4 pipelining requests sounds okay. After a bit of reading, it may be caused by looking for search results that don't exist. I've noticed something strange on google myself these past few months where if I begin browsing through the hundreds of pages of search results, the number of results pages mysteriously shortens as I go on. There aren't as many there as it makes out there are. If you try to jump to the later ones that don't exist, then it thinks you're causing trouble.
You know, I've seen that happen too. (Before all of this happened) At the bottom I would see a link that talked about omitted posts, I'd click them and my pages would go from 100 pages of results to 3. Or I would click on page 72 in the Page 1 2 3 4 5...72 list and it would be blank or it would take me back to the first page. This whole Google Sorry page issue has been happening with them since 2005. It is said that Google will not admit it's a bug. Of course, I never knew of it until it happened this time around to me. J.
I wonder why it numbers them like that if they don't exist? As for Google's Sorry page, I'm not getting it for the moment. That may change shortly though. J.
I always set mine to 8. It doesn't hurt anything, and, while most of the time the difference isn't noticeable, there are times it does make a difference.
My guess is for optimisation. Imperfect hash tables suggest more results than there actually are. Although it is a quick and easy way of filtering the results down to a manageable size. Each of these pages might be entries in a concordance sql database, but it doesn't need to look through all of those results unless it is necessary. Since most people are served by the first results page, preparing hundreds of results would be a waste of effort. It just prepares the first page and it is only when you visit the later pages that it forces google to sort through the rest. It is only then that it discovers the false positives from the hash function, and there are less results than initially suggested. That's my guess anyway.
Ah, okay. I see what you're saying. The results then are just incidental matches, which results in a lot of false positives until they're re-examined. Interesting. J.