of explaining what i'm talking about than I have. Here's a link, it should start at the appropriate section (which is only 10 minutes long). http://bloggingheads.tv/videos/2438?in=24:58&out=37:37