The Interdependency Dependence Notes from DatacenterDynamics NY   When Watson identified Toronto as a city in the United States during its recent Jeopardy competition, it inadvertently highlighted the limits of even the most powerful machines. Billed as a breakthrough in natural language processing, Watson could rapidly search through a treasure trove on information stored in memory. Watson, of course, does not really hear or see, rather it “read” text files served to it from a competition server. Watson is really, really fast, but no faster than a smart human. Congressman Rush Holt (D-NJ), for example, beat Watson at its own game on March 1.

And for all the excitement over Watson’s win over former Jeopardy champions Ken Jennings and Brad Rutter, a fair amount of discussion centered on what was the best way to fairly allow such different competitors to buzz in. Watson’s reaction time, for example, comes very close to zero. The humans, on the other hand, could read and try to anticipate the light control. Buzz too soon, and you are locked out. Too late is, well, too late. Watson does represent a breakthrough; yet at the same time, its errors suggest that Watson produces wrong results from time to time and can never know why.

Similarly, the Marriott Hotel’s computers produced wrong results when I tried to check in for DatacenterDynamics in New York. Eventually it turned out that my name had been entered wrong when I made my reservation. Understandable. But the same system couldn’t find my reservation when I produced an on-line confirmation number. What’s the point of the number (the only one on the confirmation), I asked, if it can’t be used to locate a record?

The computer didn’t make a mistake; it just lacked the ability to search on another parameter or locate similar records as a motivated person might. Brute-force computing is not thinking, however powerful it might seem.

The next morning I found myself at the opening of DatacenterDynamics, which featured 75 exhibitors, more than 1200 attendees, and five program tracks. It was simply too big for one person to cover.

DatacenterDynamics New York is one of the industry’s most impressive gatherings. Each year it seems only to get bigger, with this year’s show being no exception. Attendance seemed to be bolstered by the presence of many from New York City’s financial centers. Buzz on the floor centered around games of moving chairs and companies partnering to develop new business or finding better ways to build new data centers. In short, the show was a hot bed of business activity and well attended technical sessions. As the first big data center show of the calendar year, DatacenterDynamics New York could be the signal of a very good year for the industry.

Those of you who read this space know that DatacenterDynamics is active year round, which is one of its strengths. I recommend that you visit their April 13th event in Phoenix or May 17th in Seattle, if you are in the area.