Forum Moderators: Robert Charlton & goodroi
Today we're helping people get better search results by extending Personalized Search to signed-out users worldwide
That's a staggering statement meaning that every computer accessing Google is now being personalized, signed in or not, so any desktop, laptop or kiosk will start tracking everything everyone does and you won't be able to access the same search results from any two machines.
The possible impact to all is staggering.
We get it, you love Googlaid they can't do anything wrong. It must be nice working for such a perfect company that makes no mistakes.
No, you don't get it. Instead of making an adolescent and inaccurate statement what you apparently regard as an insult, why not address the question that I posed? Again (and try to read slowly and carefully this time):
If Google is going to personalize all of its search results, what's the point of trying to check your rankings by viewing search results through a proxy? Whatever rankings you see won't be the rankings that John across the street and Jane next door are seeing. So why jump through hoops to get results that don't reflect reality?
If Google is going to personalize all of its search results, what's the point of trying to check your rankings by viewing search results through a proxy? Whatever rankings you see won't be the rankings that John across the street and Jane next door are seeing. So why jump through hoops to get results that don't reflect reality?
I got this one the first time, and have to agree, so let me see if I can say the same thing it seems you're saying another way:
If you're getting personalized search results you cannot see what someone else sees, because your results are personalized.
If you're not getting personalized results you cannot see what someone else because their results are personalized.
The only way to accurately check where your site shows up in the actual results other people see is to be served their results, so what's the point of bothering to try to get 'non-personalized' results, except conversation and futility, much like paying any attention to the little green idiot light on the tool bar... It's only a conversation piece and it appears the only real way to determine rankings any more is through traffic and keyword referrals.
If you want to know what you rank for check your stats...
[edited by: TheMadScientist at 5:46 am (utc) on Dec. 11, 2009]
Now it's a find and find again engine, so you're the only one who can possibly mess up your results. It's simple, easy and removes the blame from them. Over time they'll have it narrowed down to one right answer for you... You'll either get YouTube, GoogleMaps, GoogleEarth, GoogleAds, Twitter or WikiPedia (after they buy the last two).
If Google is going to personalize all of its search results, what's the point of trying to check your rankings by viewing search results through a proxy? Whatever rankings you see won't be the rankings that John across the street and Jane next door are seeing. So why jump through hoops to get results that don't reflect reality?
So, here I have a product (class of products) that visitors look for once or twice a year...
Gorg keeps feeding prospects the same junk (mine or others) for 180 days - or so they say. Maybe it makes sense to benchmark clean results?
-----------------
So, most users are going to see Wikipedia, Amazon, and eBay at the top of their search results...
In the UK there are direct marketing/direct mail firms who will help you to target junk mail (the old type in paper envelopes) based on the probability that the recipient will be interested in your product. They basically use correlations between neighbourhoods and household types and buying behaviour. Similar people live in similar neighbourhoods and buy similar things. My point is that the statistics, correlation theory and probability theory used in consumer targeting are all old hat Google are (probably) just applying those statistical principals to our individual user data categorising us as users who like results skewed towards a certain mix for one category of search or another mix for another category of search.
In order to keep their statistics fresh they have to allow this to develop over time. Personalisation would fail if they pigeon hole us and we are fixed in that position. They have to offer a mix so that we can continue to make choices otherwise the whole thing is doomed to fail. We (users) would all become bored with what Google gives us we would ruin Google for ourselves. They must therefore give users skewed natural results so that personalisation remains fresh.
Strongly linked to this is search type categorisation. If they have not analysed it yet then they will some time soon realise that some categories of search work best with more personalisation and some do not. In time certain categories of search will be more personalised than others. Things that you only search for on an annual basis may be categorised one way and be 99% natural other searches that fall into a different more regular category may be heavily personalised.
In short what can be done in statistical analysis of user data is far more complex than we are assuming and when this is mixed with statistical analysis of topic and search type you multiply the complexity.
Cheers
Sid
PS I'm sure that Tedster will shortly point us to the Google patents for using probability theory on user preference data.
ie will the Firefox de-personalized search add-ons work on our browsers?
maybe we could get the whole world to search through them instead? :)
