Interesting article, truly a unique approach from all the real testing of other sites. Theory is all good, but as others mentioned it should be backed up. Anyhow I learned something new from you.
Regarding latencies, could you also explain why the latencies of Rambus increases when you add more memory? As I understand from "other sites" is that the more Rambus you have the slower it gets.
Noone really says that PC100 is competing with Rambus, nor PC133 for that matter. What all others sites say is that DDR will be the competitor to Rambus, so please don't make the same mistake as all those Intel, ZDNet etc websites that compared Rambus with PC100. Since you make a theoretical comparison you should compare Rambus to DDR and not to PC100/133.
I hope that no poor Rambus investor believed your article and bought some stock... theory seems not be the same a real-life situations.