Technologist accused of spreading Vista, Win7 FUD wasn't a real person

Normally, Betanews doesn't like to do "inside baseball" stories, that deal with the individuals in the technology journalism business and all the insights as to "how the sausage is made." I'll try to make this one as painless as possible, but it needs to be done, because the individual involved had been cited by me in Betanews stories in the past.

Yesterday morning, ZDNet Editor-in-Chief Larry Dignan revealed the results of research showing that a blogger for IDG publications, and the CTO of a testing and research firm cited by that blogger, were actually the same person. Blogger Randall C. Kennedy, a trusted InfoWorld contributor up until yesterday, was Devil Mountain Software Chief Technology Officer "Craig Barth," the author of reports over the years claiming that Windows Vista performance was slower than Windows XP, and recently that Windows 7 performance was slower than Windows Vista.

The admission was admirably acknowledged on InfoWorld yesterday by its editor-in-chief, Eric Knorr: "Integrity and honesty are core to InfoWorld's mission of service to IT professionals, and we view Kennedy's actions as a serious breach of trust." Though Knorr states Kennedy's blog was removed from the site, past posts including a handful to which Dignan linked were left visible for public scrutiny. They show Kennedy referring to Barth, Devil Mountain, or Devil Mountain's "laboratory" exo.performance.network as though it were an outside source.

In his blog post, Dignan states that his research was only able to turn up a quote from Barth in one other publication besides IDG's. That publication was probably us, and the guy citing him was probably me.

You should know that the Betanews article in question, published November 27, 2007, was a critical examination of "Barth's" research methodology. His report purportedly showed that Microsoft Office performance was much slower under Windows Vista than Windows XP. In examining the research, I discovered that "Barth" was using Office 2007 on Vista, versus Office 2003 on XP. Amid an outpouring of complaints (which actually began prior to Betanews' report), "Barth" re-tested Office performance using Office 2003 on both operating systems, and admitted the performance gap was not as wide as originally claimed.

That was the subject of that November 2007 story; and as long-time Betanews readers should recall, I've often disassembled research reports that draw an extremely favorable or extremely unfavorable conclusion about a product or service.

Betanews' second run-in with Kennedy (which we only realized was the second incident just this morning) came just last August. In our story about a CHKDSK- triggered memory leak in Windows 7 discovered by an independent researcher (not Kennedy, not "Barth"), it was Kennedy who grabbed attention by characterizing the bug as a "showstopper," and saying it threatened to postpone the then-pending launch of Windows 7.

"Oh boy!" Kennedy's post began, as though he were licking his lips. "It appears that Microsoft's glowing track record with Windows 7 is about to come to an abrupt and unceremonious end."

As I and many of my colleagues know from experience, even late-discovered bugs in Windows and other major software can be addressed by way of hot-fixes downloaded during the setup process. Microsoft confirmed my prediction that this is how it would handle the problem, and then pointed me to a blog post by company president Steven Sinofsky, who sufficiently diffused the issue. As it turned out, no show was stopped whatsoever, the CHKDSK bug was addressed in time, and few other people were ever bothered by it.

Among the first to spot deficiencies in Devil Mountain Software's research were Betanews readers. As reader HBX, a beta tester of Windows Vista Service Pack 1, wrote in December 2007, "I'm running SP1, and I agree it's a substantial improvement across the board. Everything seems much quicker, and transitions are faster within vista. I've also tested SP3 for XP and I can tell you this: I don't know where Devil Mountain got [its] numbers, but I've test both SP1 and SP3 against each other on the same hardware, and sorry to say it, but Devil Mountain is spreading FUD."

Unfortunately, over the past three years, Kennedy used both of his identities simultaneously, apparently leveraging his Kennedy identity to substantiate his "Barth" identity. In so doing, he successfully duped a handful of reputable and honorable journalists and editors. Only a few hours after Dignan's expose did Kennedy change his story in an e-mail to a frequent contact of his, ComputerWorld reporter Gregg Keizer.

As Keizer cited from Kennedy's e-mail: "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."

In the latest and perhaps saddest turn in this story, Kennedy appears to be capitalizing on his newfound infamy, perhaps as publicity for his research firm, whose list of contributors now bears Kennedy's name rather than "Barth's." In a response to Dignan yesterday, Kennedy not only accused his editors at IDG of complicity, but of actually creating a kind of false Randall Kennedy identity, in an effort, he claimed, to build a kind of "shock jock persona" capable of garnering millions of page views without being answerable to the publisher.

However, while he admitted to leading Keizer on, at the same time, he blamed Keizer for his ability to be misled: "Poor Gregg Keizer -- hey, the man was looking for an anti-Microsoft angle at every turn, and he let his zeal get the best of him. I honestly never meant any harm, especially to Gregg."
Then finally, perhaps caught up in a climax of excitement, a few hours ago, Kennedy posted an "Editorial" on the blog that used to host "Barth," claiming that the entire revelation of his false identity was orchestrated by Microsoft, perhaps in an effort to divert public attention away from some hidden, dark truth.

"It was our research into Windows 7 performance that prompted Microsoft to call in its chips," either Kennedy or "Barth" wrote, speaking apparently on behalf of both. "And call them in it did, instructing its media cronies to silence me by dragging my name through the mud and casting doubt about what is by any measure a very successful professional history."

Larry Dignan has done the world a good service by exposing this individual's duplicity, and Gregg Keizer -- another honorable journalist -- has done well by coming clean and apologizing for an oversight that was made by many other reputable reporters as well.

Kennedy's defense is that, while he himself may have been untrue to the world, his data was not. That defense would never hold up in court. But it's also not even correct: As I discovered -- as it turns out, on two occasions -- his observations and his conclusions were flawed as well. I'm a little embarrassed to discover that two of the examples I often used to demonstrate Betanews' propensity for debunking mythical claims, even in the same sentence, were actually very directly related. That's one connection I missed.

The man's personal problems are perhaps more appropriate for discussion in a psychology publication. If we as journalists and professionals can learn anything applicable to our business from this latest flameout, it's this: At this level, we're expected by our readers to be more than just conveyors of data that somebody else assembled, produced, and pre-digested for our convenience. It's sometimes way too tempting to take somebody else's pre-packaged, extreme conclusions and analyses, complete with suggested headlines for Google News consumption, and run them as-is -- proof positive that the world will end next week, that Obama was born in Uzbekistan, or that Windows 7 is a conspiracy to make the world run slower. More so now than ever, everything we're handed on a silver plate -- everything -- needs to be inspected with a microscope before being passed onto the public as the truth. We need to be the security screeners for the world's facts.