| 7:59 pm on Jan 1, 2004 (gmt 0)|
So when you search for "Google Zeitgeist" for example, should Google place the Zeitgeist at number 20 with references to it in the top 19 places? I'd rather Google place the site that it thinks is best for the user at number one. All I then have to do is to build the site that's best for the user - simple. :)
| 8:02 pm on Jan 1, 2004 (gmt 0)|
Just ask the folks at Alta Vista about their experience with rotating SERPs -- no stress at all until people stopped using it.
| 8:03 pm on Jan 1, 2004 (gmt 0)|
Biggest problem with this idea is that users want the most relevant sites at the top. My guess is that you are thinking of commercial SERPs. With those, if widgets is a commonly sold item, all the top 20 sites for "buy widgets" could be seen as interchangeable. However, on info searches often they can be ranked best to worst, which Google with PageRank tends to do a decent job with. Also, on some topics there aren't 20 worthwhile sites. As an example, for some relatively obscure 1970s bubblegum rock band, there may be only one dedicated fan site about them. All the other sites have just passing mentions of the band. By rotating the top 20, that one dedicated fan site, which before always was at the top, now may end up in the #20 spot buried. This would lead to searcher dissatisfaction by making this site hard to find in the SERPs.
| 8:17 pm on Jan 1, 2004 (gmt 0)|
It may be nice for those that have sites that don't get into the top5 but would end up causing more stress as we would all have to build 20 times more sites to ensure we get one at #1.
| 8:31 pm on Jan 1, 2004 (gmt 0)|
I'm not sure Google is bothered if we are stressed or not.
And besides that would take all the fun out of it :)
| 7:25 pm on Jan 4, 2004 (gmt 0)|
How about if there was a 'preference' in your google preferences where you controlled whether it rotated or not? For example, when searching for hardware stores, some people are looking for the big mega stores, and some are looking for really good mom-and-pop hardware stores on the corner.
If you just search for hardware store, in a search engine world tailored to the 'most `relevant`, most linked-to, etc' you probably could expect to always see the larger ones, the type with the big blue sign, or the big orange sign? The rotation might allow those searching for the lesser-known to have it come up first.
Especially in highly competitive areas, where there are more than 10 or 20 with the same PR, those would be easy to 'randomize' a bit.
| 7:58 pm on Jan 4, 2004 (gmt 0)|
SERPs aren't ordered on PR - there are many more factors than that. If you want to see the smaller sites, you can either view the second page or use the advanced search function and have the top 100 sites showing on the first page. I don't quite follow the logic of the search engine giving you what it thinks is the best twenty sites, then shuffling them. Surely Google's ability to rank the top twenty sites is why people think it's better than other SEs, even though most of the time most SEs bring up similar sites in similar positions.
| 8:14 pm on Jan 4, 2004 (gmt 0)|
They're not randomised results - they are crap results. There is a difference.
Occasionally even us commerce sites try to buy something.
When we tried to buy a dedicated server, one of the top results was a directory that returned a page about servers - with no links to servers on it.
Oh well! That must be progress.
(Does anyone remember the days when if you needed more RAM for a computer you simply typed it in - and Google came up with an expert site? I must be getting old - but complete crap is now served up by the self-proclaimed masters of the Internet because they want to line their pockets with gold.)
| 8:59 pm on Jan 4, 2004 (gmt 0)|
You couldn't find what you were looking for on searches for "ram" and "dedicated server"? Did you look at the whole page, or did you just hit "I feel Lucky"?
Then when I tried something like "pc100 ram" it all seemed to come up good. Good SERPs, good adwords, good ads at the top of the page, and good froogle entries.