Question about Dynamic Graphs of the Nimda worm

The Dynamic Graphs of the Nimda worm at http://www.caida.org/dynamic/analysis/security/nimda/ are interesting. To what degree do the graphs really tell us how many hosts are infected? Or do they just tell us how many infected hosts are observed every 15 minutes? I guess what I am asking is, is the 15 minute sample period causing the graph to under report the total number of infected sites? How likely is it that the CAIDA monitor will be visited at least once every 15 minutes by an infected system? No matter what the answer to this question is, I do like the fact that the trend is down. -Jeff Ogden Merit