"Repeat the search" - significant?

[glengara]

Any theories on the rationale behind this option?

I picked a site at random, the "contains the term" search goes from 1,360 to 61 and ends up at 402, the "link" search goes from 96 to 36 back to 94, while the "site" search starts at 521 ends at 332 but with no option to "Repeat the search".

So does the option work as a qualitative filter, an attempt to save bandwidth or what?

[tedster]

The existence of the "Repeat the search" link is the result of duplicate filtering, I think. But the changing numbers as you click into page 2, 3 etc have just been strange, historically and even now.

I think those shifts are an artifact of how Google gets the number of urls - remembering that Google's data is sharded across all kinds of different storage. The size of their data set and the storage messages in use make accurate estimates of how many urls a very challenging issue.

[g1smd]

As I have been involved in getting a lot of Duplicate Content removed from the SERPs I see that effect a lot.

I think the mis-count is caused by data that hasn't been completely erased. It's gone from the cache, but references still pop up in the list of URLs for the site.

When the search is performed, the system retrieves the number and displays it, but when you get to later pages and it goes to fetch the actual data, it realises that it isn't there and so rounds down the reported number accordingly.

There are anomolies.

I am seeing some searches where I add &filter=0 to the search URL, and the initial reported number goes down.

I see some where the actual number of returns decreases with &filter=0 added.

These are for searches that return only dozens to hundreds of results.

.

There used to be some interaction with supplemental results. I think those used to be the major cause of the anomoly.

For URLs we were looking to remove from the SERPs, the actual number would go down, and then the reported number would follow a few days later. Then the cycle would repeat again.