Hello Gregory, Tuesday, November 23, 2004, 17:27:34, you wrote: GG> So, while the circular reasoning comment is cute, I support Paul's somewhat GG> cautious approach. After all, if say we were discussing a vulnerability GG> in Win2K or something similar, we would make damned certain that the GG> thing works and worked properly and consistently before we pass it around GG> or disclose it, for fear of incurring the wrath of the population of this GG> list, for example. This is indeed the case. I had replied directly to Paul before, and will now expand a bit here on this. My problem with Paul's argument was his choice of "more respected peers". The UCB people published what they *think* was correct, gave the reasoning, and provided the sources and the raw data. So, if I want to prove/dispute/verify, I can analyse UCB's reasoning, even collect the data all over again -- which would be safer --, run other analysis, etc. I may *reproduce* it, *verify* it, I may *disprove* it. On the other hand, the others provided a _statement_ that they did not see anything weird. No data, no reasoning, no anything. Just by being "academic", and very cautious, I am forced to disregard this. I cannot prove their statement to be correct or incorrect. It's just hot air. I absolutely agree with Paul -- there is *NOTHING* to investigate in the judicial sense of the word, since there is nothing proving or suggesting a crime was commited. And, mind you, neither did the UCB people suggest a crime had been commited. They state more analysis is required. They state the results are not coherent with the models. There is, nevertheless a LOT to investigate by the academia -- specifically, UCB's assertion. Only after such an analysis -- which should either confirm or not confirm UCB's results -- can we then think on what can be done (if needed). But to state that UCB's results PROVE fraud is as hot air as the statement as I commented earlier. After all, this is statistics, and almost anyone of us that has played in this field knows how to truthfully lie with it. Now, I hope we can bury this argument and go back to the usual stuff. Huh, perhaps this IS the usual stuff... -- ..hggdh..
Attachment:
pgp00112.pgp
Description: PGP signature