One of the more interesting people I've talked with in the last two years is a figment of his own imagination.
"Craig Barth," the chief technology officer of Florida-based Devil Mountain Software, a company that makes and markets Windows performance metrics software, is, I have discovered, nobody. He doesn't exist.
Barth is, in fact, a nom de plume, which is a fancy, French way of saying "alias." The real man behind the curtain is Randall C. Kennedy, a popular, sometimes outrageous blogger for and frequent contributor to InfoWorld, a publication that like Computerworld is part of IDG. Kennedy's connection to InfoWorld was severed on Friday.
The two, Barth and Kennedy, are one and the same. The problem was that I didn't know that. The problem was that Kennedy didn't tell me he was Barth, that I didn't figure out Barth was he, and that together, they were Devil Mountain.
Devil Mountain's data, derived from a network of PCs whose owners have voluntarily downloaded and installed a stripped version of the software the company has sold to financial service firms, Wall Street traders and government agencies, has provided some interesting insights into PC use and behavior: Internet Explorer is more popular than most believed, or most recently, that Windows 7 machines are twice as likely as XP systems to run low of memory. Barth's data was unique: Microsoft rarely divulges details of the telemetric monitoring it does on Windows PCs. Microsoft declined to comment or to make someone from their Windows or telemetry teams available for an interview, for example, to respond to the memory claims.
I have spoken with the man I knew as Barth between 15 and 20 times since December 2007. There was a phone number and a man behind the phone number. The guy seemed to know his technical stuff.
But on Friday, after I confronted Barth with evidence that linked him to Kennedy -- I didn't yet know they were one and the same -- he assured me that although the two had worked together in the past, and in fact, now worked together at Devil Mountain, any allegations that he and Kennedy were the same person were ridiculous. Two hours later, I received an e-mail from Kennedy, who I'd e-mailed separately.
"Time to level with you," Kennedy wrote. "The individual Craig Barth doesn't exist. It's a pseudonym I created a decade ago while writing news copy for Windows NT Magazine. I resurrected it a few years back in an effort to separate my sometimes controversial editorial contributions to InfoWorld from the hard research content I was developing as part of Devil Mountain Software.
"What began as a simple e-mail exchange of benchmark data two years ago snowballed, as all such white lies tend to do, into the mess we have today," he added.
"Lie" it is, "white" it's not. And "mess" doesn't begin to describe the fall-out over Kennedy's disguising his identity to InfoWorld, Computerworld, and other news organizations and blogs, including the Associated Press, WindowsITPro, and Gizmodo.
Even before this revelation, last week's Computerworld story on Window 7 memory usage had raised a storm of criticism as readers and Windows bloggers reacted with disbelief to the data Barth presented. He fired back, both in Computerworld's follow-up story and on his own blog Friday.
During every interview with Barth since late 2007, I came away convinced he knew what he was talking about. On Friday, Kennedy claimed that everything but his identity was legit: Devil Mountain, which is a registered corporation in the state of Florida; the fact that he developed the performance benchmarking software, which had come out of consulting work he did for Intel; the customers the company had sold its software to; and the data from XPnet. "The research content I've published to date was always based on an unbiased interpretation of the data at hand," he said.
"I even consulted for Microsoft," he bragged during the Friday conversation as Barth, and forewarded a white paper on application virtualization he said he'd written for VMware, and which that company hosts on its site.
Obviously, that's moot now. Readers who scoffed at the data he presented last week have all that much more reason to doubt. Even people who accepted the data as valid, like me, have to wonder where the slippery slope of deception ends.
In the past, reporters could plead the "Don't shoot the messenger" defense, but things are different now. Right or wrong, many people blur the source and messenger into one persona, even when -- as for news reporters at Computerworld -- that's most definitely not the case. Personal experience pieces like this are the rare exception.
Part of a reporter's job is to evaluate the veracity of a source. I did that, but failed, for which I'm sorry.
Gregg Keizer covers Microsoft, security issues, Apple, Web browsers and general technology breaking news for Computerworld. Follow Gregg on Twitter at @gkeizer or subscribe to Gregg's RSS feed . His e-mail address is firstname.lastname@example.org.