In a Wall Street Journal article called Estimates for Web Search Results
Are Often Wildly Off the Mark, The Numbers Guy points out that too many people are using web "hits" to define how popular a certain topic is. See that article for his eloquent explanation, but a quick example will suffice.
Example: You do a web searches for New York Yankees and Cincinnati Reds and try to determine the relative popularity of each by looking at the number of times each can be found on the internet. What will skew the results is when all MLB baseball teams are listed on the same page, so you won't get a fair measurement.
In terms of measuring buzz for a topic, I recommend BlogPulse Trend Search.