[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: VMs: NSU review of Rugg (2003)...

Dear Gordon,

At 13:46 18/12/2003 +0000, Gordon Rugg wrote:
As for detecting the "signature" of a particular table, it's statistically
non-trivial. One issue is that you'd need to assume a particular breakdown of
each word into characters, pairs of characters, syllables, or whatever,before
you could begin any statistical test. If the manuscript is a meaningless hoax
containing gibberish, then there may well be wildcards in there, and deliberate
filling-in of pages using more than one table, plus deliberate creation of
unique meaningless labels - I used all these techniques, and more, when trying
to recreate a hoax. It wouldn't be a case of making some assumptions, running a
test, and calculating a p value - I think that there would be few, if any,
tests which could work with those assumptions which were valid.

If the number of tables is large enough (yet the output statistical properties remain constant), the black art of table construction becomes indistinguishable from cryptography anyway. But if the tables are (as you assert) probably few in number, then the re-use of fragments should be statistically detectable, surely?

Really, we're all bright enough here to devise scientific-sounding theories that are untestable - but what kind of scientist would assert that such a claim helps moves us toward a solution?

As for "deliberate filling-in of pages using more than one table" - this sounds an awful lot like a cop-out on your part. Why would a hoaxer bother, if it looked good enough without going to that much trouble?

Cheers, .....Nick Pelling.....

______________________________________________________________________ To unsubscribe, send mail to majordomo@xxxxxxxxxxx with a body saying: unsubscribe vms-list