However, whether the results on Ben and on me are due to malevolence or the same unreliable reporting of result counts that appear to be prevalent in the other cases, the unreliability of the counts is striking. Until recently I had used counts of results reported as a rough measure of the significance of an individual or a piece of work, or at least of their popular visibility, which is not quite the same thing as significance. Obviously I cannot use the number of results returned as a measure of anything until I understand them further. Can I use Google Scholar, or any other Google site, and assume that results have been returned with an even hand, without fear or favor, and with care for their accuracy? I suspect not.
This is such a biting damnation of Google’s search result count it garnered a response in less than 3 hours. Unfortunately the response is more public relations than substantive explanation or even rebuttal. Restricting a search should always do just that, restrict the search. If the “estimate” goes up, these numbers are not even remotely accurate regardless of how many significant digits you care about. Trying to hide behind the word “estimate” is simply disingenuous when a more accurate term would be “guesstimate.” And Google has no plans to fix this bug since they don’t want to “spend the cycles on that aspect of [their] system.” It’s only their core business. Not very important stuff at all.