In my recent post, I noted that when you perform a Google blog search repeatedly on the same string, the figure given for the number of “is occasionally anomolously different”. In fact, it was usually the first search that was apparently anomolously high.
I didn’t make a note of what that high value was then. However, I noted something curious when I repeated the experiment today using the same search strings. The number of results reported from three sucessive searches were:
- ‘Iraq women rights’ – 65,382, 2,652, 2,827
- ‘Iraq men rights’ – 48,813, 2,568, 2,730
- ‘Iraq men killed’ – 73,664, 73,603, 73,549
- ‘Iraq women killed’ – 69,964, 2,999, 3,188
Subsequent searches returned figures broadly in line with the second and third for each string. (The slight variations each time are explained by the fact that Google has multiple data centres which are not kept synchronised. Sucessive searches get dispatched to the various centres at random.)
What I can’t explain is the marked drop in the figure after the first search on three of these strings, nor can I explain why it is not happening with the ‘Iraq men killed’ search. It didtwo days ago.
The upshot of all this is that I no longer have confidence that the number of results reported is meaningful.