Cashill writes:
Unfortunately, the technology is not currently available to do a fully reliable authorship analysis. As expert in the field Patrick Juola of Duquesne University observed, “The accuracy simply isn't there.” He cautioned that for high stakes issues like this one, “The repercussions of a technical error could be a disaster (in either direction).”
That much said, preliminary QSUM analysis supports an Ayers-Obama link. Systems designer Ed Gold--with twenty years of high-level experience in image and signal processing, pattern recognition, and classifier design and implementation--volunteered to run a QSUM scan on multiple excerpts from both memoirs. “I have completed the analysis,” he wrote me, “and I think you will be pleased with the findings.” In assessing the signature of sample passages from Dreams, he found “a very strong match to all of the Ayers samples that I processed.”
Like Juola, Gold recognized the limitations of the process and of his own resources. He has volunteered to make the raw data available to more established authorship authentication experts, and I will be happy to pass that data along. Gold saw the complementary value, however, in text analysis, as did Juola, who encouraged me “to do what you're already doing . . . good old-fashioned literary detective work.”
Well, there you have it.
Of course the software isn’t perfect, but having that analysis compared to a textual analysis is fairly convincing.
So call me convinced.
It’s Ayers’ book.
To bookmark or copy.
http://www.freerepublic.com/focus/f-news/2281020/posts?page=134#134
http://www.freerepublic.com/focus/news/2281020/posts?page=80#80
http://www.freerepublic.com/focus/f-news/2281020/posts?page=120#120
http://www.freerepublic.com/focus/f-news/2281020/posts?page=122#122
History in Pictures:
http://www.freerepublic.com/focus/news/2278969/posts?page=27#27
http://www.freerepublic.com/focus/news/2278969/posts?page=37#37