I do, I've just never sorted them in the way you describe. You are correct - clearly my brain wasn't working at all that day - I completely overlooked the lower tabs. You are quite correct and I was wrong in what I said about 100% identification across the taxonomy tree - sometimes there is, sometimes there isn't. Your way of sorting the data is in effect giving an average of the capacity of the technique to give specific identification. I looked again at the breakup of the taxonomy tree on one randomly selected recent test to see where most of the loss of identification is coming from (just a quick survey - not exhaustive). For most of the phylla, there is 100% or very close to 100% identification down to the genus level. For what is commonly the second biggest phylum, Bacteroidetes, there is a small loss from phyllum to order and a further small loss in just one genus identification, the rest are virtually 100%. Where almost all the variation is coming from is within Firmicutes, in particular, within the order Clostridiales. Clostridiales is bewilderingly complex and badly in need of reclassification. Whole sections of it are lumped together into "Unclassified" or "incertae sedis" (of uncertain placement). So I think at least some of the problem lies in current inadequacy of the databases used for identification rather than in the technique itself. Because Clostridiales is just so complex identification within the grouping suffers from the same problem that limits species identification. The short lengths of DNA used in the 16S technique don't always include enough information for unequivocal identification.